Design of Engineering Experiments - Jordan …haalshraideh/Courses/IE710/Ch01_new.pdfDesign of...
Transcript of Design of Engineering Experiments - Jordan …haalshraideh/Courses/IE710/Ch01_new.pdfDesign of...
Design of Engineering Experiments
Hussam Alshraideh
Chapter 1: Introduction to Designed Experiments
October 4, 2015
Hussam Alshraideh (JUST) Introduction October 4, 2015 1 / 21
Introduction
Statistical Studies
How can we evaluate evidenceagainst global warming?
Are cell phones dangerous?
How likely are we to win thelottery?
Is there bias against women inappointing managers?
Can Champix help smokers inquitting?
Hussam Alshraideh (JUST) Introduction October 4, 2015 2 / 21
Introduction
Data
Data is information we gatherthrough experiments or observation.
Experiment on low carb diet
Data: weight of subjectsbefore and after
Survey on effectiveness of a TVad
Data: percentage who wentto Starbucks since ad aired
- The ultimate goal is to translate data into knowledge andunderstanding.
- Statistics is the art and science of learning from data.
Hussam Alshraideh (JUST) Introduction October 4, 2015 3 / 21
Introduction
Three Aspects of a Study
Design: Planning how to obtaindata
Description: Summarizing thedata
Inference: Making decisions andpredictions
Hussam Alshraideh (JUST) Introduction October 4, 2015 4 / 21
Introduction
Methods of Data Collection
Observational: used to assess association
Retrospective studies: historical dataProspective studies: current observation
Designed Experiments: used to obtain cause-and-effect relationships
Study ”Subjects” are controlled.
Subjects: The entities that we measure in a studyMeasurements may vary from subject to subject. ⇒ Variability
Hussam Alshraideh (JUST) Introduction October 4, 2015 5 / 21
Introduction to DOX
Introduction to DOX
An experiment is a test or a series of tests to identify reasons ofchange in some response variable.
Experiments are used widely in the engineering world
Process characterization & optimizationEvaluation of material propertiesProduct design & developmentComponent & system tolerance determination
”All experiments are designed experiments, some are poorly designed,some are well-designed”
Hussam Alshraideh (JUST) Introduction October 4, 2015 6 / 21
Introduction to DOX
Introduction to DOX
Example: study the effect of two hardening processes on an aluminumalloy.
Oil quenching vs. Saltwater quenching.
Objective: to determine which quenching solution produces themaximum hardness.
Run tests for each solution and compare averages.⇒ poor design
Hussam Alshraideh (JUST) Introduction October 4, 2015 7 / 21
Introduction to DOX
Introduction to DOX
Before running the experiment need to know:
Other quenching media: air
Other factors: temperature of media
# of runs at each combination: 2, 3, 5?
Order of runs
Method of data analysis
Significant difference in the average
Hussam Alshraideh (JUST) Introduction October 4, 2015 8 / 21
Introduction to DOX
Engineering Experiments
What variables are mostinfluential?
Screening experiments
Settings of x’s for optimal y?
Response surface methods
Settings of x’s to minimizevariability of y?
Robust parameter designs
Setting of x’s for minimumeffect of z’s on y?
Robust parameter designs
Hussam Alshraideh (JUST) Introduction October 4, 2015 9 / 21
History of DOE
Four Eras in the History of DOX
The agricultural origins, 1908 1940s
W.S. Gossett and the t-test (1908)R. A. Fisher & his co-workersProfound impact on agricultural scienceFactorial designs, ANOVA
The first industrial era, 1951 late 1970s
Box & Wilson (1951), response surfacesApplications in the chemical & process industries
The second industrial era, late 1970s 1990
Quality improvement initiatives in many companiesTaguchi and robust parameter design, process robustness
The modern era, beginning circa 1990
Hussam Alshraideh (JUST) Introduction October 4, 2015 10 / 21
History of DOE
William Sealy Gosset (1876-1937)
Gosset’s interest in barley cultivationled him to speculate that design ofexperiments should aim, not only atimproving the average yield, but also atbreeding varieties whose yield wasinsensitive (robust) to variation in soiland climate. Gosset was a friend ofboth Karl Pearson and R.A. Fisher, anachievement, for each had amonumental ego and a loathing for theother. Gosset was a modest man whocut short an admirer with the commentthat Fisher would have discovered it allanyway.”
Hussam Alshraideh (JUST) Introduction October 4, 2015 11 / 21
History of DOE
R. A. Fisher (1890 1962) George E. P. Box
Hussam Alshraideh (JUST) Introduction October 4, 2015 12 / 21
The Basic Principles of DOX
The Basic Principles of DOX
Randomization
Running the trials in an experiment in random orderNotion of balancing out effects of lurking variables
Replication
Sample size (improving precision of effect estimation, estimation oferror or background noise)Replication versus repeat measurements? (see pages 12, 13)
Blocking
Dealing with nuisance factors
Hussam Alshraideh (JUST) Introduction October 4, 2015 13 / 21
Strategy of Experimentation
Strategy of Experimentation
”Best-guess” experiments
Used a lotMore successful than you might suspect, but there aredisadvantages(needs good first guess, no guarantee of optimality)
One-factor-at-a-time (OFAT) experiments
Sometimes associated with the scientific or engineering methodDevastated by interaction, also very inefficient
Statistically designed experiments
Based on Fishers factorial concept
Hussam Alshraideh (JUST) Introduction October 4, 2015 14 / 21
Factorial Designs
Factorial Designs
In a factorial experiment, all possiblecombinations of factor levels are tested
The golf experiment:
Type of driver (Oversized, Regular sized)Type of ball (Balata, Three piece)Walking vs. RidingType of beverage (Water, Something Else)Time of round (morning, afternoon)Weather (cool, hot)Type of golf shoe spike (metal, plastic)Windy vs. calm days
Hussam Alshraideh (JUST) Introduction October 4, 2015 15 / 21
Factorial Designs
Factorial Designs
Hussam Alshraideh (JUST) Introduction October 4, 2015 16 / 21
Factorial Designs
No Title
Hussam Alshraideh (JUST) Introduction October 4, 2015 17 / 21
Factorial Designs
Factorial Designs with Several Factors
Hussam Alshraideh (JUST) Introduction October 4, 2015 18 / 21
Factorial Designs
Factorial Designs with Several Factors: A FractionalFactorial
Hussam Alshraideh (JUST) Introduction October 4, 2015 19 / 21
Planning, Conducting and Analyzing an Experiment
Planning, Conducting & Analyzing an Experiment
1 Recognition of and statement of problem
2 Choice of factors, levels, and ranges
3 Selection of the response variable(s)
4 Choice of design
5 Conducting the experiment
6 Statistical analysis
7 Drawing conclusions, recommendations
Hussam Alshraideh (JUST) Introduction October 4, 2015 20 / 21
Planning, Conducting and Analyzing an Experiment
Planning, Conducting & Analyzing an Experiment
Get statistical thinking involved early
Your non-statistical knowledge is crucial to success
Pre-experimental planning (steps 1-3) vital
Think and experiment sequentially (use the KISS principle)
See Coleman & Montgomery (1993) Technometrics paper +supplemental text material
Hussam Alshraideh (JUST) Introduction October 4, 2015 21 / 21