Muutke küpsiste eelistusi

Basic Experimental Strategies and Data Analysis for Science and Engineering [Pehme köide]

(Brigham Young University, Provo, Utah, USA), (University of North Dakota, Grand Forks, USA)
  • Formaat: Paperback / softback, 444 pages, kõrgus x laius: 254x178 mm, kaal: 820 g
  • Ilmumisaeg: 30-Jun-2020
  • Kirjastus: Chapman & Hall/CRC
  • ISBN-10: 036757408X
  • ISBN-13: 9780367574086
Teised raamatud teemal:
Basic Experimental Strategies and Data Analysis for Science and EngineeringSuurem pilt
  • Formaat: Paperback / softback, 444 pages, kõrgus x laius: 254x178 mm, kaal: 820 g
  • Ilmumisaeg: 30-Jun-2020
  • Kirjastus: Chapman & Hall/CRC
  • ISBN-10: 036757408X
  • ISBN-13: 9780367574086
Teised raamatud teemal:
Püsilink: https://www.kriso.ee/db/9780367574086.html
Märksõnad:
Every technical investigation involving trial-and-error experimentation embodies a strategy for deciding what experiments to perform, when to quit, and how to interpret the data. This handbook presents several statistically derived strategies which are more efficient than any intuitive approach and will get the investigator to their goal with the fewest experiments, give the greatest degree of reliability to their conclusions, and keep the risk of overlooking something of practical importance to a minimum.





Features:



















Provides a comprehensive desk reference on experimental design that will be useful to practitioners without extensive statistical knowledge





Features a review of the necessary statistical prerequisites





Presents a set of tables that allow readers to quickly access various experimental designs





Includes a roadmap for where and when to use various experimental design strategies





Shows compelling examples of each method discussed





Illustrates how to reproduce results using several popular software packages on a supplementary website











Following the outlines and examples in this book should quickly allow a working professional or student to select the appropriate experimental design for a research problem at hand, follow the design to conduct the experiments, and analyze and interpret the resulting data.





John Lawson and John Erjavec have a combined 25 years of industrial experience and over 40 years of academic experience. They have taught this material to numerous practicing engineers and scientists as well as undergraduate and graduate students.

Arvustused

"The purpose of this book is to educate scientists and engineers on statistical strategies for developing and improving products and processes, because: 'Companies that use these strategies as standard operating procedures can expect large cost reductions in manufacturing, improved product quality, and reduced lead time for the introduction of new products and/or manufacturing methods' (Preface). The material covered in the book has been taught in a one-semester course and portions of the book have been taught in workshop settings." ~Robert Easterling, Technometrics

"This book aims to present the fundamentals of data analysis and interpretation for the sciences and engineering. Overall, the topics are very informative, and the book is dense in useful knowledge. This material is appropriate for undergraduate or graduate level students; it will be of great use to practitioners across a range of technical fields and for professionals who wish to gain a stronger understanding of experimental design and statistics. There is an emphasis on practical working knowledge, with many examples that are detailed enough to seem realistic. The book starts with basic definitions and concepts related to experimental design and does not presuppose a background in this area. Abundant supporting black-and-white figures and tables allow readers to consider the implementation of statistical concepts. A nice benefit is the exposure to a variety of graphical ways to present dataâfrom dot plots and dot diagrams to histograms, boxplots, three-dimensional surface plots, etc. The work mentions open source programs as an alternative to commonly used proprietary software, such as Excel and Minitab. The volume is strongly recommended for all readers in the sciences and engineering fields." ~M. R. King, Vanderbilt University

List of Figures
xiii
List of Tables
xix
Preface xxiii
Author Bios xxv
1 Strategies For Experimentation With Multiple Factors
1(16)
1.1 Introduction
1(1)
1.2 Some Definitions
2(1)
1.3 Classical Versus Statistical Approaches to Experimentation
3(3)
1.4 Why Experiment? (Analysis of Historical Data)
6(1)
1.5 Diagnosing the Experimental Environment
7(2)
1.6 Example of a Complete Experimental Program
9(4)
1.6.1 Chemical Process System
9(1)
1.6.2 Step One: Screening
9(2)
1.6.3 Step Two: Crude Optimization
11(1)
1.6.4 Step Three: Final Optimization
12(1)
1.7 Good Design Requirements
13(2)
1.8 Summary
15(2)
1.8.1 Important Terms and Concepts
15(2)
2 Statistics And Probability
17(32)
2.1 Introduction
17(1)
2.2 Graphical and Numerical Summaries of a Single Response-Variable
18(7)
2.2.1 Dot Diagrams
18(1)
2.2.2 Sample Mean, Variance, and Standard Deviation
18(2)
2.2.3 Histograms
20(2)
2.2.4 Five-Number Summaries and Boxplots
22(3)
2.3 Graphical and Numerical Summaries of the Relation Between Variables
25(2)
2.4 What to Expect from Theory
27(9)
2.4.1 Probability Density Functions and Probability
27(1)
2.4.2 Expected Values, Variance, and Standard Deviation
28(3)
2.4.3 Normal Distribution
31(1)
2.4.4 Central Limit Theorem and Distribution of Sample Means
32(4)
2.5 Using Theory to Help Interpret Experimental Data
36(10)
2.5.1 Comparing the Mean of Experimental Results to a Standard
36(2)
2.5.2 Comparing the Means of Two Experimental Conditions
38(1)
2.5.3 Comparing the Means of Several Experimental Conditions
39(4)
2.5.4 Comparing Observed Data to the Normal Distribution
43(3)
2.6 Summary
46(1)
2.6.1 Important Equations
46(1)
2.6.2 Important Terms and Concepts
47(1)
2.7 Exercise
47(2)
3 Basic Two-Level Factorial Experiments
49(34)
3.1 Introduction
49(1)
3.2 Two-Level Factorial Design Geometry
50(2)
3.3 Main Effect Estimation
52(2)
3.4 Interactions
54(2)
3.5 General 2k Factorial Designs
56(2)
3.6 Randomization
58(1)
3.7 Example of a 23 Factorial Experiment
59(4)
3.7.1 Background and Design
59(2)
3.7.2 Calculation of Effects and Interactions
61(2)
3.8 Significance of Effects and Interactions
63(7)
3.8.1 Statistical Significance of Results
63(1)
3.8.2 Pooled Variance
64(2)
3.8.3 Statistical Significance of Results for Fly Ash
66(1)
3.8.4 Interpretation of Results for Fly Ash Example
66(2)
3.8.5 Example of Computer Analysis of Data
68(2)
3.9 Example of an Unreplicated 24 Design (Stack Gas Treatment)
70(6)
3.9.1 Background and Design
70(1)
3.9.2 Graphical Analysis
71(1)
3.9.3 Numerical Analysis
72(3)
3.9.4 Interpretation of Results
75(1)
3.10 Judging Significance of Effects When There Are No Replicates
76(3)
3.11 Summary
79(2)
3.11.1 Analysis of 2k Experiments
79(1)
3.11.2 Important Equations
80(1)
3.11.3 Important Terms and Concepts
80(1)
3.12 Exercise
81(2)
4 Additional Tools For Two-Level Factorials
83(28)
4.1 Introduction
83(1)
4.2 Number of Replicates Needed for Desired Precision
83(4)
4.3 Results in Equation Form
87(4)
4.4 Testing for Curvature
91(4)
4.5 Blocking Factorial Experiments
95(6)
4.5.1 Calculating the Error of an Effect (sE) in Blocked 2k Experiments
97(1)
4.5.2 Two Block Example
98(2)
4.5.3 Designs for Blocked Factorials
100(1)
4.6 Split-Plot Designs
101(6)
4.6.1 Split-Plot Example
103(4)
4.6.2 Designs for Split-Plot 2k Experiments
107(1)
4.7 Summary
107(2)
4.7.1 Important Equations
108(1)
4.7.2 Important Terms and Concepts
109(1)
4.8 Exercises
109(2)
5 General Factorial Experiments And Anova
111(24)
5.1 Introduction
111(1)
5.2 Multiple Level Factorial Designs
112(1)
5.3 Mathematical Model for Multiple Level Factorials
113(3)
5.4 Testing the Significance of Main Effects and Interaction Effects
116(1)
5.5 Example of a Multilevel Two-Factor Factorial
117(2)
5.6 Comparison of Means after the ANOVA
119(11)
5.6.1 Least Significant Difference (LSD) Method
120(1)
5.6.2 Tukey's Method of Comparing Means after the ANOVA
121(1)
5.6.3 Orthogonal Contrasts
122(3)
5.6.4 Other Applications of Orthogonal Contrasts
125(1)
5.6.4.1 Orthogonal Polynomial Contrasts
125(1)
5.6.4.2 Analysis of Unreplicated Multilevel Factorials
125(5)
5.7 Analysis of Blocked and Split-Plot Factorial Experiments with Multilevel Factors
130(2)
5.7.1 Analysis of a Blocked Factorial by ANOVA
130(1)
5.7.2 Analysis of a Split-Plot Factorial by ANOVA
131(1)
5.8 Summary
132(2)
5.8.1 Important Equations
132(1)
5.8.2 Important Terms
133(1)
5.9 Exercises
134(1)
6 Variance Component Studies
135(28)
6.1 Introduction
135(1)
6.2 Additivity of Variances
135(1)
6.3 Simple Experiments for Estimating Two Sources of Variability
136(1)
6.4 Estimation of Variance Components Using ANOVA
137(2)
6.5 Graphical Representation of Data from Simple Experiments for Estimating Two Sources of Variability
139(2)
6.6 Components of Variance---Multiple Sources
141(6)
6.6.1 Introduction
141(1)
6.6.2 Nested Designs for Estimating Multiple Components of Variance
141(6)
6.7 Staggered Nested Designs
147(5)
6.7.1 Design and Analysis with Staggered Nested Designs
147(2)
6.7.2 Staggered Nested Design Example with Four Sources
149(3)
6.8 Sequential Experimentation Starting With Components of Variation---Case Study
152(7)
6.8.1 Introduction
152(1)
6.8.2 Staggered Nested Experiment
153(2)
6.8.3 Follow-up Split-Plot Experiment
155(3)
6.8.4 Conclusions
158(1)
6.9 Summary
159(1)
6.9.1 Important Equations
160(1)
6.9.2 Important Terms
160(1)
6.10 Exercises
160(3)
7 Screening Designs
163(40)
7.1 Introduction
163(2)
7.2 Cause-and-Effect Diagrams
165(2)
7.3 Fractionating Factorial Designs
167(1)
7.4 Fractional Factorial Designs
168(16)
7.4.1 Constructing Half Fractions
170(1)
7.4.2 Confounding in Half Fractions
171(2)
7.4.3 Simple Example of a Half Fraction
173(1)
7.4.4 One-Quarter and Higher Fractional Factorials
174(2)
7.4.5 Fractional Factorial Design Tables
176(1)
7.4.6 Example of Fractional Factorial in Process Improvement
177(3)
7.4.7 Advantages of Fractional Factorial Designs
180(3)
7.4.8 Resolution of Fractional Factorial Designs
183(1)
7.5 Plackett--Burman Screening Designs
184(9)
7.5.1 Tables of Plackett--Burman Designs
185(1)
7.5.2 Using the Tables of Plackett--Burman Designs
186(1)
7.5.3 An Example of Using a Plackett--Burman Design
187(6)
7.6 Other Applications of Fractional Factorials
193(3)
7.6.1 Fractional Factorial Designs for Estimating Some Interactions
193(1)
7.6.2 Designs for Blocked Factorials in Fractional Arrangements
193(3)
7.7 Screening Designs with Multiple Level Factors
196(4)
7.7.1 Combination of Factors Method (Pseudofactors)
197(1)
7.7.2 Collapsing Levels (Dummy Levels)
198(1)
7.7.3 L18 Orthogonal Array
198(2)
7.8 Summary
200(1)
7.8.1 Important Terms and Concepts
200(1)
7.8.2 Important Formulas
200(1)
7.9 Exercises
200(3)
8 Regression Analysis
203(34)
8.1 Introduction
203(1)
8.2 Method of Least Squares
203(4)
8.2.1 Estimating the Slope of a Straight Line Through the Origin
205(2)
8.3 Linear Regression
207(9)
8.3.1 Estimating the Slope and Intercept of a Straight Line
207(2)
8.3.2 Statistical Significance of Coefficients
209(3)
8.3.3 Confidence Intervals for Coefficients
212(3)
8.3.4 Precision of Predictions
215(1)
8.4 Multiple Regression
216(9)
8.4.1 Introduction
216(1)
8.4.2 Estimation of Coefficients
217(3)
8.4.3 Statistical Significance of Coefficients
220(2)
8.4.4 Confidence Intervals for Coefficients
222(1)
8.4.5 Precision of Predictions
223(2)
8.5 Quantifying Model Closeness (R2)
225(2)
8.6 Checking Model Assumptions (Residual Plots)
227(4)
8.6.1 Checking for Normal Distribution of Errors
228(1)
8.6.2 Checking for Constant Variance
229(1)
8.6.3 Checking for Independence of Errors
230(1)
8.6.4 Checking for a Mean of Zero for the Errors
230(1)
8.7 Data Transformation for Linearity
231(2)
8.8 Summary
233(1)
8.8.1 Important Equations
233(1)
8.8.2 Important Terms and Concepts
234(1)
8.9 Exercises
234(3)
9 Response Surface Designs
237(26)
9.1 Response Surface Concepts and Methods
237(1)
9.2 Empirical Quadratic Model
238(3)
9.3 Design Considerations
241(3)
9.4 Central Composite Designs
244(3)
9.5 Central Composite Design Example
247(4)
9.6 Graphical Interpretation of Response Surfaces
251(5)
9.7 Other Response Surface Designs
256(4)
9.7.1 Box--Behnken Designs
256(2)
9.7.2 Small Composite Designs
258(1)
9.7.3 Example---Coal Gasification Modeling
258(2)
9.8 Summary
260(2)
9.8.1 Procedure for Design of Experiments for RSM
260(1)
9.8.2 Important Terms and Concepts
261(1)
9.9 Exercises
262(1)
10 Response Surface Model Fitting
263(24)
10.1 Introduction
263(1)
10.2 Estimation of Coefficients in a Quadratic Model
263(3)
10.3 Checking Model Assumptions (Residual Plots)
266(2)
10.4 Statistical Check of Model Adequacy -- Lack of Fit (LoF)
268(3)
10.5 Trimming Insignificant Terms from a Model
271(4)
10.5.1 Justification
271(1)
10.5.2 Deleting Statistically Non-Significant Coefficients from Model
272(3)
10.6 Exploring the Response Surface
275(4)
10.6.1 Analytical Interpretation of Response Surfaces
276(3)
10.6.2 Numerical Methods for Interpreting Response Surfaces
279(1)
10.7 Precision of Predictions
279(2)
10.8 Summary
281(2)
10.8.1 General Procedure for Analysis of Data from RSM Design
281(1)
10.8.2 Important Equations
282(1)
10.8.3 Important Terms and Concepts
283(1)
10.9 Exercises
283(4)
11 Sequential Experimentation
287(30)
11.1 Introduction
287(1)
11.2 Augmenting Screening Designs to Resolve Confounding
288(5)
11.3 Application of Sequential Experimentation in a Process Start-up
293(7)
11.4 Sequential Analysis without Follow-up Experiments
300(11)
11.4.1 Sequential Analysis of Data from a Plackett--Burman Design
300(8)
11.4.2 Sequential Analysis of Screening Designs to Find Quadratic Optima
308(3)
11.5 Summary
311(1)
11.5.1 Important Terms and Concepts
312(1)
11.6 Exercises
312(5)
12 Mixture Experiments
317(42)
12.1 Introduction
317(1)
12.2 Models for Mixture Problems
318(3)
12.2.1 Overview of Modeling
318(1)
12.2.2 Slack Variable Models
319(1)
12.2.3 Scheffe Models
320(1)
12.2.3.1 Scheffe Linear Model
320(1)
12.2.3.2 Scheffe Quadratic Model
320(1)
12.2.3.3 Scheffe Special Cubic Model
321(1)
12.2.3.4 Scheffe Full Cubic Model
321(1)
12.3 Experimental Designs for Mixture Problems
321(12)
12.3.1 Unconstrained Problems
321(2)
12.3.2 An Example
323(2)
12.3.3 Constrained Mixture Problems
325(2)
12.3.3.1 Pseudocomponents
327(1)
12.3.3.2 Extreme Vertices Design
328(5)
12.4 Data Analysis and Model Fitting for Constrained Mixture Problems
333(10)
12.4.1 Least Squares Model Fitting Examples
333(4)
12.4.2 A Procedure for Choosing the Correct Model
337(1)
12.4.3 Interpretation of Fitted Models
338(3)
12.4.4 Identifying Optimum Conditions
341(2)
12.5 Screening Experiments with Mixtures
343(6)
12.5.1 Designs for Screening Experiments with Mixtures
343(3)
12.5.2 An Example
346(3)
12.6 Mixture Experiments with Process Variables
349(7)
12.6.1 Designs for Mixture Experiments with Process Variables
349(1)
12.6.2 Models for Mixture Experiments with Process Variables
350(1)
12.6.3 An Example
351(5)
12.7 Summary
356(1)
12.8 Exercises
356(3)
13 Practical Suggestions For Successful Experimentation
359(4)
13.1 Introduction
359(1)
13.2 Points to Remember
359(4)
Appendix A 363(6)
Appendix B 369(42)
Check Figures for Selected Exercises 411(2)
Bibliography 413(4)
Index 417
John Lawson is a professor of statistics in the Department of Statistics at Brigham Young University, Provo, Utah. John Erjavec is a retired professor and chair of the Department of Chemical Engineering, University of North Dakota, Grand Forks, North Dakota.