BASIC DESIGN OF EXPERIMENTS - 3 DAYS

ADVANCED DESIGN OF EXPERIMENTS - 2 DAYS

J.G. Noguera & Associates

Design of Experiments (DOE) is an off-line quality improvement methodology that dramatically improves industrial products and processes thus enhancing productivity and reducing costs. Input factors are varied in a planned manner to efficiently optimize output responses of interest with minimal variability.

About the Seminars:

Participants learn the basic principles of DOE, and receive detailed guidance on how to implement the concepts - both manually and with software. SAS/JMP, RS/Discover, and Design-Expert software are used for computer demonstrations. A student version of DOE KISS (Keep It Simple Statistically) is provided with the textbook. The required statistics are simplified and provided as needed so that students are not inundated with theory. Participants gain hands on experience with actual industrial examples, use of the Statapult (shown below), Quincunx (bead board), and computer simulations. Basic Design of Experiments is covered in the first three days. Advanced DOE is given on days four and five, covering topics such as Taguchi, Response Surface Methods, and the latest developments in Variation Reduction. By taking the full five days, attendees will have a solid understanding of the BEST approaches to DOE.

Statapult

Seminar Materials:

The seminar textbook provided is Understanding Industrial Designed Experiments, 4th Edition, by S.R. Schmidt and R.G. Launsby. A student version of the DOE KISS software is included.

Who Should Attend?

Participants would include managers, engineers, technical staff, and supervisors involved in product or process design and optimization in manufacturing, quality or process control.

Seminar Outline:

BASIC DESIGN OF EXPERIMENTS (3 days)

• Introduction (DOE)
  - What is DOE?
  - Why Do a Designed Experiment?
  - The Four Stages of Quality Improvement
  - Concepts
  - Problems With Interpreting Routine Operating Data
  - DOE vs. One-Variable-At-A-Time
  - Terminology
  - Types of Experimental Designs
  - Implementation
• Single Factor Experiments
• Two-Factor Factorial Design
  - Calculating Main Effects and Interactions
  - Importance of Randomization
  - Industry Example
• Video - "Planned Experimentation"
• Three-Factor Full -Factorial Design
  - Graphical Interpretation of Effects and Interactions
  - Statistical Concepts
  - Determining Statistical Significance
  - Probability Plots
  - Quincunx (Bead Board) demonstration and exercise
  - Design and Analysis using DOE software
• Fractional Factorials
  - Power and Economy of Fractional Factorials
  - Defining Relation
  - Confounding
  - Resolution
  - Factor Assignment
  - Sequential Use of Fractional Designs
  - Statapult Exercise
• Other Screening Designs
  - Plackett-Burman
  - Taguchi Arrays
• Blocking
  - Benefits of Blocking
  - Blocking for Full Factorial Designs
  - Blocking for Fractional-Factorial Designs
• Model Verification
  - Residuals Analysis
  - Analysis of Variance
  - Pure Error and Lack of Fit
  - Curvature Test
  - DOE Software Analysis and Graphical Tools
• The Sequential Approach

ADVANCED DESIGN OF EXPERIMENTS (2 days)

• Taguchi Methods For Robustness
  - Taguchi's Philosophy
  - Quality Loss Function
  - Parameter Design
  - Tolerance Design
  - Control and Noise Factors
  - Implementing Taguchi Methods with DOE Software
  - Critique of Taguchi Methods
• Further Concepts In Variation Reduction
  - Taking Advantage of Noise-Control Interactions
  - Use of RSM to Determine Robust Operating Conditions
  - Using Fractional Factorial Designs to Efficiently Reduce Variation
• Response Surface Methodology (RSM)
  - Introduction to RSM
  - Path of Steepest Ascent/Descent
  - Yield Simulation Example
  - Types of RSM Designs
  - Regression Methods
  - Data Transformation
  - Including Qualitative Factors
  - Contour Plots
  - Dealing with Multiple Responses
  - Flow Chart for Application of RSM
• Other DOE Methods
  - D-Optimal
  - Mixtures
  - Evolutionary Operation

Return to Home Page