Overview of Taguchi concepts of quality engineering
Understanding what Design of Experiment (DOE) is and what’s new with the Taguchi approach is essential for developing application expertise. Although, detail statistical theory of DOE is beyond the scope of this seminar, basic principles of DOE and the standardized steps practiced today are covered in detail. You will understand: - How quality is defined as consistency and the statistical terms used to measure it  - Standardization introduced by Taguchi  - How a team approach can be beneficial to application effort

Measuring cost of quality by Loss Function
Conventionally, the cost of lack of quality is measured by cost of rejection at production. In the Loss Function, the ill effects of poor quality when the product is in use, is included in the cost. You will learn how the Loss Function is used to quantify the dollar benefits from improved designs.

Review basic concepts in experimental design
Experimental designs for simultaneous study of multiple two level factors introduces the basic principles in DOE. Types of factors, effects of factor level on the size of experiments, and the desired qualities of the orthogonal arrays are discussed in detail. You will understand: - Types of factors - Multiplicity of levels - Orthogonal array vs. one factor at a time experiments

Project objective and Overall Evaluation Criteria
When project objectives are numerous, it is likely that they are measured using subjective as well as objective evaluation criteria with different units of measurements. When the goal is to seek a design that is overall best, it becomes necessary to combine all evaluation criteria into a single number. A popular scheme is to define an Overall Evaluation Criterion (OEC). You will get an insight into: - Why it is difficult to combine engineering measurements into a single quantity - What is the need for combining multiple evaluation criteria into a single index - Rationale behind the scheme to combine subjective and objective evaluations into one number.

Experiments designed using orthogonal arrays
Orthogonal arrays present opportunities to handle numerous experimental situations using a few simple guidelines. A small number of arrays, each of which can be used for multiple experimental situations, can also be used for: - Experiments with all factors having two, three or four levels - Experiments involving factors at mixed level - Example analysis for Main Effect and Optimum Condition.

Experiments to study interaction
The use of standard orthogonal arrays, although it keeps experiment design process simple, may not always produce satisfactory results. For better accuracy, the study of interaction effects and subsequent corrections may be necessary. Detailed treatment of interaction design and study will help you understand: - Trade off between factors and interactions - Testing for presence of interactions without sacrificing columns - Testing for relative  influence of interaction - Necessary condition for test of significance - Influence of interaction on experimental strategy

Basic analysis and strategy for experimentation
Compromising what you would like to study and what you can afford to study is a key decision in planning the experiment. Because interaction between factors under investigation is inevitable, and since the size of the experiment becomes prohibitively large when all interactions are included in the study, a balance between the number of factors and the interaction becomes extremely important. Knowledge about how to test for presence of interaction allows you to make a practical compromise between the two.

Experiments with mixed level factors
Factors selected for studies may not all have the same number of levels. Mixed levels (2, 3 and 4) factors can often be handled by using the standard 2-level arrays such as L-8, L-16 and L-32. To enhance your experiment design capabilities, you will learn how to:  - Upgrade 2-level columns into a 4-level column - Downgrade (dummy treatment) a 4-level column into a 3-level column - Downgrade (dummy treatment) a 3-level column into a 2-level column - Design 15 different experiments using an L-8 array.

Combination Design (special design tool)
In some experimental situation, conventional design approach makes the experiment too large; some special technique could potentially save experiments and time. Combination Design is such a special technique.

Strategy for Robust Designs
Variation in sample performances is due to uncontrollable, or "Noise" factors. To reduce variation is to minimize the influence of the uncontrollable factors. Conventional approach has been to investigate the uncontrollable factors and attempt to control their influence. But Taguchi offers revolutionary Robust Design concept which, instead of going after the uncontrollable factors, attempts to reduce their influence by adjusting the controllable factors. As part of this new experiment design strategy, you will learn: - New attitudes toward uncontrollable factors - How to design an experiment with Outer array  - How to achieve more "bang for the buck" from your experiment.

Experiment layout for Systems with Dynamic Response (if time is available)
Although common DOE projects under investigations are static in nature, there are some that exhibit response, which changes (Dynamic) depending on a specific factor (Signal). A quick review of the topic is intended to provide understanding of: - The nature of dynamic systems - How the response characteristics are determined - Procedure for carrying out the experiments.

Analysis of Results 
Although detail analysis of less importance in this seminar, complete statistical calculations are presented through five separate example experiments. Review of analysis steps through these examples will provide better understanding of the: - Main effect study for influence of factors - ANOVA for relative influence of factors - Optimum condition for estimation of performance improvement  - Cost savings expected from the improvement - Conservative predictions - Confidence level and confidence interval (C.I.) - Transformation of S/N data into measured units.

Brainstorming for experimental designs
Although, experiment planning is the first step in the application process, its full value cannot be realized until one is completely familiar with the experiment design and analysis techniques. Thus a final summary of the planning process is presented to reaffirm: - the role of the new disciplines in the workplace - Order of discussions in the planning session - Considerations for selecting participants for the planning session.

Computation of Savings using the LOSS FUNCTION
Consistent with the intent of the project, reduction of variation is the primary improvement desired. When analysis is performed using Signal to Noise Ratio (S/N) of the results, quantified improvement in terms of variation reduction can be computed. More than variation reduction, management generally value information in terms of cost benefit (return on investment) achievable from the improved design.

Design and analysis using computer software
Attendees will work as a group to apply the technique in their own project. The steps involved in real life applications, from experiment planning to run confirmation tests are traced. For the sake of class projects, results are assumed and analysis performed as if they were real results. All attendees learn how to analyze results using Qualitek-4 software. Instructor provides the software for the class.