Download the Article (PDF, 94 KB)
What literature on Design for Six Sigma (DFSS) do you find most useful and why?
Compiled by Ramón León, associate professor, department of statistics, University of Tennessee; associate editor, Six Sigma Forum Magazine
Design for Six Sigma (DFSS) is a catchall phrase that has yet to have a precise meaning. Whatever the ultimate definition turns out to be, any design activity that leads to products or processes performing at a six sigma level should be considered part of DFSS.
Design for reliability is one such area, and Gerry Hahn has been kind enough to provide his recommended references.
Another area that should be considered part of DFSS is Robust Design (RD) or Taguchi Methods in which one uses experimental design to find product or process parameter settings that reduce product or process sensitivity to uncontrollable variation. In this area, one of the best references is Madhav Phadkes book Quality Engineering Using Robust Design, Prentice Hall, 1989.
Below are responses from five noted DFSS experts.
Gerry Hahn, consultant and adjunct professor at Rensselear Polytechnic Institute
Design for reliability (DFR) is one of the most critical elements of DFSS. Reliability is the ability of a product to satisfactorily perform its designated function over its customer intended lifetime. It has been referred to as quality over time and is a major bottom-line assessment of product performance in customers minds, especially at the time of repurchase.
In the past, while in a ship and fix mindset, manufacturers spent billions of dollars each year correcting field failures. A key goal of DFR is the design of products with high built-in reliability to avoid or mitigate premature malfunctions in field operation. This is not an easy task. Unlike most other characteristics, reliability cannot be easily evaluated upfront, and extensive testing is required for validation. DFR requires insightful engineering analysis and physically based testing, including accelerated and degradation testing. Initially, prime emphasis needs to be on physics of failure concepts and finding reliability problems expeditiously.
DFR is such a hot topic that much of what has been written about it is still company confidential. More readily available material is just appearing. You might want to search the Web for publications and courses, including:
DFR requires specialized statistical tools. This is due to the presence of censored data on unfailed units, the predominance of distribution models other than the normal such as the Weibull and the high interest in distribution tail percentiles such as the time by which 1% of the units fail. These topics are discussed in detail in books dealing with product life data analysis:
An overview of relevant statistical tools is provided in chapter 48 of J.M. Jurans Quality Handbook, fifth edition (McGraw-Hill, 1999). Also, I co-authored a series of articles with Necip Doganaksoy and Bill Meeker on reliability for Quality Progress. These deal with reliability improvement issues and tools (May 1999), a case study (June 2000), the use of degradation data (June 2001) and analyses by individual failure modes (June 2002).
Thanks to Necip Doganaksoy, Jim Loman, Bill Meeker and Chris Stanard for their valuable input.
Roger Hoerl, manager of the applied statistics lab at GEs global research center
The literature on DFSS is extremely sparse relative to the literature on define-measure-analyze-improve-control (DMAIC). I think the main reason for this is that DFSS is not as understood as DMAIC. Therefore, companies and consultants are much more reluctant to share their DFSS approaches, fearing they will be giving away a competitive advantage. Amazon.com lists 65 titles on Six Sigma. However, only six of these mention DFSS in the title, and two are not yet published.
Heres what I recommend:
Six Sigma and the Evolution of Quality in Product Development, Six Sigma Forum Magazine, Larry Smith, November 2001. This article provides a high level conceptual approach to DFSS and shows how the DFSS tools, which differ somewhat from the DMAIC tools, fit into the high level process. This is a good introduction to DFSS.
Design for Six Sigma: The Revolutionary Process for Achieving Extraordinary Profits, Subir Chowdhury, Dearborn Trade, 2002. While there are points I might quibble with, it provides a step-by-step approach to DFSS and is probably the best book to date on the subject.
Leading Six Sigma: A Step by Step Guide Based on Years of Experience With GE and Other Six Sigma Companies, Roger Hoerl and Ronald Snee, Financial Times/Prentice Hall, October 2002. While not focused on DFSS, this book will present an approach to DFSS we found to be very successful.
You can also find commentaries on DFSS at such popular popular Web sites as www.isixsigma.com and www.sixsigmaforum.com, though these may not provide the level of detail youre looking for. Hopefully, the literature on DFSS will begin to increase at an exponential rate.
Andy Sleeper, DFSS Master Black Belt at Six Sigma Consultants
Setting the PACE in Product Development: A Guide to Product and Cycle-Time Excellence, Michael E. McGrath, Butterworth-Heinemann, 1996. Fundamental tools for teams developing new products, including core teams, structured product development and technology management are clearly explained in this book. This is a valuable reference and improvement guide for any engineering leader.
The Power of Product Platforms: Building Value and Cost Leadership, Marc H. Meyer and Alan P. Lehnard, Free Press, 1997. This book shows how to apply platforms and leverage them as essential tools for streamlined product development.
Miller and Freunds Probability and Statistics for Engineers, Irwin Miller, John Freund and Richard Johnson, Prentice Hall, 1999. Many engineers are inexperienced in applying basic statistical tools. This standard undergraduate text includes clear explanations of common techniques with several examples of real engineering applications. I have used this book frequently in my work with engineering teams.
Measuring Process Capability, Davis Bothe, McGraw-Hill, 1997. The best book of its kind, this reference explains a wide range of capability indexes and how to properly apply them. Of particular value is the thorough discussion of situations omitted from standard statistical process control training, such as process stability, capability confidence bounds and non-normality.
Analysis of Messy Data, Volume I: Designed Experiments, George Milliken and Dallas Johnson, CRC Press, 1992. Most real experiments depart from the neat, completely randomized, two-level designs taught in most design of experiments classes. This book helped me understand and apply appropriate methods for analyzing a wide variety of experiments.
Tolerance Design: A Handbook for Developing Optimal Specifications, Clyde M. Creveling, Prentice Hall, 1996. This is a thorough and clearly written reference for mechanical engineers seeking to predict and minimize the effects of variation on systems they design.
Shin Taguchi, president of the American Supplier Institute
I have been teaching and consulting in the area of RD using Taguchi Methods for the last 19 years, and over the last three years, I have been involved in developing and delivering DFSS to several companies. I believe the heart of DFSS includes quality function deployment (QFD), Pughs concept selection along with the theory of inventive problem solving (TRIZ) and RD optimization.
First, appropriate technology or product planning is done strategically using QFD to set targets and specifications based on the voice of the customers. Then, the engineers develop and select certain concepts, after which they optimize and validate the design.
The following is the list of books I recommend for learning more about DFSS:
The Power of Six Sigma, Subir Chowdhury, Dearborn Trade, 2001. This book is an excellent resource for Yellow Belt training. It is written in a story format and takes only a few hours to read.
Design for Six Sigma (see Hoerls recommendation).
Robust Technology Development, Kenzo Ueno, ASME Press, 1997. This book is an excellent introduction to the concept of Robust Design. Ueno discusses key concepts based on measuring energy transformation. He also touches on management pitfalls you may encounter while trying to change the company culture.
Taguchi Methods for Robust Design, Yuin Wu and Alan Wu, ASME Press, 2000. This book dives deeper into RD optimization. It provides abstracts for 37 case studies with various ideal functions for mechanical, electrical and chemical systems.
Systematic Innovation: An Introduction to TRIZ, John Terninko, Alla Zusman and Boris Zlotin, CRC Press, 1998. This book clearly explains TRIZ and gives numerous real examples. After reading this book, anyone can apply TRIZ and be tremendously successful.
Faysal Khalaf, design for Six Sigma technical specialist and certified Six Sigma Black Belt at Ford Motor Co.
I recommend the following technical materials:
Design and Analysis of Experiments, D.C. Montgomery, John Wiley & Sons, 2000. This book serves as a foundation for different types of experiments and helps guide Master Black Belts, Black Belts and Green Belts through the details of how to analyze and understand the data.
Taguchi Methods, S.G. Peace, Addison Wesley, 1993. This book will help engineers understand the basics of Robust Design. A supplement to the classical design of experiments, it explains how mechanical systems such as transmissions or engines can be designed following RD.
Reliability in Engineering Design, K.C. Kapur and L.R. Lamberson, John Wiley & Sons, 1977. This book is a good reference and will help readers understand the fundamentals of reliability and reliability as a science. An interesting topic covered in this book is probabilistic engineering design and the concept of strength stress distributions, which DFSS focuses on.
Minitab 13 StatGuide, part of Minitab Software, release 13, Minitab, 2000. The Minitab help menu provides a good first-hand learning experience. Choose a subject, and you are on your way to finding a wealth of knowledge and quick references at your finger tips.
Product Design: Techniques in Reverse Engineering and New Product Development, Kevin Otto and Kristin Wood, Prentice Hall, 2000. This is one of the first books to discuss the functional requirements of product design at the concept phase. It covers the techniques needed to represent a complicated design by its functional model and to decompose the overall functions into manageable subfunctions. This ultimately leads to efficient experimentation and optimization.
The Principles of Design, first edition, Nam Suh, Oxford University Press, 1990. This book introduces the concept of axiomatic design, an alternative method to functional modeling. Sometimes, checking the adequacy of the design is important, and Suh illustrates the importance of two main axioms of independence and information. He helps engineers minimize coupling and maintain independency among product functional requirements.
Any statistical process control textbook will help you understand process variations and process capabilities.
Any failure mode and effects analysis handbook will be useful.
Six Sigma Bestsellers*
The Current Literature department features information about relevant Six Sigma publications and articles that readers will find useful. Please send your recommendations to the editor at firstname.lastname@example.org.