12.10

A History of Six Sigma

By Dr. Wole Akpose        

The immediate origin of Six Sigma can be traced to its early roots at Motorola, and specifically to Bill Smith (1929 - 1993). Mr. Smith was born in Brooklyn, New York, in the winter of 1929 at the start of the Great Depression. Little has been written about Bill Smith beyond the fact that he graduated from US Navy Academy in 1952 and studied at the University of Minnesota School of Business. Bill Smith was an employee of Motorola and a Vice President and Quality Manager of Land based Mobile Product Sector, when he approached then chairman and CEO Bob Gavin in 1986 with his theory of latent defect.

The core principle of the latent defect theory is that variation in manufacturing processes is the main culprit for defects, and eliminating variation will help eliminate defects, which will in turn eliminate the wastes associated with defects, saving money and increasing customer satisfaction. Variation is measured in terms of sigma values or thresholds. The threshold determined by Smith and agreed to by Motorola is 3.4 defects per million opportunities (3.4 DPMO), which is derived from sigma shifts from specifications.

Motorola adopted the concepts and went on to win the first ever Malcolm Baldrige Excellence Award in 1988, just two years after Bill Smith’s introduction of Six Sigma. This award has been credited with fueling the popularity and widespread adoption of Six Sigma beyond its early roots as a process control methodology into an organization improvement philosophy, and the leading inheritor or usurper (depending on what side of the argument you choose) of Total Quality Management (TQM).

The tools and techniques used by Bill Smith were by no means new when he developed them into a coherent strategy for improving Motorola’s production processes. Indeed, the concept of Statistical Process Control has been attributed to Walter Shewhart (1891 - 1967) and his student, Edwards Deming (1900 - 1993).

Walter Andrew Shewhart was a physicist, engineer and statistician often referred to as the father of statistical process control. A native of New Canton, Illinois, Shewhart attended the University of Illinois and earned his Ph.D. in Physics from University of California Berkeley in 1917. When Walter Shewhart joined the Western Electric Company^[1] inspection of engineering department in 1918, industry quality control was mostly limited to inspecting finished products and removing defective items (similar to how many organizations still do quality control today). The problem was that in the early part of the twentieth century, Bell Telephone engineers were trying to improve telephone transmission reliability, but because amplifiers and other equipment used for transmission were buried underground, making changes after installation was quite expensive, and there was a growing need to minimize incidents and rates of failures and repairs or defective components.  Dr. Shewhart reportedly approached his boss, George D. Edwards, with a prepared memorandum about one page in length, the highlight of which was arguably the first schematic of a control chart and short explanatory text. Shewhart identified variation as the principal culprit for defects in the device manufacturing process and he pointed out that reducing variation, and keeping the process in control would improve quality — he surmised that continual process adjustment in reaction to non-conformance actually increases variation and degrades quality.

Dr. Shewhart suggested that there are two key causes of variation: assignable-cause (or special-cause variation), and chance-cause (or common-cause variation). He introduced the control chart to distinguish between the two, stressing that bringing a production process into a state of statistical control, where there is only chance-cause variation, and keeping the chance-cause variation in control, is it possible to predict and manage a process cost effectively or economically.

Assignable Cause variation is unanticipated, emergent or previously unknown  phenomena within the system. This type of variation is inherently unpredictable (even probabilistically), and can be outside of the historical experiential base. It is usually characterized as a signal within the system itself and often a surprise to practitioners or workers in the process.  Examples of special cause variation include operator failure (e.g., an operator falling asleep on the job), faulty controllers (e.g., software failure), machine malfunction, computer crashes, power surge, abnormal web traffic (sudden, unexpected surge in clicks), absent operators or some other unexpected (and unplanned) system breakdown.

Chance Cause variation is well-known, expected part of the system with predictable variations. Variation may be irregular, but is within an historical experiential base and may also lack significance in individual high or low values. In a way, common cause variation can also be described as the inherent system noise. Examples of chance cause variation include inappropriate procedures, poor designs, poor maintenance, lack of clearly defined procedures, poor working conditions, substandard raw materials, inadequately trained personnel, quality control errors, incomplete testing, vibrations in industrial or manufacturing process, normal system wear and tear, computer response time, etc.

Despite his intellectual heft and the clear value of his work to Bell Telecommunication and other organizations, Shewhart’s best contribution may have been his mentee and student, Edwards Deming. William Edwards Deming was born in the Fall of 1900, earned his BSc in electrical engineering from University of Wyoming at Laramie in 1921, a Master of Science degree from the University of Colorado in 1925 and later a Ph.D. from Yale in 1928 (his graduate degrees were in mathematics and physics). Deming had an internship at Bell Telephone Laboratories in the late 1920s while studying at Yale, where he also met Dr. Walter Shewhart. He worked at the U.S. Department of Agriculture (USDA), and later the Census Bureau. Post World War II, he worked as a consultant to the Japanese government, under Gen. Douglas MacArthur as a census consultant. While in Japan, Deming made a significant contribution to Japan’s reputation for innovative high-quality products, and later Japanese economic power, and is highly regarded as the single non-Japanese individual with most impact on that nation’s manufacturing and business outcomes and growth during and post Marshal Plan.

Deming studied electrical engineering between 1917 and 21 at the University of Wyoming at Laramie and obtained graduate degrees in Physics and Mathematics from University of Colorado (M.S. - 1925), and Yale University (Ph.D. - 1928). In 1936, he also studied under Sir Ronald Aylmer Fisher and Jerzy Neyman at the University College, London, England. Deming interned at Bell Telephone and was introduced to Dr. Walter A. Shewhart by Dr. C.H. Kunsman of the USDA in 1927. Deming found great inspiration in Shewhart’s work on statistical process control and the control chart, and as he took more interest in the application of statistical process control to industrial production and management, he realized that Shewhart’s ideas could also be applied to the processes by which enterprises are led and managed. In 1939, he edited a series of lectures delivered by Dr. Shewhart at USDA into a book, Statistical Method from the Viewpoint of Quality Control. His uncanny ability to distill Shewhart’s work into a more accessible form helped popularized those ideas and their application later in the century, and accounts for his great influence on the economics of the industrialized world, post second world war. One of Deming’s key philosophical viewpoints can be characterized by the following equation:

1. When people and organizations focus primarily on quality as defined by

Quality = (Results of work efforts)/(Total costs)

quality tends to increase and costs fall over time

2. When people and organizations focus primarily on costs, costs tend to rise and quality decline over time.

For his contributions to quality control, Deming is often called the father of modern quality management. His work is the foundation for much of Total Quality Management (TQM). He developed the System of Profound Knowledge, also known as Deming System of Profound Knowledge, which says that all managers must possess the following key knowledge:

• Appreciation of a system - An understanding of the overall system which includes suppliers, producers, customers (or recipients) of goods and services.

• Knowledge of variation - The range and cause(s) of variation in quality, and the use of statistical sampling in measurements.

• Theory of Knowledge - The concepts that explain knowledge and what can be known.

• Knowledge of psychology - The concept of human nature.

These ideas are fundamental to TQM, which is defined by the American Society for Quality (ASQ) as a management approach to long-term success through customer satisfaction. As the guardian of TQM, ASQ has developed a set of certifications built around the management practices described in Deming's book, Out of the Crisis:

• Create constancy of purpose (Corporate Vision and Mission) for improving products and services

• Adopt the philosophy of improvement (or TQM)

• Cease dependence on inspection (or assessments) to achieve quality.

• Stop awarding contracts (or businesses) based on price alone - emphasis on total cost of ownership and added value.

• Continuously improve every planning, production and service process.

• Institute on-the-job training (as a part of continuous improvement)

• Institute leadership - focusing on helping people and machines do a better job.

• Eliminate fear, so that employees work more effectively for the enterprise.

• Eliminate organizational silos, helping employees see the global view of the organization and better appreciate their role in the overall output of the enterprise.

• Eliminate slogans, exhortations and targets for the workforce - The assumption being that the bulk of the cause of low quality and low productivity is the system (or processes) which are often beyond the power of the workforce.

• Eliminate numerical quotas for workforce and numerical goals for management - instead substitute leadership for quotas.

• Remove barriers that rob the hourly worker of his/her rights to pride of workmanship and/or eliminate the annual merit system.

• Institute a vigorous program of education and self-improvement for everyone in the enterprise.

• Engage everyone in the enterprise in the task of organization transformation.

The journey to process quality is a long one, and the pace has accelerated since the days of Henry Ford’s Just-In-Time (JIT) production system, where the emphasis was on cost control, rather than defect reduction; to Edwards Deming’s TQM, which focused on elements of statistical process control as well as organization transformation; to Toyota Production System (TPS) which emphasizes elimination of waste and continuous rapid improvement (using many tools similar to those used in Lean Enterprise); and to Six Sigma at Motorola. Today, it seems, the baton of quality has been passed to an idea called Six Sigma (and increasingly Lean Six Sigma).

While Six Sigma continues its evolution as a methodology and a philosophy, its increasing popularity and increasing widespread adoption continues to fuel debates about its value, and even about concepts like standardization, certification as well as the role of the core metric — 3.4 DPMO.

The term Six Sigma was coined by Bill Smith in 1986, while at Motorola. It was coined as a target for defect-free product manufacturing. The term was derived from the idea that process capability can be described by product or service deviation from specification. For example, if a widget specification is diameter of between 0.01mm and 0.015 mm and for every ten million widgets, 34 are outside the specification, then the process capability sigma level is said to be 3.4 defects per million opportunities or in sigma term, 6σ. It is important at this point to note that there is some mathematical magic in these numbers. Indeed, the number 3.4 DPMO requires a 1.5 sigma shift correction over the long term ( i.e. a process that has a 3.4 DPMO capability in the short term is expected to have a 4.5 sigma capability in the long term). While this fact has been assailed by critics of the metric, it is important to recognize that the spirit of the Six Sigma movement is to eliminate all defects. Six Sigma just happens to be a very reasonable starting metric. If a process has absolutely zero defects in the short term, it may indeed approach 3.4 DPMO in the long term.

Endnotes

[1] Western Electric Company was the manufacturing arm of AT&T from 1881 to 1995, it was also the purchasing agent for the member companies of the Bell System  -  http://en.wikipedia.org/wiki/Western_Electric

References

Walter A. Shewhart : http://en.wikipedia.org/wiki/Walter_Shewhart 

William Edwards Deming: http://en.wikipedia.org/wiki/W._Edwards_Deming

Mikel J. Harry : http://www.mikeljharry.com/milestones.php   http://www.pqa.net/ProdServices/sixsigma/W06002009.html

List of Six Sigma Companies : http://en.wikipedia.org/wiki/List_of_Six_Sigma_companies

Six Sigma: http://en.wikipedia.org/wiki/Six_Sigma

Control Chart: http://en.wikipedia.org/wiki/Control_chart

Toyota Production System: http://www2.toyota.co.jp/en/vision/production_system/

Seven Basic Tools of Quality :  http://en.wikipedia.org/wiki/Seven_Basic_Tools_of_Quality

Plan Do Check Act : http://en.wikipedia.org/wiki/PDCA

Engineering Statistics Handbook, National Institute of Standards and Technology (NIST) -  http://www.itl.nist.gov/div898/handbook/

Thomas Pyzdek. Six Sigma Handbook

Geoffrey Tennant. Six Sigma: SPC and TQM in manufacturing and service. ISBN 0 566 08374 4

Malcolm Baldrige Excellence Award Criteria http://www.nist.gov/baldrige/

Dr. Wole Akpose is the Membership Development Chair for Region 2 and a member of the IEEE ITC&O and the Individual Benefit and Services Committee. He is the founder of HNT Solutions, a technology consulting company and a technology manager and occasional faculty member at Morgan State University. He is also Six Sigma certified.

Comments may be submitted to todaysengineer@ieee.org.