Genichi Taguchi (1924-2012) was made an honorary member of the American Society for Quality (ASQ) in May 1998, one of many awards and commendations bestowed on him. In support of his nomination it was said that his leadership in the quality control field was unsurpassed, and his influence would be felt for a long time in engineering, quality fields and industry sectors, throughout the world.
Taguchi is famous for his pioneering methods of modern quality control and low-cost quality engineering. He is the founder of what has come to be known as the Taguchi method, which seeks to improve product quality at the design stage by integrating quality control into product design, using experiment and statistical analysis. His methods have been said to fundamentally change the philosophy and practice of quality control.
Life and career
Born in Japan in 1924, Taguchi initially studied textile engineering until the escalation of the Second World War where he then served in the Navigation Institute of the Japanese Navy. In 1948 he worked in the Ministry of Public Health and Welfare and in the Institute of Statistical Mathematics of the Ministry of Education, meeting the renowned statistician, Matosaburo Masuyame, who nurtured and honed Taguchi's statistical skills. Here he also got recognition for his contribution to industrial experiments dealing with the production of penicilin.
In 1950, Taguchi joined the Electrical Communication Laboratory (ECL) of Nippon Telephone and Telegraph Company, gaining six years' experience in experimentation and data analysis while developing telephone switching systems. The commercial benefits resulting from his ECL work helped Taguchi to earn the Deming prize in 1960, for his contribution to the field of quality engineering. He went on to win this award, one of Japan's most prestigious commendations, a further three times.
Taguchi was awarded his doctorate by Kyushu University in 1962, after working with industrial statisticians (and beginning his work on the signal to noise ratio) at Bell Laboratories in the US. He continued working for ECL in a consulting role and became part of the associate research staff of the Japanese Standards Association, where he founded the Quality Research Group. In 1964, he took up a professorship at Aoyamagokuin University in Japan, where he spent the next 17 years developing his methods. Attending some of Taguchi's lectures was the founder of Sony and personnel from a struggling car manufacturer named Toyota.
Throughout this time, Taguchi methods were largely unheard of outside Japan. He developed his concept of the quality loss function in the early 1970s, but it was during the 1980s that Taguchi methods became established, when he revisited AT & T Bell Laboratories in the US, as director of the Japanese Academy of Quality.
After that, American interest from companies such as Xerox, Ford and ITT in Taguchi's methodology increased. In 1982, Taguchi was involved in seminars for Ford executives, and the next year he became executive director of the Ford Supplier Institute (later known as the American Supplier Institute). He was also further honoured in 1986, receiving the Indigo Ribbon from the Emperor of Japan for his contribution to Japanese economics and industry (and in 1990 the Blue Ribbon Award from the Emperor of Japan). He was also awarded the International Technology Institute's Willard F. Rockwell Medal for combining engineering and statistical methods to achieve rapid improvements in cost and quality by optimizing product design and manufacturing processes.
Throughout much of this time, Taguchi was also operating as a full-time consultant to various major companies in the US, Japan, China and India.
Apart from occasional work with, for example, Lucas Industries, Taguchi's ideas only became known in Europe from 1986, when the Institute of Statisticians organised a conference in London. The UK Taguchi Club (later, the Quality Methods Association) was formed the following year and, since then, Taguchi methods have been in regular and widespread use in the West in a diverse range of industries although particularly in the car industry. Taguchi died in June 2012.
Taguchi developed methods for both online (process) and offline (design) quality control. This formed the basis of his approach to total quality control and assurance within a product's development life cycle. His approach emphasised improving the quality of product and process prior to manufacture (that is, at the design stage) rather than the more traditional approach of achieving quality through inspection.
Quality loss function
Taguchi's approach differed from the traditional one of manufacturing a product within a specification based on tolerances equally spaced around a target value. He developed a concept of quality loss occurring as soon as there is a deviation away from the target value, and worked in terms of quality loss rather than just quality. He defined quality loss as: 'the loss imparted to society from the time the product is shipped', and this related the loss to society as a whole. Thus, it included both company costs such as reworking, scrapping and maintenance, and any loss to the customer through poor product performance and lowered reliability.
A loss function curve can be calibrated by using information from the customer. A target value is identified as being the best possible value of a quality characteristic. Taguchi associates a simple, quadratic loss function with deviations from the target. Thus:
- The smaller the performance variation, the better the quality of the product.
- The larger the deviation from the target value, the larger the loss to society.
A loss will occur even when the product is within the specification allowed, though it is minimal when the product is on target.
After the design engineer has determined the costs of parts being manufactured out of specification, this information can be used to justify expenditure on quality improvement, enabling decisions to be made on firm cost and quality grounds. Thus, the 'quality gain' from changing a design might sometimes be estimated as not worth while, though ensuring that a product is produced at a quality level acceptable to the customer remains an important consideration.
Signal to noise ratio
One of Taguchi's most innovative ideas was to utilise a 'quality' measure called the signal to noise ratio, which was then used by communications engineers to find the strength of an electrical signal. Taguchi applied this measure to everyday products, and used it as a measure to choose control levels that could best cope with changes in operating and environmental conditions, or noise.
Robust quality of design
On the basis of the signal to noise measure, Taguchi was able to develop the concept of robustness, which enables a product to be designed to be less affected by noise. Given normal variations in process operations, the product in question would be less likely to fail acceptable quality criteria.
Product design improvement
During the product design and production engineering phases, Taguchi set out three steps that must be followed.
- System design: This may involve the development of a prototype design and will determine the materials, parts, and assembly system to be used. The manufacturing process has also to be considered.
- Parameter design: Taguchi's parameter design aimed to find the most cost-effective way of controlling noise. Taguchi process and design improvements are gained by identifying easily controllable factors and settings that minimise performance variation. Controllable factors are design factors that a designer can set or easily adjust. The specified value becomes the signal. Uncontrollable factors are noise, or external variations, and a higher signal to noise ratio means better quality. Taguchi found that if controllable factors were set at optimal levels, the product would be robust to external changes. This was achieved through parameter design applied at the design (offline) stage to reduce or remove the effect of noise factors, and design in robustness. Experiments were designed using orthogonal arrays which (rather simply described) were a series of rows and columns allowing the effects of different factors to be extracted, and separated out. Taguchi was not the inventor of the orthogonal array, but this type of experimentation moved away from the traditional approach of testing one factor at a time and, instead, tested many factors at the same time. His new approach dramatically reduced the number of experiments and prototypes required and, in consequence, costs were much lower. He developed various experimental designs which allowed the variability of the noise factors on each controllable factor setting to be simulated. The settings that minimised variability could then be determined.
- Tolerance design: If parameter design failed, Taguchi suggested using tolerance design to identify the most crucial noise factors. Tolerences could be reassigned so that the overall variability was reduced to acceptable levels.
Invest last not first
Taguchi placed much emphasis on initially optimising the product and process to engineer product quality (parameter design) into the system. Using low cost materials and components was a vital feature of this, and money was spent on higher cost items only when necessary (tolerance design).
In perspective – from Deming to Taguchi
It was W. Edwards Deming who first recognised the importance of moving quality control backwards from inspection to proper process control, notably via Statistical Process Control (SPC). Taguchi moved quality control even further back, to the design stage, thus completing the total quality loop. Taguchi's techniques and statistical experimental designs for offline quality improvement complemented SPC for online quality improvement. Deming's philosophy for management quality improvement encompassed both.
It has been said that Deming's work inspired a revolution in the old management culture whilst Taguchi inspired evolution. Certainly, Deming provided mainly a theory for management, while Taguchi provided important techniques for improving a process at every stage, from design to production, and for keeping the improved processes under control.
Key works by Genichi Taguchi
With Elsayed, E. & Hsiang, T. Quality engineering in production systems. New York, McGraw Hill, 1989
The system of experimental design. vols 1 & 2. New York, Kraus International Publications, 1987
Online quality control. Tokyo: Japanese Standards Association, 1986
Introduction to quality engineering: designing quality into product and processes. Tokyo, Asian Productivity Organisation, 1986
Offline quality control. Nagoya, Central Japan Quality Control Association, 1980
Key works by others
Bendell, T. et al. Taguchi methodology within total quality. Kempston, IFS Publications, 1990
Logothetis, N. Managing for total quality: from Deming to Taguchi and SPC. New York: Prentice Hall, 1992
Belavendram, N. Quality by design: Taguchi techniques for industrial experimentation. Hemel Hempstead: Prentice Hall, 1995
Ealey, L. The methods of a quality master. McKinsey Quarterly, 1994 (4) pp.3-17
Bendell, T. et al. The future role of statistics in quality engineering and management. Journal of the Royal Statistical Society Series D (The Statistician), 48 (3) 1999, pp.299-327