The Power Of To Impact Change In u.s. Companies

L.P. Sullivan

.-

he advancement of "quality engi­ been neutralized by success stories stand the concepts since this re­ T neering" as a technology to si­ in the application of Taguchi Meth­ quires many years of application. mUltaneously reduce cost and im­ ods. However, some benefit has (Today there are only six or seven prove quality was developed by the been derived through this controver­ specialists in the U.S. that have full noted Japanese special­ sy and debate. Traditional statistical Taguchi knowledge.) Many U.S. ist, . Since 1960 Dr. methods are more in focus and companies, however, have been Taguchi has received the Deming "quality engineering" using very successful at applying Taguchi Prize in on four separate oc­ has emerged as a separate technol­ Methods. Therefore, a large number casions for his contribution to the ogy. of specialists exist in supplier com­ field of quality engineering. On Oc­ Advancements in traditional sta­ panies. tobElr 7, 1986, in Pittsburgh, PA, Dr. tistics have been promoted by J. When Ford introduced this tech­ Taguchi received the most prestig­ Stuart Hunter, who received the nology to suppliers (October 1982), ious prize from the International Deming Award for Applied Statistics ITT was the pioneer in case study Technology Institute: The Willard F. in December 1986. This award is for development and they now have Rockwell Medal for Excellence in excellence in the application of sta­ over 2000 case studies. Jim Pratt, Technology. tistics and we are very pleased that ITT director of statistical methods, This medal is awarded to a Dr. Hunter has been recognized for was the first supplier executive to maximum of three individuals each his work. fully appreciate the power of year (no more than one from any The American Experience Taguchi Methods. Today he is a continent) for excellence in technolo­ As one of the major advances leader in this technology. Other gy. In particular, it recognizes the in the history of manufacturing in­ companies who have learned these recipient's contributions toward the dustries, Taguchi Methods are be­ methods include Sheller Globe, generation, transfer, and application ginning to have great impact on Aerojet Ordnance, Allied, Ford, Gen­ of technology for the betterment of changing the way U.S. companies eral Motors, Chrysler, Eaton, Dana, mankind. Following is a portion of operate to simultaneously improve General Tire, Johnson Controls, Po­ the inscription on Dr. Taguchi's quality and reduce cost. This paper laroid, Copeland, General Electric, medal "... Taguchi's major contri­ outlines several changes currently Goodyear, Kelsey-Hayes, Mercury bution has involved combining engi­ being experienced in certain U.S. Marine, United Technologies, Flex neering and statistical methods to companies. Products, Diversitech General, achieve rapid improvements in costs My first introduction to Taguchi Xerox, AT&T, Texas Instruments, and quality by optimizing product Methods was at Nippondenso in Davidson Rubber, A.O. Smith, Allen­ design and manufacturing process­ March 1982. Professor Yuin Wu. Bradley, Control Data, and the Budd es. His methods represent one of who translated Dr. Taguchi's works, Company. the major advances in the history of began supplier training for Ford in Applications experts have manufacturing industries." April 1983. Five years later, I am still emerged in each of these compa­ Many papers have been written learning about this technology, nies. Within Ford, there is a very over the past four years to evaluate which is not covered in traditional aggressive effort to apply Taguchi the technical aspects of Taguchi statistical literature. (The term Methods and develop case studies. Methods and to compare them with Taguchi Methods refers to parame­ Dr. W. H. Moore in the Body & As­ traditional . Also, ter design, tolerance design, the sembly Division heads up a special there have been a great number of quality , on-line quality task force on Taguchi Methods, and debates in the U.S. promoted by control, design of using she is one of the most knowledge­ well known statistical specialists. I orthogonal arrays, and methodology able practitioners in U.S. industry believe the controversy has now applying for the evaluation of mea­ today. Bob Marshall, the Ford Body suring systems). Due to the recent popularity of Taguchi Methods in the U.S., many specialists are applying these meth­ ods. Few individuals really under-

18 Target & Chassis engineering director, is Explanation of Taguchi Methods and Quality Function personally sponsoring the develop­ Deployment ment of many case studies and the presentation of results in symposium Taguchi Methods and Quality Function Deployment (QFD) are major type reviews. elements of an overall business strategy known as Total There are over 6000 case stud­ (TQC). TQC is sometimes also referred to as CWCC-Company Wide ies in U.S. companies. Japanese Quality Control. companies are the leaders in the Total Quality Control originated in Japan as a in significantly application of Taguchi Methods. improving business operations through improved quality. One often misun­ Nippondenso, for example, com­ derstood aspect of TaC is that quality encompasses not just product "good­ pletes over 2600 applications per ness" but economics as well. The Japanese do not consider the NASA or year primarily to optimize product military high-cost approach to be quality! This can be illustrated with the design. Most Japanese companies accomoanying table. and several U.S. c'Jmpanies are Quality using Quality Function Deployment Low High (QFD) as the horizontal mechanism to identify conflicting design require­ I Low Junk ments where Taguchi applications Desired to would be most beneficial. o A good explanation of Taguchi C,) High Bad Business NASA Methods and QFD (see Fig. 1) was written by Bill Eureka, reliability The excellent Japanese companies have moved into the high quality/low manager for General Motors, CPC cost quadrant through the use of TQC techniques, of which Taguchi Methods Group. Ford and Chrysler have writ­ and QFD are the most powerful tools. ten similar definitions which are Taguchi Methods are a collection of techniques for reducing costs being fostered by many supplier through reduction of variation, which will always result in quality improve­ companies. (See the May and June ment! The primary methodology is based on statistical techniques which 1986 issues of Quality Progress for Genichi Taguchi has reduced to a simple form that hourly production workers more detailed reports on Quality in Japan can readily utilize. Taguchi Methods allow designs and manufactur­ Function Deployment). ing processes to be optimized in a manner which makes them insensitive to During the past three years in factors beyond the manufacturer's direct control (such as environmental the U.S., we have seen a significant usages and customer abuse), resulting in "forgiving" designs and processes. change in quality thinking and in en­ Thus designs become easier to build and manufacturing processes become gineering technology through the ap­ more capable, allowing for reduction (or elimination) of expensive inspection plication of Taguchi Methods. Al­ and rework operations. though this change has developed slowly (due to statistical controver­ Quality Function Deployment is a business planning tool which may sy), we are beginning to see where be applied at any point in the business, but is most effective when initiated at this technology will have a greater the earliest stages of product development. When applied at this point, impact on U.S. products than any customer requirements are defined, and a sequence of plans evolve for other individual concept or method ultimately identifying critical product characteristics and manufacturing proc­ yet devised. ess parameters-resulting in satisfaction of the original customer require­ ments. To Impact Change The use of QFD and Taguchi Methods has been instrumental in helping During a recent study mission Japanese companies to improve quality, reduce cost, cut product develop­ to Japan, the president of Toyoda ment time in half, and achieve major competitive market advantages. Gosei made some interesting re­ marks regarding change within a Fig. 1. company. Mr. Nemoto told us that Source: Bill Eureka, reliability manager for General Motors, CPC Group. managers at Toyoda Gosei are pro­ moted on the basis of change; how budget levels. In fact, U.S. manag­ ists have been successful in main­ they were able to impact change in ers who go beyond their job to cre­ taining the status quo for U.S. indus­ their job to help achieve company ate change are considered maver­ tries. In recent years there has been growth objectives (i.e., market icks and they rarely advance to a modest change in company opera­ share). He contrasted this with tradi­ higher levels in the company. U.S. tions through the application of sta­ tional U.S. managers who are evalu­ executives are uncomfortable with tistical process control (SPC). How­ ated on how well they perform with­ managers who promote change! ever, traditional Design of Experi­ in prescribed guidelines, specific The real value of Taguchi Meth­ ments (DOE) has had no impact at areas of responsibility, and fixed ods is to impact change in U.S. all unless modified for the Taguchi companies. Traditional statistical I> theory has created very little change in 30 years. U.S. statistical special-

Summer 1987 19 way of thinking. It is the change that and a small company. Since March Leadership for the supplier was is important because without 1983, ITT has trained approximately provided by C.D. Lauer (Ford vice change we cannot improve the 1200 engineers in Taguchi Methods president of purchasing) who in May competitive position of U. S. prod­ and completed over 2000 case stud­ 1984 declared that Taguchi Methods ucts. The balance of this paper will ies. They have calculated cost sav­ would form the basis lor quality im­ outline several changes that are de­ ings of $35 million from this effort. provement and cost reduction in the veloping within U.S. companies Sheller Globe, which has trained future and all suppliers should be­ which are the direct result of 120 engineers and completed 225 come trained in this new quality Taguchi Method thinking and appli­ case studies, reports a $10 million technology. In May 1985, L. Ross cations. cost savings. Flex Products, with 12 (Ford executive vice president) di­ Change #1 - Definition of Quality engineers trained and 75 case stud­ rected that all engineers at Ford be One of the most important ies, shows a $1.4 million savings. trained in Taguchi Methods. Taguchi changes in U.S. companies resulting In some companies, there is re­ Methods are beginning to take hold from Taguchi Method applications is sistance to case studies due to the at General Motors. An article in the the redefinition of quality and the cost and production interruptions of January 1987 issue of Automotive meaning of quality improvement. running experiments. At Dana Cor­ Industries magazine quoted Gary Since the central theme of parame­ poration, the first nine experiments Dickenson, the General Motors CPC ter design is to reduce cost by re­ cost $33,200 to complete but result­ engineering director, as follows: ducing variability, the definition of ed in a direct cost savings of "We're using Taguchi Methods to quality can no longer be "conform­ $447,000 (not a bad payback for improve aligning the front suspen­ ance to specification" limits. The low-level quality improvement ef· sion. We have an eight-man team meaning of quality improvement, fort). A large portion of reported studying all variabilities such as tire ther!lfore, is changed from problem cost savings is passed on to the pressure, hysteresis in the tire bead, solving to reducing variability around customer, enabling these companies and so on. I want all our people target values. The important point to be more competitive and gain trained in Taguchi Methods." here is how to measure quality im­ larger market share. Quality im­ I think this change in thinking provement! provement is automatic since the about quality is truly a great ad­ In the traditional U.S. way of way in which cost savings were real­ vancement for U.S. industries and it thinking, you measure quality im­ ized was through quality improve­ is a direct result of Taguchi Method provement by evaluating defect ment (by reducing variability). applications by supplier companies rates or the frequency of field prob­ Although engineers at with leadership from Ford. lems. Through the use of Taguchi and Xerox were the first to experi­ Change #2- Financial Control Methods, you measure quality im­ ment with Taguchi Methods in the System provement based on annual cost United States, Ford was without a The Quality Loss Function savings. Through parameter design, doubt the leader in changing compa­ (QLF) for which Dr. Taguchi re­ quality improvement is automatic ny policy through this more enlight­ ceived a Deming Literature Prize and therefore need not be measured ened way of thinking. In a March 3, has the potential for great change in as quality but rather as cost im­ 1983, policy the financial system of U.S. compa­ provement. Therefore, the whole letter, Mr. Peterson (then president nies. The QLF creates an economic focus of quality improvement is to and current chairman) stated, "Em­ perspective which tends to redefine reduce cost. Since other corporate ployees and suppliers must begin to the traditional cost control guidelines performance indicators are cost­ think of the quality of products and under which most U.S. companies based, this more enlightened think­ services not only in the customer operate. Every major company has ing transforms quality from a series sense but relative to the quality of a financial system with payback­ of emotional encounters to a busi­ the processes that produce them. In ratios for capital expenditure and ness strategy for all employees to this regard, we should focus on product design improvement. If a follow. methods to reduce the variability of quality improvement idea does not As I mentioned earlier, there process output, not just meeting meet these guidelines, it cannot be are now over 6000 case studies specifications". implemented. Based on the tradi­ from U.S. companies on the applica­ W.E. Scollard, Ford vice presi­ tional U.S. definition of quality (con­ tion of Taguchi Methods. In quantify­ dent of engineering and manufactur­ formance to specification limits), ing the results of these case studies ing, led a company-wide effort to these guidelines prevent improve­ we rarely see quality (i.e., defects) imbed this thinking in all activities. A ment where part characteristics are as a measure of success. Usually statement from Scollard's presenta­ within print tolerances (see Fig. 2). results are evaluated on the basis of tion to all top executives at Ford on The QLF quantifies annual cost annual cost savings. May 9, 1983, reads as follows; "Our savings as product characteristics For example, let's take results new quality thinking should be re­ improve toward target values even reported by three separate compa­ duced process variability around the when they are within spec. This puts nies: one large, one medium size, nominal as an operating philosophy for never-ending quality improve­ ment."

20 Target a whole new economic perspective Quality Loss Costs: on quality and encourages continual Taguchi VB. Conventional Loss Functions improvement as a method to reduce Conventional Loss Function (Based on conformance to costs. specifications) 100% LOSS For example. General Tire in Any item which measures inside Akron, OH. has been working for tolerance (conforms to specifica- L some time to improve tire conicity tion) has no loss. Any item measuring outside tolerance should and reduce cost. Using the Taguchi be a reject-therefore 100% loss. Loss Function to evaluate internal I loss by sorting as compared with Taguchl Loss Function Lower QU.lily Upper (Based on amount of any deviation Tolerance Ta~gel Tolerance loss by dispersion. they calculated a Limit potential savings over three times from target value) Va!lue Limit Any item deviating from a target I I the amount currently booked in the val ue has a loss associated with 'It. 100% LOSS cost control system. This savings Losses approximate an inverted normal (bell-shaped) curve. funded design and process improve­ L ments for reduced variability in tire conicity. As a result, General Tire is changing the financial system to rec­ Fig. 2. Traditional and Taguchi loss functions are compared. ognize annual savings calculated by the loss function. This process Change #4-Manufacturlng isting equipment has been im­ opens up a whole series of oppor­ Process Improvement proved using Taguchi Methods. tunities to reduce costs by improving Taguchi Method applications reo Before a project is approved, the quality. Similar changes in financial suit in significant changes to the ITT plant manager must present a controls are being implemented by manufacturing system. Traditional Taguchi case study showing how Sheller Globe and several other thinking in U.S. companies holds the current process operation and supplier companies. that parts within spec are OK and tooling was optimized. This guide­ as long as parts are within spec. line effectively reduces cost and Change #3-Englneering Target nothing further is required. Several improves process capability. Values changes that we see developing in • Less reliance on control charts In order for Taguchi Methods of U.S. companies which apply param­ parameter design to be effective. the and process capability analysis. eter design and on-line quality con­ Companies which have imple­ product planning function and engi­ trol methods are: neering activities must establish tar­ mented Taguchi Methods for on­ get values to meet each characteris­ • Continuous improvement through line quality control have reduced tic of the "voice of the customer." variability reduction (discussed in the need for Statistical Process Change #1). Spec limits or tolerances which the­ Control activities and changed the oretically represent the "voice of the With the idea of continuous im­ focus from quality control to cost engineer" are no longer acceptable. provement. traditional quality con· reduction. Traditional SPC is not This requires a change in marketing trol inspection using go-no-go cost based and therefore tends to intelligence as well as engineering gaging is replaced by target or conflict with other in-plant cost knowledge to develop target values best-value analysis. This change systems. On-line quality control which represent the best level for effectively re-directs quality effort methods developed by Dr. function. fit. or appearance. from problem solving to reducing Taguchi include cost equations to When target values are estab­ variability. It is very exciting to determine optimum inspection and lished. the engineering function must see all employees (product engi­ testing intervals. Today only a be directed to optimize product de­ neers. maintenance. tool set-up, very few U.S. companies under­ signs early in the product develop­ line operators. etc.) change their stand on-line quality control tech­ ment stage, especially where con­ thinking to focus on target values. nology while in Japan it is used flicting requirements exist. Trade-offs • New guidelines for funding proc­ extensively. Nippondenso is prob­ and compromises are eliminated ess improvement. ably the most advanced in the where possible or moved to the very New guidelines for funding proc­ use of these methods. last stages of product development. ess improvement result from this Change #5- Horizontal Technical Compromises made in the early change. The idea is to optimize stages-done in many U.S. existing machinery using parame­ U.S. companies are generally companies-become built in and ter design before a project is ap­ strong in vertical deployment and therefore preclude product optimiza­ proved for new, more modern weak in horizontal interaction. tion for robust function. This method equipment. Many U.S. companies Taguchi Methods training and case forces a significant change in the are quick to fund new processes study development. which requires way many U.S. engineering activi­ or tooling without first optimizing cross-functional team interaction. ties operate today. old equipment. At ITT for exam­ creates great change in how a com­ ple. there is a revised guideline in pany operates to improve quality. several operational groups which prohibits capital spending until ex-

Summer 1987 21 The traditional method of as­ assessment of cost. The first Japan junction with the American Supplier signing problems to the areas most Standards Association publication is Institute (ASI) will promote these responsible for solutions and prepar­ entitled, General Tolerancing Rules changes through education, training, ing reports on corrective action sta­ for Plastic Dimensions-JISK-7109. and case study development. The tus simply does not work. It only Several U.S. supplier companies are 1987 Annual Trend Forecast pre­ serves to perpetrate emotional inter­ conducting pilot studies of this stan­ pared by a special ASQC committee action and "finger pointing." Compa­ dard to determine changes to be outlines how this organization can nies which foster team training in made in traditional methods. be most effective in assisting the Taguchi Methods with case stUdy Tolerancing in many U.S. engi­ transformation of U.S. industries. Al­ development and presentation of re­ neering activities historically has though several strategies are out­ sults realize significantly stronger been derived from rule-based meth­ lined in this report, the first relates horizontal interaction. It is the most ods from S.A.E. or S.M.E. stan­ to Understanding New Quality Tech­ powerful internal change to maxi­ dards. More often, however, toler­ nologies. It is stated as follows: mize quality improvement and cost ances are established through "There is a desperate need to reduction! "engineering judgment" or provided educate masses of engineers in The presentations of case stud­ by suppliers who have determined Taguchi principles. Quality profes­ ies to large groups of engineers and the "best we can hold" on a particu­ sionals and selected hourly workers executives is an essential ingredient. lar dimension. Therefore, we find need similar understanding, espe­ When Dr. Taguchi first became a di­ significant variability in tolerances cially in loss function analysis. rector of the American Supplier Insti­ from company-to-company for ge­ "The society needs to take an tute, he insisted on case study pres­ neric processes and parts that are aggressive role in this area. Colleg­ entations in a symposium format. similar in function. es and universities need to be en­ (He was instrumental in developing The important point is an eco­ couraged to familiarize their under­ this review method at Nippondenso). nomic one: Japanese standards graduate and graduate students Since April 1984, after four such using the loss function are derived with these concepts. The society symposia conducted by ASI, we to achieve the lowest cost and best should consider functioning as in a fully appreciate the value of this quality simultaneously. The standard­ role similar to the J. U.SE to take a technique. ization aspect becomes an added firm stand on the education and Several supplier companies in­ advantage to improve efficiencies of training necessary for managers, cluding ITT, Sheller Globe, and engineering activities in a variety of engineers, technicians, and factory Allen-Bradley are holding regular in­ industries. The establishment of floor personnel, and to develop ternal symposium-type reviews for General Tolerancing Rules based on common material. Taguchi Methods. Chrysler held its economic as well as functional con­ "The society should consider a first symposium in November 1986 sideration results in a significant joint venture with the American Sup­ and Ford began quarterly symposia change through the use of Taguchi plier Institute in this area. in January 1987. Technical interac­ Method Technology. We see this "In every case emphasis needs tion is promoted through thousands change emerging in several U.S. to be on the effectiveness of the of case studies presented in forums companies including Ford, ITT, education process. The case study with top management. This effec­ Sheller Globe, and The Budd Com­ approach should be utilized wher­ tively changes internal deployment pany. ever possible rather than passive from vertical (voice of the executive) Conclusion lecture type education . .." to horizontal (voice of the customer). New quality technology based The ultimate benefit of Taguchi Change #6-Industrial Standards on Taguchi Methods was first intro­ Methods education, training, and ap­ The application of Taguchi duced to the U.S. automotive indus­ plication will be to significantly short­ Methods tends to standardize as try in March 1982. It resulted from a en the product development cycle, well as promote new quality technol­ special study mission to Japan led improve quality, and reduce cost. As ogy. When evaluating QFD case by the author which was followed by a result, many U.S. industries will studies in Japanese companies, it many supplier applications. In the become more competitive in world became clear where design engi­ beginning, this effort was fostered markets. The credit for this transfor­ neers are using Taguchi Methods to by an aggressive Ford drive to im­ mation and continued leadership is establish part tolerances based on prove quality and reduce cost. What clearly due to aggressive applica­ economic as well as functional con­ evolved are fundamental changes in tions by U.S. automotive supplier siderations. The Japanese Stan­ U.S. engineering and quality control companies. To them we will be for­ dards Association is now developing methods which greatly changed in ever grateful. General Tolerancing Rules for ge­ the operating aspects of U.S. com­ neric components based on cost as­ panies. These operational changes Author: lawrence P. Sullivan is chair­ pects of dispersion, process capabil­ will drive further improvement in the man and chief executive officer of the ity, and the value of signal-to-noise competitive position of all U.S. in­ American Supplier Institute. ratios. Included is the variability of dustries. measurements with a corresponding The American Society for Quali­ CD ty Control (ASQC) working in con-

22 Target