John Terninko, Ph.D.
Responsible Management Inc. & Ideation International Inc.
The Ninth Symposium on Quality Function Deployment, June 10, 1997
Taguchi's approach to robust designs has been in North America since 1981. QFD arrived in 1984, and TRIZ, the newcomer, arrived publicly in 1991. Each contributes to one aspect of the design process. Together they become an unbeatable powerhouse of Customer Driven Robust Innovations. QFD gathers and translates customer requirements (the voice of the customer) into design requirements (the voice of the innovator/engineer). The prioritization of desired improvements and the corporation's performance measures in QFD become, in TRIZ, the so-called "initial useful functions." The output of the QFD is used to rank the many innovative concepts generated by the TRIZ methodology. The TRIZ methodology provides only the concepts of a design -- not the design details. Taguchi's methodology determines the design specifications for a product to be insensitive to uncontrolled influences. This paper discusses these and other linkages between these powerful quality tools. The synergy formed becomes the ideal design process.
A1 Matrix, Contradictions, Creativity, House of Quality, Ideality, Innovation, Loss Function, QFD, Quality Function Deployment, Robust Design, Systematic Innovation, Taguchi, Taguchi Methods, Theory of Inventive Problem Solving, TIPS, TRIZ, Useful Function, Voice of Customer Table.
The Essence of QFD
The fundamental principle of QFD is to gather all relevant information about the customer and use this information to drive the design of a product or service. Several tools have been integrated into the QFD process to facilitate the deployment of this information throughout the design and manufacturing processes, and to all relevant organizational functions. The primary function of QFD is to identify important issues and link priorities and target values back to the customer.
QFD started as a means to improve market share by reducing the gap between the customer's desires and the product's performance.
Identification of the Customer Segments and using the Analytic Hierarchy Process (AHP) to prioritize the customer is an effective first step. Gathering the voice of the customer and the context of use is an important second step. Again AHP can be used to prioritize the verbatim information. The context of use defines the constraints in which the product must function or the service be performed. These constraints often create design bottlenecks for which QFD provides no means for resolution. For this reason the Theory of Inventive Problem Solving (TRIZ) is a welcome improvement to past innovation tools.
The Essence of TRIZ
When Genrich Altshuller completed his research of the world patent base he had identified four key learnings: 1. there are five levels of invention; 2. inventive problems contain at least one contradiction; 3. there are standard patterns of evolution; and 4. the same principles are used in many inventive designs and can therefore be considered solution patterns.
Exhaustive study of the world's patents reveals that the same principles have been used in innovative solutions to problems in different industries, sometimes with many years elapsing between applications. Access to this information is one of the contributions of TRIZ. For example, the principle of separation by means of increasing the pressure on some material and then suddenly returning it to one atmosphere is the method by which artificial diamonds are cracked, sunflower seeds and cedar nuts are shelled, the tops and insides are removed from peppers, powdered sugar is made, filters are cleaned, and rust is removed from metal parts. All these applications use a sudden pressure drop and all require engineering effort in order to determine the best system and operating conditions. Splitting diamonds requires 2000 atmospheres where processing peppers requires only five. Shelling cedar nuts uses high-pressure water where shelling sunflower seeds uses air.
A visit to a museum of technology reveals that the same patterns of evolution exist in very diverse products. One such pattern is the migration from exerting no control over operating conditions to human control to automatic control. Another pattern is the transition from a single-function product to one which has multiple functions which may be either homogeneous or heterogeneous. The evolution of the duplicating machine illustrates this pattern. Starting with simply coping one page, today addition functions include digital storage of the master, printing on two sides, collating, mixing feed stock and binding. TRIZ offers eight patterns of evolution containing 280 lines of evolution from which tomorrow's products can be designed today.
Using the TRIZ methodology, it is possible to generate concepts for reducing negative effects and improving the performance of existing designs. TRIZ includes four analytical tools used to structure the innovative problem and six knowledge-base tools used to point in the direction of solution concepts.
Innovation Situation Questionnaire (ISQ)............
Gathers all the relevant data for analysis.
Problem Formulator Takes a single problem statement and, through the use of linked cause-and-effect statements, generates an exhaustive list of more explicit problems.
Algorithm for Inventive Problem Solving (ARIZ)
An alternative way to structure problem definitions for more difficult problems.
Substance-Field Analysis (Su-Field) Models a problem into three components for breakthrough thinking with regard to system structure and energy sources.
Patterns/Lines of Evolution Descriptions of the sequence of designs possible for a current design. One line, for example, describes the evolution of a system from a macro level to a micro level. Illustrating this is the system used to support and transfer glass during the manufacturing process. Over time, smaller and smaller rollers were used until eventually the use of molten tin (i.e., rollers at the molecular level) was incorporated.
Inventive Principles & Contradiction Table
Design contradictions between two performance parameters (drawn from a total of 39) may be resolved by using one or more from among 40 innovation principles. Successfully-used principles for 1201 contradictions are presented in a matrix. As an example, if the design effort to reduce the weight of a moving object is detrimental to force, the matrix suggests using principles
8 Counterweight (i.e., a lifting force),
10 Prior Action (arrange items beforehand),
18 Mechanical Vibration (use resonant frequency), and 37 Thermal Expansion (change dimensions using heat).
Separation Principles The simultaneous occurrence of two mutually exclusive conditions can be resolved using the separation principles. For example, in a coating process which entails dipping parts in a bath of coating solution, the throughput can be increased by increasing the solution temperature. Doing so, however, reduces the usable lifetime of the solution bath. The principle of separation in space suggests that the bath solution be cold but that the part be hot.
76 Standard Solutions Generic system modifications for the model developed using Su-Field Analysis. For example, if a system which employs a so-called mechanical field is not effective enough, changing to the use of an electrical field -- a major paradigm shift -- is suggested.
Effects Physical, Chemical, Geometric, and other effects offer "free" resources commonly forgotten and sometimes even incompatible with the system as designed. The Seebeck Effect describes the phenomenon whereby an electrical flow is created in the presence of two different materials at two different temperatures. This effect can be used to generate electricity in remote locations such as on a space satellite.
System of Operators Universal operators are recommendations which are potentially applicable to any situation, such as Excessive Action (e.g., painting can be accomplished by dipping the component and spinning off the excess paint). General operators are recommendations applicable toward improving functionality and eliminating undesired effects, such as Elimination of a Harmful Action (e.g., through isolation). Specialized operators are used to improve specific parameters or features of a product/process, i.e., Improve Useful Actions (examples include those for increasing reliability).
All off the TRIZ knowledge-base tools yield concepts which require system design and engineering to satisfy the needs of the current innovative problem. Taguchi's approach to robust designs facilitates the engineering of the system.
The Essence of Taguchi
Classical Quality Control uses upper and lower specifications as boundaries between acceptable and unacceptable performance. Performance between the specifications is considered equally acceptable. Taguchi states that the customer's increased dissatisfaction with any deviation from the target value can be approximated by a quadratic function. The limits are not the engineer's specification but the customer's tolerance limits. The lower customer tolerance limit is shown as LCT in figure 1.
Variation is now critical, as seen by the average loss (average customer dissatisfaction) calculation. The average loss is the product of a constant times the product of the variance and the square of the deviation from the target (T).
The basic building block of Genichi Taguchi's philosophy is the idea of reducing the effect of uncontrolled factors while assuring the performance is on target. Which performance (A or B) is best is now dependent upon both the average and the variation. Uncontrolled factors are the internal variations of the components or the external variation of the environment. The concept of the loss function use the electrical engineering signal-to-noise ratio used to, maximize the ratio of useful energy to wasted energy.
Parameter investigations in the laboratory must be conducted in controlled chaos. An example of this is to vary temperature, humidity and component precision -- parameters which are normally controlled -- during data collection to simulate the variance which would naturally occur. Design values are thus obtained which result in close-to-target performances and which have little variation when artificially generated variation is introduced. A system which incorporates these values will not be sensitive to naturally-occurring sources of variation.
If a designed experiment is conducted using eight different combinations of design variables the artificial sources of variations would be the result of several combinations of noise factors. The noise factors could include different levels of humidity, two different ages of materials and a variation in components.
Dynamic applications of Taguchi's thinking are useful for future models of a product. The Ideal Function of a design represents the theoretically-perfect relationship between performance and a signal input to a device. In the case of a spring, the ideal design would provide a linear relationship between applied force and the length of the spring, and ignore the effects of fatigue, temperature, etc. (as we so often did in school). Dynamic applications look for design parameters which increase the linearity of the relationship by making the response independent of the sources of variation. These applications are particularly important for measurement devices and control of manufacturing processes.
Taguchi methodology assumes that both the ideal function and the system design are known.
The QFD, TRIZ and Taguchi Synergy
QFD, TRIZ and Taguchi's methods fit together like a three piece jigsaw puzzle (figure 2) to form a complete picture of the design process. Missing from QFD is bottleneck engineering and optimization. Bottleneck engineering can be overcome with the solution concepts generated via TRIZ. TRIZ is weak, however, in the areas of customer-driven requirements and optimization. QFD provides the customer input and Taguchi provides the process for determining the best parameter values for a
robust design. Taguchi's methods lack the customer-driven priorities and the tools required for system definition. These are provided by QFD and TRIZ, respectively.
The linkage is rather elaborate and can recycle through the sequence of QFD, TRIZ, Taguchi several times. Using the five-step model presented in "Step-by-Step QFD: Customer-Driven Product Design," the process starts with customer analysis. The voice of the customer is prioritized and sorted into Functions, Demanded Quality, Performance Measures, Failure Modes, Concepts, and Manufacturing.
The gathering of data for the customer context table should be attentive to the needs of both TRIZ and Taguchi. TRIZ cultivates a thorough understanding of the constraints, resources, historical solutions, and harmful and useful functions of a system. The visit to the gemba is the best source for this information. Taguchi's method requires an understanding of the sources of variation and freedom to change design parameters. Gathering this information at the gemba is the ideal source.
If QFD is applied to an existing design, and the organization chooses to first use the House of Quality, then the TRIZ methodology reduces the constraints identified in the roof of the house. Often an existing design contains conflicts between performance measures used to evaluate the design. These same performance measures are used during technical benchmarking. Taguchi suggests that sources of variation should be included to obtain realistic data. The TRIZ methodology can be used to protect a company's design by providing for a patent fence around a competitor's design.
How QFD is helped by TRIZ and Taguchi
The five-step model of QFD will be used with additional comments for comprehensive QFD. A count for the number of impacts TRIZ and Taguchi have on the QFD are recorded after each example.
The first step is the identification of customer segments, selecting criteria for ranking the segments, and ranking the segments. Recognizing the power of TRIZ to improve designs never considered becomes possible, and the new demands are possible to satisfy.
TRIZ (1) allows a more aggressive attitude because of the possibilities offered by the technical lines of evolution.
The second step is to understand the customers needs and environment by going to the gemba. The customer's needs are sorted into demanded qualities, functions, reliability issues, solutions, safety and failure modes. Because "ideality" utilizes resources, the voice of the customer context table should add the system and environment resources to the "who," "what," "where," "when" and "how" information.
The experienced practitioner of TRIZ (2) would look for the resources available to provide for a more ideal design, as well as consider historic constraints, and useful and harmful functions.
With Taguchi (1), identifying the sources of variation and the ideal function offer another perspective.
The third step maps the subjective demanded quality information of the customer into the objective measures of performance used by the engineer. The matrix used is often called the House of Quality. If a product is a model upgrade, it is important to identify conflicts between different performance measures. These conflicts influence the compromises in target values set after doing a technical benchmarking.
There are several aspects of TRIZ which can help these activities. The most profound is to ignore the conflict because, with TRIZ, the degree of conflict is often reduced.
With TRIZ (3), the conflict is not used to establish compromises in performance, but to start the TRIZ analysis for physical and technical contradictions.
Taguchi (2) can take advantage of the positive interactions.
Part of the thought process for selecting target values is the relative performance of the existing design against the competition, and the anticipated direction of the competition. The lines of evolution used to create new designs by TRIZ can be used to forecast the competition's future designs.
TRIZ (4) can be used to resolve conflicts both technical contradictions and physical contradictions and thus the setting of target values need not be compromised.
This allows for the application of patent fences to stop the competition and/or protect the existing and future designs.
Taguchi (3) offers the loss function as a more effective measure for technical benchmarking.
Many organizations stop after completing the House of Quality and continue with their existing design process.
TRIZ (5) improves manufacturing equipment by looking at problems and conflicts.
TRIZ (6) improves the manufacturing process by looking at process evolution.
TRIZ (7) reduces cost via ideality, which looks at system and environmental resources.
The fourth step is concept generation. This is one of the most powerful aspects of TRIZ. The inputs for this analysis are the target values for the performance measures, priorities and conflicts. The reliability, manufacturability, cost, and environmental impact can be added to these.
TRIZ (8) would generate many alternatives as defined by the performance measures.
In TRIZ (9), different lines of evolution identify possibilities for designing tomorrow's product today, and thus bypass the competition's existing patents, and excite the customer in the Kano sense.
Taguchi (4) will determine the best design values for a robust design for each considered concept.
Step five looks at an organizations knowledge base of their manufacturing process. They have concerns with the process, the equipment and the capability of their current technology. Each of these can be significantly improved by TRIZ and Taguchi.
TRIZ (10) provides for a search for technologies which can improve the products made from one manufacturing process, thereby increasing the breadth of the product base.
With TRIZ (12), the design of the equipment used to make a product can be enhanced by looking at the evolution of the equipment. This often results in a quantum jump in equipment design, thereby improving product quality and profitability.
Using Taguchi's (5) robust operating conditions for the important operating conditions is only a beginning.
Taguchi (6) provides for finding the appropriate signal to select the exact desired output for the production process.
With Taguchi (7), developing a database upon designed experiments for important operating conditions is preferable to depending on a soon-to-retire employee.
A total of twelve TRIZ hooks and seven Taguchi hooks are identified for five-step QFD.
Comprehensive QFD has more depth in the analysis. The flowchart looks very much like Bob King's matrix of matrices, but is a true flowchart. Kings matrices are just a listing of possible analyses. The TRIZ and Taguchi hooks in the five-step QFD are also present in the comprehensive QFD. The additional hooks in the comprehensive QFD were mentioned previously as activities to be continued in an organizations design process.
Comprehensive QFD expands the beginning and the end of the process by using the Seven Process Tools in the beginning and including delivery and service at the end. The middle analysis is increased by a factor of four.
The Ideation Methodology has taken all the pieces of so-called "classical" TRIZ and integrated them into a unified whole. The analytical and knowledge-base tools have been expanded and modified to eliminate the weaknesses of classical TRIZ. Ideation has developed additional tools for comprehensive and directed product evolution, for eliminating future failures, and for formulating problems associated with products, processes, services and interpersonal issues.
Comprehensive QFD includes all the other information identified and sorted from the customer. Functional analysis links the demanded quality of the customer to system functions and rearranges the priorities accordingly.
Using TRIZ (13) and Ideation's Problem Formulator, normal function analysis is extended by identifying harmful effects and defining related problem statements.
Designing a world-class, customer-driven product is possible by using:
TRIZ (14) as incorporated in Ideation's directed product evolution in the concept generation stage.
Rather than belabor the point that there are many instances in which TRIZ can complement QFD, it can simply be noted that anyone who has been involved in a QFD project recognizes that there are many times that the project team is searching for direction. TRIZ with the formulation process and the structured way of searching through the 400 system operators with 1,600 links forming associative chains guide the design team toward increasing a system's ideality.
Dr. John Terninko
John has taught QFD and Taguchi philosophy to corporations in North America, Central America and Europe for 12 years. He is a recipient of the 1985 Taguchi Award for promotion and application. He has integrated his diverse experience base (electrical engineering, operations research, organizational development, teaching, and management consultation) to develop an approach to the problem-solving required for QFD and utilizing the synergy with Taguchi's philosophy. In 1995 he began integrating TRIZ into QFD.
John has applied QFD to health care, the automotive industry, consumer products, service industries, and durable products. He adapts the QFD process to suit the unique needs of each project.
John has provided TRIZ training and consultation in North America and Europe for both manufacturing and design audiences.
John is a principal with Responsible Management Inc., cofounder and director of the QFD Institute, cofounder and director of the QFD Network, and director of education, training and publications for Ideation International Inc. John is also a TRIZ specialist.