INDUSTRY NEWS

Inspiration Resources for Development & Design

Competitive Advantage Through New Product Development (Unifi)

New Product Innovation in Children's Wear (VF Playwear)

NCRC Partners Center of Excellence has Grand Opening

Design For Six Sigma versus Traditional New Product Development:
Just Hype or Fundamentally Different?

RESEARCH NEWS

Plasma Treatment of Textiles: Changing Fiber Surfaces

RELATED ARTICLES

National Textile Center (NTC) Review - 10th Annual Conference

INSIDE TATM

TATM Welcomes Nancy Powell - Textile Designer

New Curriculum in Medical Textiles

Anni Albers Scholars Program - Textile Design

TATM Graduate Program

College of Textiles Distance Education TOTE Program

Design For Six Sigma versus Traditional New Product Development:
Just Hype or Fundamentally Different?

Dr. A. Blanton Godfrey, Dean and Dr. Timothy G. Clapp, Professor
College of Textiles, North Carolina State University

In the past few years there have been many stories about Six Sigma Quality™ fueled by rather remarkable publicity about the success of General Electric, Honeywell, and other companies. Most companies focused their efforts on cost savings and cycle-time reductions (Table 1), but a few expanded their efforts into design, often called DFSS or Design for Six Sigma (Borgtsen, 2002). Case studies have appeared in numerous journal articles and even in the Wall Street Journal and the New York Times. Is there something fundamentally different about DFSS or is it just the most recent evolution of new product development methods?

Design for Six Sigma is a rigorous approach to designing products and services and their enabling processes from the beginning to ensure that they meet and exceed customer expectations and outperform competitive products and services. Design for Six Sigma has in part become a design methodology of choice because of the extraordinary success some companies, such as General Electric, have had with its implementation. The resulting publicity has influenced many corporate leaders. DFSS has helped create in many companies a successful corporate environment based on data driven decision-making. Data driven decision-making has moved the decision point from experience and "gut feel" toward facts. The need for data driven decision-making is most critical in the new product development process when product characteristics are defined. Experience and gut feeling have positive aspects but they are inefficient when: (1.) the R&D organization works far from the market and lacks an understanding of customer needs and requirements, and (2.) turnover of staff hampers the organization's experience base. Real data and efficient measurement methods give unbiased facts where assumptions and opinions are eliminated. The challenge is to measure, analyze and interpret relevant data using suitable tools and methods, to make management decisions, create design specifications, and to develop processes to produce goods and services that meet and exceed customer expectations and give an edge in the competitive market place.

It is not surprising that there is debate about difference in the Design For Six Sigma (DFSS) and traditional New Product Development (NPD) processes. Almost all companies implementing DFSS have built it on top of their existing new product development process. Traditional NPD processes include Stage Gate (Cooper, 2001) and Product and Cycle-time Excellence (PACE) (McGrath, 1996). GE has formalized its DFSS process and patented their process for implementing DFSS (Martin, 2001). Some have integrated several new ideas into their existing new product development process and called the resulting new process DFSS. For example, Ford Motor Company has used inventive problem solving, axiomatic design, and DFSS concurrently (Smith, 2001). To further complicate the matter, many of the tools used in DFSS are not new. Some companies as part of their new product development processes have used most of these tools. Another complication in discussing new product development is that in most companies there is not one standard process. There are often different processes in different divisions. There are different processes for developing software and hardware. There are different processes for developing new services and products. Furthermore, many companies do not have standardized new product development processes even in divisions or product lines. Different teams of developers may use quite different development processes.

Therefore, one of the main differences we see in DFSS is the formalization of the development process with standardized tools and a structured development process that is taught in similar way across the company. This new process is often used for products and services and for both software and hardware. It is often also used for process design. The details of the application may vary, but the organization applies the steps of the process in a similar, structured way. A second major difference is the use of sophisticated statistical tools. Most of these tools are not new, but they have not been widely used in the new product development process by many companies. Other non-statistical tools are also formalized and used in the process in a structured way. A third major difference is the intense structured training all design teams receive in the DFSS process. This is quite different from the usual practice of introducing one new tool or method at a time and giving complete freedom to design teams on which methods and tools they use and when in the process they use these tools and methods.

The training is usually four or five weeks spread across four to six months. The design and development teams are trained as teams while developing a new product or service. The first week is focused on DEFINE. During this week the emphasis is on market research and Quality Function Deployment (QFD). A preliminary Failure Mode and Effects Analysis (FMEA) is usually completed. One of the main goals of this week is to create a basic product or service definition along with the objectives and a first draft of the driving customer requirements called the Critical to Quality list (CTQs). The teams also start weighting the CTQs using information from customer surveys, competitive analyses, and knowledge of the market and market trends. During the first week the team also completes a first draft of a detailed project plan, the first draft of a functional process diagram, and the product or service functional block diagrams.

The second week is MEASURE. During this week the instructors present a formal, but brief, introduction to probability and statistics with an emphasis on variability and the components of variation. Many of the measurements used in the first week to characterize customer needs and wants and competing product performance are questioned. Many of the basic assumptions used in preliminary design decisions have used existing measurements or measurements provided by key suppliers. These are questioned and a formal Measurement Systems Analysis (MSA) is performed for critical areas. During this week it is not unusual for the design team to experience some uncomfortable surprises. They often find that measurements they have been using for several past designs are so unreliable as to be totally useless.

The third week is ANALYZE. During this week many standard statistical methods are introduced (or for some designers reviewed). These include one and two-sample t-tests, Chi-Square and F tests for comparing two variances, and Bartlett's and Levine's tests for comparing multiple variances. Analysis of Variance methods (ANOVA) are also introduced for comparing multiple means. Correlation and regression are introduced to begin exploring functional relationships. The focus of much of this work can be explained by a simple formula: Y = f(X), where Y is the Critical to Quality (CTQ) identified earlier as a priority and X is a vector of input variables thought to be related through a function f to the performance of Y.

Let's illustrate this idea with a simple example. Suppose the breaking strength of a warp yarn is one of our CTQs. The breaking strength may be a function of many different possible variables: type of fiber, linear denisty, twist, processing, sizing, etc. The designer wants to identify not only which variables (the Xs) determine the final breaking strength (the Y), but the designer also wants to know the functional relationship (f). The designer is also interested in eliminating those input variables (the Xs) that have little relationship and can be ignored. The designer is also interested in making many comparisons and design decisions. Statistically comparing both the differences in averages and variances of different input conditions or variables allows the designer to make fact-based decisions.

Another major difference in DFSS and traditional product development processes is the emphasis not just on the product or service design but also on the processes by which these products or services will be produced. Design for Manufacturability or Assembly methods (DFM and DFA) are incorporated as are their service process counterparts (design for operations). During the ANALYZE week the fact that there are always numerous design alternatives is stressed. In traditional product development too often the first idea is pursued or ideas selected by opinion rather than by formal, structured statistical tests. During this week a simulated process capability study is usually performed.

The fourth week is DESIGN. During this week many of the key design decisions are made. The main statistical methods taught and used during this week are experimental designs. Both reliability prediction and reliability estimation are included, and a formal FMEA is completed. It is during this week that the design team members learn how to select the optimum (or near optimum) levels for the critical input variables (the Xs) to optimize the outputs (the Ys). For most products and services, there are a number of CTQs (the Ys). Sometimes there are conflicts, and the design teams have to make compromises among the CTQs. Using our example above, the maximum breaking strength available for our warp yarn may be in a yarn with poor dyeability characteristics. It is during this week that the design teams learn many new tools to help them make trade-off decisions. They end the week with a plan for completing the specifications of all of their Critical to Quality variables (the Ys) and for the critical input variables (the Xs).

The fifth week is VERIFY. During this week many of the analyze methods and tools reappear as prototypes are tested and compared to objectives and competing product designs. A major part of the week is developing the control plans and standard operating procedures that will be handed off to the operating forces who will produce, deliver and perhaps install and maintain the product or service. Advanced techniques used by many organizations at this stage include accelerated life testing, process simulations, and reliability predictions. Some organizations will also develop the reliability estimation plan during this step and decide what data will be collected, the sample in the tracking study, and the persons responsible for analysis of the data and reliability growth plan.

In each week of training, a similar teaching method is used. New concepts, methods, and tools are introduced and immediately examples and case studies are used to show how these methods and tools are used. Software tools support all of the exercises. The exercises are first done following step-by-step examples by the instructor, and then small groups work together on the next exercises. Then the actual design and development teams work together on their own product and process developments. Sometimes during Black Belt DFSS training there are only one or two people from a project. Then these people have the responsibility of training their other team members (sometimes called Green Belts) in the weeks between their own classes.

Although some organizations have implemented DFSS as an evolutionary step in their continuing improvement of their new product development process, there are some fundamental differences. For most organizations, DFSS is far more documented, formalized, and structured than their former development process. There is a far more formalized emphasis on understanding customers' needs and wants and determining the Critical to Quality variables (the CTQs). A formalized process is used for collecting data on existing products, competitive products, and trends. Far more sophisticated statistical tools are used - all with standardized, easy-to-use software support. Training is done in a learn-do-do-do fashion with actual development projects being carried out during the training. The design teams are expected to have a much broader role ranging from the early market research through concepts, basic designs, detailed designs, production process, design transfer, operating procedures, and field tests and support.

In summary, companies with an informal, unstructured new product development process will see DFSS as a revolutionary approach for developing new products. Companies with a formal new product development process will view DFSS as the next step for improving their new product development process. In either case, DFSS can be successfully deployed in an organization to reduce the risk of new product failures and thereby increase the efficiency and speed at which new products are developed.

REFERENCES

• Borgsten, Jonas E. (2002), 6s in New Product Development, Masters Thesis, The Royal Institute of Technology, Sweden.
• Cooper, R.G, (2001), Winning at New Products- Accelerating the Process from Idea to Launch, 3rd Edition, Perseus Publishing, Cambridge Massachusetts.
  
• Martin, Arlie Russell et al. (2001), System for Implementing a Design for Six Sigma Process, US Patent Number 6,253,115.
  
• McGrath, M. E, (1996), Setting the PACE in Product Development, Rev. Ed, Butterworth-Heineman, Boston, Massachusetts.
  
• Smith, Larry R. (2001), "Six Sigma and the Evolution of Quality in Product Development," Six Sigma Forum Magazine, Volume 1, Number 1, November 2001, pages 28-35.

AUTHORS
Dr. A. Blanton Godfrey, Joseph D. Moore Distinguished University Professor
Dean, College of Textiles
North Carolina State University
Raleigh, NC 27695
blanton_godfrey@ncsu.edu

Dr. Timothy G. Clapp, Professor
Textile Engineering, Chemistry, and Science Department
North Carolina State University
Raleigh, NC 27695
timothy_clapp@ncsu.edu

Table 1. Fortune 500 companies using Six Sigma (Borgsten, 2002)

Company Grade Comment

* Lowest level of implementation, ** moderate level, and *** highest level.

RETURN TO TOP