
「6 Sigma」 is a method for solving scientific problems that uses standardized deductive processes and statistical verification as well as overall financial effect as a basis to audit results. Due to the results seen after its adoption by Motorola and GE, domestic and foreign companies promoted its use and it became widely accepted as a solution of scientific problems.
The basic spirit of Six Sigma is to combine together management worksite experience and verifiable evidence statistical evidence and use them together to solve the problems that occur in the industry. There are five important steps called DMAIC described below:
1. Define: | Choose and verify the problem that you want to work on and perform a cost effect analysis before starting work. |
2. Measure: | Convert the problem into measurable data and collect the above data to measure what the current status of the problem is. |
3. Analysis: | Analyze the above data and use the data determine what major factors affect quality output. |
4. Improve: | Improve the major factors above that affect quality and verify that the defined problem passes through improvement methods that can provide an effective solution. |
5. Control: | Make the improved methods and processes permanent and pass it in writing to the actual project leader. |
In simpler terms, the problem is expressed in the form y = f (x1, x2, x3,…, xn). Of these, y is the defined problem and x1, x2, x3,…, xn are the actual factors that influence y. While exploring the source of the problem and solutions, a continuous process of observation, measurements and verification must be used. The problem (y) is found through observation. The main factors behind the problem (x1, x2, x3,…, xn) are determined through measurement and testing statistical methods are employed to verify the factors that are found and then analyze, improve and control them. Generally speaking, Six Sigma solution method can be summarized into seven items:
- Improvement methods are obtained based on y and x cause and effect relationships.
- Between determining the problem and improvement methods, a solution can be obtained by continually working on and checking the problem.
- Problems need a clear operational definition to have an improvement process that is precisely targeted.
- Problems should be quantified whenever possible to facilitate measurement and verification.
- The solution analysis process must be data driven.
- Using fresh and previously unused methods for solution thinking achieves greater results.
- Hypothesis testing statistical processes must by verified by returning to a practical level.
The ultimate objective when using six sigma methods is continuous improvement of quality to achieve a defect rate of just 3 or 4 parts per million. Some the work is done in lots on the production line, it is almost impossible to produce one million parts at one time. Therefore, this defect rate is almost equivalent to a zero defect level. Using regular statistical distribution charts below such as extending six sigma from the right and left ends, we can clearly that almost all products are included within the conforming product requirements. There are virtually no parts with natural defects so we can at least reach the 3 to 4 part per million defect level.
Six Sigma at the Define and Measure stages is mainly used to all of the existing past experience of the project leader to determine what major factors affect the project target (y). By entering the Analyze, Improve and Control stages, data is needed to determine the analysis and verification data that affects target quality.
All of the data collected at the Analysis, Improve and Control stages must be based on the latest population samples. Therefore, during the sampling process, the random sample concept must be carefully followed. In addition, a sufficient number of samples must be collected within cost limitations to ,show the attribute variations between the samples.
From the samples collected, we can obtain the arithmetic mean (x) and standard deviation (σ). In fact, all statistical techniques are derived from these two numbers. Standard deviation can be thought of as the distance of the sample to the mean. As the data spreads (disperses), each sample has a distance to the mean. This distance is expressed as xσ. So, 1σ means that a single sample distance to the mean has a standard deviation of distance. 2σ is the two standard deviation distance from the mean. Based upon the basic logic of normal distribution, the farther the distance from the mean (different from many values) means the data is more special and more bizarre. When the data distance from the mean exceeds 3σ on either the right or left edge (especially bad or good compared to test grade), the data is seen as being an abnormal and special number.
In Six Sigma thinking, we hope that the standard deviation will shrink between mean’s upper specification limit (USL) and lower specification limit (LSL) and 6σ will differentiate from the right and left. This standard deviation shrinkage means that the product variation is being reduced and the homogeneity is increasing. The yield rate will then naturally increase. As the specification range is significantly reduced, the occurrence of strange and bizarre (defects in other words) will become increasingly infrequent. When you reach the 6-sigma level, only 3 or 4 defects per million parts are allowed. This number is real production conditions is virtually at the zero defect level. As the number of defects decrease, quality and cost naturally decline as well. When a company reaches the six sigma level, their Cost of Poor Quality (COPQ) may also drop significantly.
Below is the relationship between each standard deviation and defect quality and cost:
Below is the Six Sigma level, probability, defect number relationship table:
(Source of the above two data tables: James M. Lucas, “The Essential Six Sigma,” Quality Progress, January, 2002)
On top of Six Sigma, there is the Design for Six Sigma (DFSS) concept that is used for product and service process reengineering. Since Six Sigma focuses on process improvement of current processes, DFSS concentrates on process generation. This concept is derived from the belief that quality comes from design and not from inspection or manufacturing. Therefore, the focus of the entire DFSS is how redesign processes to satisfy customer requirements to allow resource inputs that can earn their proper value and produce products and quality that are demanded by the customer.
The belief that quality comes from design undermines the concept that testing or manufacturing determines quality. The focal point behind this is not only reducing COPQ but also putting “prevention is better than cure” into action by strengthening the financial effects of operational performance and effectively achieving the risk control goals of the company. Therefore, as more resources are put into prevention cost, the inspection cost (regular quality assurance work) and external failure cost (regular sales work and customer service responsible) can be reduced so that reductions in overall quality defect costs deliver improved operation results. When the prevention cost are used in the R&D stage of product or service processes, the company’s overall standard deviation level naturally increases and may perhaps reach 6 Sigma levels (only 3 or 4 defects per million parts).
Since all product and design process must be reviewed and redesigned for DFSS, different management procedures gradually develop for theoretic and practical fields due to their different outlook which different from the basic logic of Six Sigma DMAIC. The various types of DFSS induced management processes generally have the following points:
- DMADV Method : Define→Measure→Analyze→Design→Verify
- IDOV Method : Identify→Design→Optimize→Verify
- DMADOV Method : Define→Measure→Analyze→Design→Optimize→Verify
- DMCDOV Method : Define→Measure→Characterize→Design→Optimize→Verify
- DCOV Method : Define→Characterize→Optimize→Verify
- DCCDI Method : Define→Customer→Concept→Design→Implement
- DMEDI Method : Define→Measure→Explore→Develop→Implement
- DMADIC Method : Define→Measure→Analyze→Design→Implement→Control
- RCI Method : Requirement→Concept→Improvement
The above nine methods are more alike than different and they are quite similar to the DMAIC methods. If simply classified, the upper half of DFSS work decides what things should be done to design the right products and services. The focus of the second half is on how to correctly design the right things from the first half. According to Evans & Lindsay(2005), Six Sigma DMAIC methodology was first proposed in 1987 by Bill Smith, a reliability engineer at Motorola. The IDOV version from Goh & Xie (1994) in the reputable publication Total Quality Magazine proposed the use limits for Six Sigma methodology. As for DMADV, it was proposed by Joseph & Zion (2002) in the Journal of Change Management. No matter which method is used, they all must find a way to effectively satisfy customer requirements (search for Critical to Xs: CTX), establish effective procedures that conform to customer uirements and reduce defective quality and cost.
To ensure that DFSS can effectively solve the above three factors, DFSS management processes become extremely important. According to Brue & Launsby (2003), DFSS design processes can be separated into five stages:
- Determining what the actual problem is
- Converting the above actual problem into a statistical problem
- Expressing the statistical problem as a transfer function and clearly defining the Y = f(x1, x2, x3, … xn) relationship
- Statistical solution
- Practical solution
The above Y = f(x1, x2, x3, … xn) relationship for DFSS can define X1, X2, X3, … Xn in physical, mathematical and simulation terms. There are many tools that can be used with DFSS such as:
- Quality function deployment, QFD
- Axiomatic design
- Theory of Inventive Problems Solving: TRIZ/TIPS
- Design for X
- Design of experiment, DOE
- Taguchi methods
To make the entire development process clearer and easier to follow, C2C Solution organized the DFSS process into 14 factors in 2002. Because the content is simple and comprehensive, they are described briefly below:
- Verify project problem and scope (business case & project plan)
- Understand the potential and actual customer requirements (understanding customers better than they do (VOC+ MOC))
- Document and prioritize each customer requirement and periodically update (document & prioritize the customers’ “needs”)
- Deploy requirements based on quality requirements and then convert into executable work for determination of future production targets (develop metrics & set product goals)
- Convert customer requirements into engineering specifications to facilitate design and verification (product function analysis)
- Designs need to conform to design failure mode effect and analysis method to ensure that the possible failure during the product lifespan are effectively controlled (design & process FMEA’s)
- Use “does it add value” standard to examine current design process and then perform the necessary simplification (applying trimming technique)
- Invite personnel to participate in brainstorming to come up with the best solution to improving current processes (25+ strategies for innovation)
- Use various tools to evaluate the possible effects of process improvements and select the best solution (concept selection)
- Use production and design combination (Design for Manufacturability: DFM) concepts, the robust design logic of the Taguchi method and quality function deployment methods to select the most suitable design specifications (detailed product design)
- Use value analysis to examine process appropriateness and necessity to determine whether or not process functions need to be reorganized, simplified, combined or eliminated (process function analysis)
- Use quality function deployment and robust design logic of the Taguchi method to find the best solution for robust design (detailed process design)
- Use various quality control mechanisms to ensure that the previously determined best process and product solutions can continue forward and not be return to the original point (production control)
- Use continuous improvement methods to build results and satisfy the derived requirements of internal and external customers and eliminate possible waste (kaizen)
In other words, DFSS focuses on the redesign of systems and prevention of design defects so performance cannot be measured in the short term. Conventional Six Sigma works on the improvement of current process that can produce results in a short period of time but attention is paid only to eliminating current production defects. Therefore to seek greater performance and a higher 6 Sigma level, DFSS training and requirements have become one leading indicator of a company that continually strives for excellence.
To effectively promotion statistical quality concepts and resolve internal process bottlenecks, TXC Corporation started to promote 6 Sigma activities in 2004 and hired a reputable corporate consulting company to provide counseling. The programs that were a part of this activity included internal process control and external supplier management. R&D, quality assurance, process control and administrative work were all included in the work so that data management, focus on quality and continuous improvement concepts were incorporated into both internal and external work and brought into the entire supplier chain.
When promotion of the activity first started, the company established a 6 Sigma Promotion Committee and appointed a manager to specialize in statistical quality management. A promotion plan was also drafted that created a culture verification standards and verification incentives. At the beginning of 2004, the Quality Assurance Center and consulting company were placed in charge of the planning work. Afterwards, in order to integrate company resources and unify the counseling process, responsibility has been handed over to Management Center supervisor until now. Starting in 2005, the internal resources that were built up were put into use for self-guidance and review of projects. An external management consulting company was hired for verification to internally refine industry skills and external verify review capability.
At the same time, a vote was held to select the 6 Sigma promotion activity logo to focus people around a central goal. After two rounds of enthusiastic voting, the logo recommended Chen Hsien-Chen at the Purchasing Center was clearly favored by the Evaluation Committee and stood out from the rest of the other 18 qualifying logos.
The winning logo incorporated the company’s CIS symbol, the six sided attribute of the company’s quartz crystal products and Greek letter 「σ」(symbol for standard deviation). The logo is surrounded on all four sides by a circular border which symbolizes service processes that are well rounded and bring satisfying results.
The activity has been promoted now for over ten years. As of 2014, eight people have been certified as black belts. Ten have earned green belts and a total of 119 personnel have undergone training. We will continue launching new improvement projects to spread scientific problem solving methods (DMAIC) to each person who works for the company.
To ensure the success of 6 Sigma activity promotion, the company not only established the 6 Sigma Promotion Committee but also entered into cooperation with reputable corporate consulting companies such as Ahead, IEG and BMGI to provide green belt, black belt classes on a yearly and stage-by-stage basis and project guidance. In addition, the company created and revised existing internal management procedures in a steady and gradual manner to clearly define the incentive system and job responsibilities for project leaders. After this, course training was conducted, guidance and verification at each stage of projects was provided and product results were announced. 6 Sigma activity spirit and solution methods were highlighted from each stage so that they could be firmly incorporated into the daily actions of each employee and the effect could spread to other colleagues at work. When the project was formally closed out, the President, Chief Technology Officer (CTO), Vice CTO, QA Center senior manager and black belt supervisor were invited to give a formal report and evaluation, give recommendations and make additional requests. The follow-up tracking of the financial effects were done after the project was closed out and those with significant effects were submitted by the Management Center to the President to be rewarded.
Looking back from the beginning of 2006, all on the projects that passed internal review were shared with personnel that did not participate in project work in order to spread the range of knowledge, strengthen control ability and expand the 6 Sigma platform throughout the entire company. After 2007, we adopted an exact copy concept to spread the 6 Sigma activity to our China subsidiary (TXC Ningbo) to put continuous improvement and data management in effect there. Below are the steps and process used by the company to promote 6 Sigma:
2010 Planning Summary
Promotion of this year’s Six Sigma activities is an extension of the results from the black belt internal training provided by Ahead International Management Consulting Group that was hired in 2009. Five to six present stage customers were chosen from the project bank and improvement work was done on the projects that were most closely related to the customer. Starting in March, the company held a five day green belt training course. After complete the five to six projects at the beginning of September, Design for 6-Sigma (DFSS) training was held for R&D personnel so that Six Sigma solution methods and spirit could be incorporated into the basic corporate culture of our company.