Top Banner
6 Design for Six Sigma (DfSS) 6.1 PRESENTATION The introduction lecture on this subject has been presented by Leszek Biskup-Köstner from Siemens. Leszek defined Design for Six Sigma as a method to design a new process/product (or redesign a fundamentally non- competitive process) to satisfy customer requirements. The need to apply the Six Sigma approach more upstream than in production is well-known to come up after clearing up the mud. This is well-illustrated with the tree in Fig. 6.1. The phrase low-hanging fruit has been quite often used at the first conference during warning discussions about making the proper choice of BB-projects. Fig. 6.1 Famous tree to illustrate the classes of projects to attack After solving and root-analysing the “ground-fruit” and “low-hanging fruit” problems, the insight takes place that more than 60% of the roots of the problems can be found in the development phase. At Philips we learned that also from our experience with IOA (Industrial Opportunity Assessments). Fig. 6.2 is based on data from about 200 audits internal at Philips [1].
14

6 Design for Six Sigma (DfSS)

Nov 28, 2014

Download

Documents

Sixsigmacentral

 
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: 6 Design for Six Sigma (DfSS)

6 Design for Six Sigma (DfSS)

6.1 PRESENTATION

The introduction lecture on this subject has been presented by Leszek Biskup-Köstner from Siemens. Leszek defined Design for Six Sigma as a method to design a new process/product (or redesign a fundamentally non-competitive process) to satisfy customer requirements. The need to apply the Six Sigma approach more upstream than in production is well-known to come up after clearing up the mud. This is well-illustrated with the tree in Fig. 6.1. The phrase low-hanging fruit has been quite often used at the first conference during warning discussions about making the proper choice of BB-projects.

Fig. 6.1 Famous tree to illustrate the classes of projects to attack

After solving and root-analysing the “ground-fruit” and “low-hanging fruit” problems, the insight takes place that more than 60% of the roots of the problems can be found in the development phase. At Philips we learned that also from our experience with IOA (Industrial Opportunity Assessments). Fig. 6.2 is based on data from about 200 audits internal at Philips [1].

Page 2: 6 Design for Six Sigma (DfSS)

Fig. 6.2 Root causes of major changes in production

The most impressive picture presented by Leszek was Fig. 6.3, because it illustrates very well the use of Sigma level at the lowest possible level of “steps”, i.e. parts and process-steps. By accepting a too low sigma level at these details you fall into the old pitfalls.

Design - SimulationDFSS Steps

Spec Limits

USL LSL

Standard

DefinitionCTQStep

Step

Sigma

ProcessSigmaOK?

VERIFY

Redesign

Yes

No

I.e. design Scorecards

Fig. 6.3 Design Scorecard, showing the deeply deployed Sigma level

as key quality parameter In analogy with the DMAIC phases in Six Sigma for manufacturing the following five steps for Design for Six Sigma are defined (shown in Fig 6.4):

n Define n Measure n Analyse n Design n Verify

Page 3: 6 Design for Six Sigma (DfSS)

Fig. 6.4 Overview of the structured five-step approach

of Design for Six Sigma

Table 6.1 Activities to be done & entities to be taken care of during Define-stage

1. Define1 1.1 Strategy

• Business goals • Project goals • Risk Analysis • Influence of Customers and Processes

1.2 Concept • Goals and scope • Customers • Customer requirements (high level) • Technical requirements • High level plan • Problems

1.3 Project plan • Resources • Project requirements • Time line • Milestones • Authorisations

So, write down why you are doing the project and how you split it up do be able to do your project management properly. This is very common, the 1 The original sheets (see Appendix 4) mention also the tools to be used in each stage; no unknown tools were mentioned. Simulation is highly advocated.

Page 4: 6 Design for Six Sigma (DfSS)

extra Six Sigma kind of start-up a project can not be read in these phrases, but if you combine the column “step sigma” from Fig. 6.3 with this Table than you realise that the intention is to have the goal of planning for Six Sigma in mind together with all “common” steps of the Product Creation Process.

Table 6.2 Activities to be done & entities to be taken care of during Measure-stage

2. Measure 2.1 Customer Segmentation

• Identify Customers • Segment Customers • Prioritise Customers

2.2 Customer Requirements • Customer Selection • Data Collection • Prioritisation • Interviews

2.3 CTQ & Needs Analysis • Prioritise Customer Needs • Determine CTQs • CTQ Analysis • Risk Analysis

Understanding Customer needs was stated as being pivotal to a successful project. Customer Segmentation, surveys, CTQs2 (Critical to Quality) specification and using a structure tree for overview as preparation for building a House of Quality (QFD) were mentioned as the detailed steps in the Measure Phase. I like to add a proper use and understanding of Kano's model in this phase as well.

Table 6.3 Activities to be done & entities to be taken care of during Analyse-stage

3. Analyse 3.1 Process Requirements

• Define processes/product functions • Allocate CTQs to process/functions • Benchmark best performance and

processes 3.2 High Level Design

• Process Description • Deploy process/functional requirements to

design requirements • Define critical resources

3.3 Process capability • Define evaluation criteria • Obtain customer feedback • Finalise design requirements

2 More or less the same as the key parameters in the APQP (Advanced Product Quality Planning) approach of Ford from the early 1990s

Page 5: 6 Design for Six Sigma (DfSS)

• Risk Analysis

Starting with the overall requirements (i.e. CTQs) you will identify the functions required of your product or service.

Page 6: 6 Design for Six Sigma (DfSS)

Table 6.4 Activities to be done & entities to be taken care of during Design-stage

4. Design 4.1 Process Design

• Detailed process description • Define critical to process (CTP) • Control points, measurements

4.2 Simulation • Determine process capability • Process simulation • Risk analysis

4.3 Verification Plan • Develop control strategy • Develop control strategy • Develop pilot test plan

The design step is similar to the analyse step with the major difference being the level of detail of your design activities. In this step you get a good overview which of the steps/parts will be critical if you have used the design scorecards (see Fig. 6.3) properly. This is the most important extra that Six Sigma has brought above the normal way of working.

Table 6.5 Activities to be done & entities to be taken care of during Verify-stage

5. Verify 5.1 Execute Pilot

• Pilot testing • Documenting results

5.2 Analyse Results and Implementation • Compare results to specifications • Start-up and testing • Training

5.3 Project Closure • Handout process documentation • Transition to process management • Project closure

The planned workshop after the introduction lecture on DfSS was skipped because of the not-planned Six Sigma Club discussions.

6.2 AWARENESS WORKSHOP

Our Process Control group at CFT advocates and practices already for more than ten years that you must move up-stream with improving activities. So, I agreed completely with the presentation of Leszek, but did not find a really new approach and/or new tools. As a result of mentioning this during informal discussions I got a copy of the sheets of an internal GE DfSS Awareness Workshop. Fig. 6.5 shows the place of this DfSS Awareness Workshop in the total Training Roadmap. Also the Six Sigma Institute of Mikel Harry is willing to train you in DfSS (Fig. 6.6). From this bundle of

Page 7: 6 Design for Six Sigma (DfSS)

sheets3 I have taken some sheets to give extra focus on the need to apply Six Sigma in development as well.

Fig. 6.5 Example of GE Capital Quality Training Roadmap

3 Copies of this internal GE DfSS Awareness Workshop can be ordered from the author.

Page 8: 6 Design for Six Sigma (DfSS)

Fig. 6.6 Training program of the SSDI4 institute

Of course the workshop starts (see Fig. 6.7) with rephrasing the importance of Six Sigma within General Electric, but Jack Welch also states that, if you are not convinced, ".. you should take your skills elsewhere". Jack repeated this message in USATODAY (1998-02-27): “With Six Sigma permeating much of what we do, it will be unthinkable to hire, promote or tolerate those who cannot, or will not, commit to this way of working”. Quite clear!

Fig. 6.7 Earnings at GE, but also a quite serious warning

4 Six Sigma Design Institute from Dr. Mikel Harry and Dr. Douglas Mader

Page 9: 6 Design for Six Sigma (DfSS)

Fig. 6.8 Detailed structure of the DfSS process at General Electric and the preferred tools to use

The DfSS process at GE is structured and very detailed, and it looks as if nothing new can be read there. However, let us go into more detail for the first step. In the Identify phase, customer wishes and product requirements need to be identified to find all CTQ (critical to quality) variables. Not only their target values, but also their limits. The broader definition of CTQ is:

Those characteristics of an item which, if non-conforming, may prevent or seriously affect the unit performance, reliability, producibility, or customer satisfaction of a component.

In my experience, QFD is seldom used and if it is, the action is stopped after the first house. In this workshop, QFD is called the system for translating customer requirements into company requirements at each stage from research to product development, to engineering, and manufacturing to marketing/sales and distribution. Applying the QFD tool in its full strength delivers critical-to-quality characteristics (CTQs) from the second house, key manufacturing processes from the third house and key process variables from the fourth house. By the same deeply deployed use of the FMEA and RCA tools, the change of forgetting a CTQ is minimised. The sheets of the workshop illustrate that, after finding all CTQs in the identify phase, the root causes of the CTQs must be found in the design phase, using e.g. simulation and finite element methods (FEM) to model the relations.

Page 10: 6 Design for Six Sigma (DfSS)

A more general picture (Fig. 6.9) lists the three main classes of questions to be asked: • Which parameters have great influence on Customer Satisfaction? • Which failure modes can be expected? • What failures did already happen and how can we prevent them? The tools GE uses are not new (Fig. 6.9), they are operational at Philips as well. However, the drive to find all CTQs and to be only permitted to continue if the sigma level is acceptable (see also Fig. 6.3) quit rightly gives this approach the name Design for Six Sigma.

Fig. 6.9 Three main tools to identify CTQs (=parameters critical to quality)

Looking for a metric to follow the implementation path through the entire company, GE "only" uses the percentage of new drawings reviewed for CTQs and the classification in simple sigma level categories of the so chosen parameters. This still does not look very dramatic.

Page 11: 6 Design for Six Sigma (DfSS)

Fig. 6.10 Six Sigma Design metrics

However, the next slide makes it impressive (of course you can ignore the message by convincing yourself that the figures are window dressing, but please keep in mind that the sheets are taken from an internal workshop), because of the speed of improvement. The change of 59% of all drawings reviewed for CTQs via 73% to 84% is unbelievable large. And also the improvement in sigma level from the bad category smaller than four to the next category is quick.

Fig. 6.11 Overview of the performance during the first three quarters 1996

at GE That's the difference between a lot of the Six Sigma companies and their competitors: the knowledge is (maybe) the same, but GE (and others) really DO it.

Page 12: 6 Design for Six Sigma (DfSS)

6.3 DFSS AT THE 1999 ASA QUALITY & PRODUCTIVITY RESEARCH CONFERENCE

Preparing myself for the Lloret conference, especially for the DfSS subject, I downloaded several presentations from the 1999 ASA Quality & Productivity Research Conference. Reading these lectures strengthened my feeling that the Six Sigma approach is having quite an impact, but did not bring very much new with respect to methodology. But as with the scheme of Fig. 6.8 it is easy to overlook the depth of DfSS, so I want to mention two points: 6.3.1 Fast Probability Integration 6.3.2 Engineering product

Fig. 6.12 Fast Probability Integration is a numerical method to speed up simulation

6.3.1 Fast Probability Integration

Fast Probability Integration is a numerical approach for probability estimation, based on the approximate response surface and first-order (mean) and second-moment (standard deviation) representations of the uncertainties. Liping Wang [10] claims to do a two-minute FPI run instead of an eight-hour Monte Carlo run for the same simulation (robust high-pressure turbine efficiency) and even 65 minutes instead of a ten-day run (GE aircraft engine study).

Page 13: 6 Design for Six Sigma (DfSS)

6.3.2 Engineering product

Gavin Finn introduces the concept op “engineering product” in his article [8] as follows: "Advances in technologies for the engineering design process have brought this part of the product realization process far forward enough to be ready for Six Sigma approaches for engineering. New design and engineering technologies have done for the engineering process what Henry Ford did for manufacturing. No longer is the development of a new model (or new product) the handiwork of a lone, skilled artisan. Now, teams of engineers and designers, using common tools and methodologies collaborate in a highly orchestrated process to conceptualize, detail, specify, and build almost every common and uncommon product, from computer to toys. The evolution of new work methods and tools for the engineering design process has resulted in the creation of an engineering product, namely the digital product model. This product can and should be subject to the rigor of a Six Sigma quality assurance process in much the same manner as a physical product would be. This engineering product, or virtual product, saves time in product development by eliminating the need for a physical mock-up, and allows for early detection of interferences between components, and a number of trade-off, or optimization studies. The approach suggested here is to focus a Six Sigma program on the digital model, in addition to the Six Sigma programs for the manufactured product. This new quality focus will, by virtue of its intrinsic higher quality yield, also improve the product and process quality with respect to the manufactured product." By consequently applying the dpmo philosophy to the engineering product, Gavin illustrates how the Six Sigma way of thinking can be profitably used up-stream long before an actual prototype exists. Gavin states that the available commercial design quality systems are able to analyse the digital model in such detail. For a detailed description I recommend reading the original article [8]. Overviewing the whole concept of Design for Six Sigma, I came to the following conclusion:

Companies that are able to get momentum in Design for Six Sigma will speed up their PCP and avoid transferring from development to production too early.

Page 14: 6 Design for Six Sigma (DfSS)