ford dfss

esign for Six Sigma (DFSS) has been publicized as a product develop-ment approach that complements the Six Sigma problem solvingmethodology. Promoted as “Six Sigma goes upstream,”1 DFSS encour-

ages systematic anticipation of customer needs and disciplined application ofscientific and statistical methods to meet those needs. Many organizations,enthusiastic to build on Six Sigma momentum, generated their own DFSSprocesses before a standard template emerged.

While Six Sigma is widely recognized by the DMAIC acronym that repre-sents its five standard phases (define, measure, analyze, improve, control),DFSS has no standard acronym. Therefore, organizations have adopted a vari-ety of approaches that resulted in acronyms such as IIDOV, CDOV, IDOV,DMADV, DCOV and IDEAS.2 Despite these naming differences, all versions ofDFSS share fundamental strategies and tools that promote a common goal: tocreate a data driven product development culture that efficiently produceswinning products.

Initial Assumptions

Ford decided to launch “consumer driven Six Sigma” in 1999, with an ini-tial focus on two things:

1. Training Black Belts (BBs).

2. Completing DMAIC projects. Management set targets for the percentage of each division trained and dol-

lar savings resulting from projects. It created a central, prioritized list of allcorporate quality issues to aid the coordination of improvement efforts withavailable resources.

The implementation strategy assumed people throughout the organization

D E S I G N F O R S I X S I G M A

Design for Six Sigma at Ford

DFSS PROVIDED A

FRAMEWORK FOR

A DATA DRIVEN

CULTURAL CHANGE

IN PRODUCT

DEVELOPMENT.

D

By Nathan R.

Soderborg,

Ford Motor Co.

S I X S I G M A F O R U M M A G A Z I N E I N O V E M B E R 2 0 0 4 I 15

DFSS enhanced Ford’s existingproduct development system.

would be more likely to buy into Six Sigma if resultswere quickly realized from the company’s short-term(four to six month) “find and fix” DMAIC projects.That buy-in was considered necessary to start a culturechange. And while significant energy was focused onachieving the short-term targets, the importance ofdeveloping DFSS for the long term was also recog-nized.

A small development team with representativesfrom various divisions embarked on a plan to developDFSS for Ford. The team held benchmarking discus-sions with early DFSS adopting companies and lis-tened to consultants familiar with those efforts. Muchof what was learned paralleled quality and reliabilityinitiatives already under way at Ford.

The team settled on a four-phase process: define,characterize, optimize and verify (DCOV). The DCOVframework aligned with Ford’s existing product devel-opment (PD) system and built on the disciplines ofsystems, robust and simultaneous engineering, all ofwhich had been in use at the company for more than10 years. Introductory training in related tools wasalready available through the Ford technical educa-tion program (FTEP), an online curriculum all engi-neers are expected to pass (see Table 1).

Implementation began with Ford’s PowertrainDivision, which wanted to apply DFSS to the develop-ment of new engines and transmissions. Work wasdivided into a large number of subprojects with anactive BB assigned to each. Growth in the number ofprojects was driven by a senior vice president, whodemanded regular status reports.

Key to progress was the strategic placement of aDFSS Master Black Belt (MBB) as manager of thedepartment responsible for the computer modeling of

new engine designs. This manager teamed up with anengine program manager to act as working level cham-pions for DFSS. Soon other divisions were launchingDFSS by identifying beta projects and training teams tocarry them out. Issues with the training and executionof these early projects highlighted implicit assump-tions that needed reevaluation. For example:

• The rationale for DFSS would be clear to midlev-el management, and DFSS would be embraced asa natural extension of Six Sigma.

• Projects would easily integrate into the company’sstandard PD process because the DFSS steps hadbeen carefully aligned to that process.

• Similar to DMAIC, steps to project completioncould be detailed in a prescriptive set of proce-dures using a standard toolset.

• The DFSS training should follow the BB model,in which all participants receive extensive train-ing in Six Sigma tools.

The following sections address these assumptionsone at a time, explaining why they were flawed andoutlining implementation changes required to main-tain momentum of the DFSS rollout.

DFSS Rationale

DFSS proponents argue the benefits reaped fromDMAIC defect reduction multiply when Six Sigmarigor is applied to defect prevention. But even when anorganization has achieved significant DMAIC success,management may be legitimately skeptical of DFSS.

At Ford it became clear early on that DFSS accept-ance could not be achieved riding on the coattails ofSix Sigma. Basic questions such as “Isn’t DFSS just goodengineering with a fancy brand name?” and “The toolsare not new—so what is new?” needed to be clearlyanswered up front. To answer these questions, the teamconcluded Six Sigma culture, which establishes statisti-cally verifiable facts as an underpinning of all productdecisions, could be a lever to increase the influence ofFord’s ongoing quality engineering initiatives.

DFSS packages methods and tools in a frameworkthat promotes cultural change under a recognizedbrand name that helps overcome an initial resistanceto change. It is most useful if it generates permanentbehavior changes that outlast its own life as a brand.

Given the DFSS toolset is not substantially new, therationale for DFSS should not focus on tools. Overtime, DFSS at Ford has emerged as a scientificapproach to PD that leverages Six Sigma culture. Ithas become a means to reinstill rigorous deductive

Design fo r S ix S igma at Fo rd

16 I N O V E M B E R 2 0 0 4 I W W W . A S Q . O R G

Table 1. Ford Technical Education Program

Ford technical education program subjectsWeb based, self-paced training with tests

Process fundamentals: DMAIC, DCOV* Reliability

System engineering fundamentals Statistical engineering

Applied consumer focus Experimental design

Design verification and production validation Parameter design

Failure mode and effects analysis Tolerance design

Eight discipline (8D) problem solving Process control methods

* Added in 2004; all other subjects were available prior toDFSS deployment.

DMAIC = define, measure, analyze, improve, control.DCOV = define, characterize, optimize, verify.

and inductive reasoning in PD processes. It requires: • Identifying customer desires.

• Developing validated transfer functions (mathe-matical models) that describe product perform-ance through objective measures.

• Correlating these objective measures to customerdesires and effectively assessing the capability tomeet those desires well before product launch.

• Applying transfer function knowledge to optimizedesigns to satisfy customer desires and avoid fail-ure modes.

Six Sigma culture aids implementation of thesesteps by providing:

• A cross company common language for problemresolution and prevention.

• A mind-set that demands the use of valid data indecision making.

• An expectation across the organization thatresults should be measurable.

• A disciplined project management system to helpachieve timely results.

None of the elements of this approach are revolu-tionary, but together they provide a template for success.Based on experience applying DFSS, Frank McDonald,a VP of Cummins Inc., observed: “[DFSS] may soundlike the latest new way if not for the fact that it really ismore like the development of the old way to a higher

standard that feels a lot like the right way.”3

Recently, some DFSS proponents have contendedthat organizations should begin Six Sigma deploymentwith DFSS instead of DMAIC. Ford’s experience indi-cates otherwise. Short-term DMAIC projects led to rapidelimination of a significant number of issues and result-ed in measurable financial benefits. This success boost-ed confidence in Six Sigma, and continued institution-alization helped build a culture receptive to DFSS.Furthermore, once BBs had successfully completed afew DMAIC projects, they were better equipped to sup-port the longer and more complex DFSS projects.

Project Integration Within Existing Processes

Even when the rationale for DFSS is clear, practicalreservations about its alignment with existing processesmay remain. Ford had been evolving its PD system foryears when DFSS was introduced. Natural fears arosethat DFSS was intended to replace this system, poten-tially causing disruption and confusion. To counterthose fears, senior management made it clear DFSSwould not replace, but would augment, the existing PDprocess in specifically designated areas.

DFSS projects would demand a deeper dive into theestablishment of customer connected performance tar-gets (Y’s) and transfer functions that link those targets tocritical design characteristics (X’s). This approach is con-sistent with the experience of other DFSS implementers:“We shouldn’t tell our engineers we’re discontinuing the


The Powertrain Division was the first to implement DFSS.


process they’ve been working with for 10 years andreplacing it with DFSS. We should integrate DFSS deliv-erables into the current development process and askproject managers to commit to providing them,” saidDoug Mader, president of SigmaPro.4

Integrating DFSS deliverables into the current PDprocess sounds simple, but it involves real challenges.For example, PD entails at least three different cate-gories of scope: the creation of a new product or tech-nology, the evolution of a next generation productfrom a current design or the enhancement of an exist-ing product. The rhythm and timing of DFSS projectsassociated with these efforts will vary depending onthe category (see Table 2).

Boundaries are not always well defined, but the cat-

egories provide initial guidance for establishing proj-ect expectations and timing. Breakthrough projectshave the most design latitude and typically require themost time to complete. Evolutionary projects have lessdesign latitude, and their timing is tied to standard PDmilestones. Adjustment projects have the least designlatitude, and they are generally expected to be com-pleted as soon as possible.

In addition to delivering an optimized design, eachDFSS project at Ford is required to institutionalizelearning and transfer functions by developing designguidelines and training that can be reused in thefuture. This effort to reduce waste in the PD factory iswhat distinguishes an adjustment DFSS project from aDMAIC project.

To reduce waste in the product development factory, each DFSS project develops reusable guidelines and training.

Some practitioners use different process phases fordifferent categories of projects. For example, theauthors of Design for Six Sigma in Technology and ProductDevelopment use invent, innovate, develop, optimize andverify (IIDOV) to describe DFSS for technology devel-opment and concept, design, optimization and verify(CDOV) to describe product commercialization(design).5 To reduce confusion, Ford uses the DCOVabbreviation for all categories of DFSS projects.

A second challenge in integrating DFSS deliverablesinto the current development process is determiningthe manager to which they should be assigned. DFSSprojects typically deal with the most challengingdesigns, such as new concepts or technologies linkedto a compelling business case or complex systems that

have exhibited problems in the past. In the first case,no natural project owner may have yet evolved in theorganization. In the second case, the natural ownermay not be clear or may be disputable because prod-uct attributes and system interface issues often cutacross system boundaries.

Even if the organization dictates a clear owner, thefact existing designs are underperforming indicates aneed for a new approach. In each of these cases, it iscritical to apply basic project definition and manage-ment techniques. These include:

• Ensuring the team is led by someone who hasresponsibility for the design and can engage theappropriate technical expertise.

• Providing mentoring in statistical and PD tools—


Table 2. DFSS Project Categories

Project category Design latitude Timing

I. New product or product breakthrough: Open—exploration of alternative Driven by market opportunities • Introduce new product, concept or technology. concepts allowed. and technology feasibility; linked • Add a new feature to next generation product. to technology development plans.

II. Evolutionary change in next generation product: Moderately constrained— Driven by market trends and• Upgrade performance. concepts often already selected, competitive pressures; linked • Maintain performance while lowering cost. assumptions for component to existing product program • Prevent problems. reuse limit design space. milestones.

III. Adjustment to existing product design: Heavily constrained—limited to Driven by waste reduction goals;• Optimize performance within existing production constraints. “running changes” to existing linked to yearly budget targets,• Reduce current costs. designs. similar to DMAIC projects.• Fix problems.

DMAIC = define, measure, analyze, improve, control.

Ford uses the four-phase define, characterize, optimize, verify (DCOV) abbreviation for all categories of DFSS projects.

for example, assigning BBs and MBBs or theirequivalents.

• Securing an executive Champion to actively sup-port and review the project.

• Committing the team and Champion to a projectcharter that outlines goals aligned with organiza-tional objectives, scope matched to team re-sources, roles and responsibilities, and key proj-ect milestones with delivery dates.

The importance of active senior management lead-ership cannot be overestimated at the project or orga-nizational level. In a division whose senior leaderactively drove DFSS from the start, project numbersgrew exponentially over the first three years of imple-mentation. In a similar sized division, in which DFSSdeployment leaders solicited projects from mid-levelmanagement, project numbers remained in the lowteens for the same time period.

Process Flexibility

The variety of scope addressed by DFSS projectsmeans DFSS needs to be more comprehensive andflexible than DMAIC. In DMAIC, BBs identify andreduce the frequency of defects generated by existingprocesses. The majority of work is analysis. In DFSS,teams must not only anticipate and prevent defects,they must anticipate and meet explicit and latent cus-tomer needs.

DFSS must address all three types of quality repre-sented in Noriaki Kano’s model: basic, performanceand excitement.6 DFSS requires a balance of analysisand synthesis—the synthesis of new designs and designstandards from a combination of deductive and induc-tive reasoning based on experience, observation andtheory. While analysis often proceeds linearly or is atleast guided by procedure, synthesis proceeds itera-tively and sometimes unpredictably as connections aremade between disparate ideas. Such iteration is aninherent part of the PD progression from the concep-tual to the concrete. Elements of each DCOV phaseare repeated as designs proceed from concept to com-puter model to prototype to production.

Compounding the challenge of DFSS is the addeduncertainty about future customer desires and futuresources and levels of production and usage variabilitythat will affect product performance. Assessing thesetypes of uncertainty is often not possible by collectingfrequency data from existing processes and makingstatistical inferences. Such phenomena are either notrandom or their random nature is not discernible.

So in addition to being equipped with statisticaltools, DFSS teams must have knowledge about nonsta-tistical methods outside the typical BB curriculum,including methods for obtaining consumer insight,cascading customer desires down to component spec-ifications, designing for robustness, mistake proofingand verifying designs with small sample sizes for test-ing. This methodology expertise, typically provided by


DFSS project numbers grew rapidly during the first three years of implementation.


BBs and MBBs, must be combined with deep techni-cal expertise in the product functionality in question.An ideal situation occurs when a BB from a function-al area finishes certification and returns to the area tosupport related DFSS work.

The Ford team initially devoted time to devising asingle DFSS flowchart. Attempts to capture any sub-stantial detail were finally abandoned because theyended up being redundant to existing PD flowchartsand failed to adequately address different project cat-egories and represent iteration in the process. It wasalso difficult to clearly represent the proper relation-ship between tools and outcomes in a single chart.

The team, therefore, decided to produce two docu-ments:

1. A tool matrix to outline which tools might beapplied at each project phase (see Table 3).

2. A project checklist to delineate intended out-comes at each phase. It is used with the under-standing some items may not apply to a particularproject as determined by the MBB and executiveChampion.

Training

One of the key DFSS implementation challenges istraining. Ford first applied a typical BB training modelto entire teams. Team members attended two weeks ofclassroom training during the early months of a proj-ect. They studied a curriculum that included in-depthcoverage of process and tools.

Participants in these pilot teams had two major con-cerns:

1. Some general tool and process information was

redundant to FTEP.

2. Not all the specialized tools presented wouldapply to a particular project. For those thatwould—due to the extended length of DFSS proj-ects—the team members worried they would for-get much of the training by the time they neededto apply it. They concluded it was inefficient totrain all team members in all tools.

As a result of the feedback, DFSS training wasrestructured. To enhance continued adherence to theprocess, teams now meet periodically in workshopscorresponding to each project phase (D, C, O and V).Instead of teaching tools, the workshop teaches theprocess and describes the results expected from theteam. The bulk of the time is devoted to helping theteam make progress on its specific project by inter-preting data and formulating strategies for the techni-cal work ahead. By the end of each workshop, the teamhas a written work plan for the project’s next steps.

A team process leader, typically a trained BB, guidesthe application of Six Sigma tools. An MBB facilitatesthe workshops, collaborates with team leadersthroughout the project and provides additional train-ing in advanced DFSS tools as required.

To supplement knowledge of basic DFSS toolsalready provided through FTEP, Ford’s office of theTechnical Fellow in Quality Engineering sponsors reli-ability and transfer function seminars for the engi-neering community at large and targeted to DFSSteams. Baseline DMAIC skills are also reinforced asengineering organizations near completion of a cor-porate mandate requiring all engineers to achieveDMAIC Green Belt status. The current flexible train-ing model in Table 4 (p. 22) rationally leverages


Table 3. Ford DFSS Tools Matrix

D Consumer Market Brand Noriaki Quality Bench- Quality Quality Quality lossinsight research analysis Kano function marking history: history: function

model deployment surveys repairs

C Concept P-diagram System Axiomatic Dimensional Failure mode Measurementgeneration and design variation and effects system Systemand functional Reliability analysis analysis analysis engineeringselection Design of diagrams and robust DFSS

experiments engineering Target scorecard

O Numeric/ Parameter Tolerancedesign

Analytical Statistical Design forsetting and

heuristic design design reliability tolerancing manufacture/verification

optimization and assembly Process robustness capability

V Model validation and Design verification Robustness/reliability demonstration Control planassessment

uncertainty assessment plan and report


existing training programs while driving projects for-ward on a just-in-time basis.

Implementing DFSS can be challenging for anyorganization. Experience at Ford revealed the chal-lenges are more likely to be cultural and organization-al than technical. Through adjustments based on les-sons learned, DFSS at Ford has emerged into what it istoday: an enhancement to the current PD system thatleverages the company’s extensive Six Sigma skill baseand culture to fundamentally understand product per-formance and deliver customer satisfying, robustdesigns. It adapts to wide ranges of scope and allowsflexibility in the choice of tools to achieve the requiredresults. It augments existing training with just-in-timeworkshops to help teams efficiently deliver results.

ACKNOWLEDGMENTS

Many people have contributed to the success of DFSS at Ford. Ihave had the pleasure of collaborating with Dave Amos, Tim Davis,Agus Sudjianto, Tom McCarthy, Chris Gearhart, Bob Thomas, EllenBarnes, Garry Smith, Fred Johns, Sam Hamade, Mahesh Vora, JohnKing and Belinda Hodge.

REFERENCES AND NOTES

1. Kai Yang and Basem El-Haik, Design for Six Sigma: A Roadmap for ProductDevelopment, McGraw-Hill Professional, 2003.

2. IIDOV = invent, innovate, develop, optimize, verify.

CDOV = concept development, design development, optimization, verify certification.

IDOV = identify, design, optimize, validate.

DMADV = define, measure, analyze, design, verify.

DCOV = define, characterize, optimize, verify.

IDEAS = identify, design, evaluate, assure, scale-up.

3. C.M. Creveling, J.L. Slutsky and D. Antis Jr., Design for Six Sigma inTechnology and Product Development, Prentice Hall PTR, 2003, p. xvi.

4. Douglas P. Mader, “DFSS and Your Current Design Process,” QualityProgress, July 2003, pp. 88-89.

5. Creveling, Design for Six Sigma in Technology and Product Development, pp. 8and 51-52, see reference 3.

6. Charles Berger, et. al., “Kano’s Methods for Understanding Customer-Defined Quality,” Center for Quality of Management Journal, Fall 1993, p. 3.


WHAT DO YOU THINK OF THIS ARTICLE? Please share

your comments and thoughts with the editor by e-mailing

[email protected].

Table 4. Training To Support DFSS

Audience

All DFSS team DFSS team DFSS MasterKind of training engineers members Black Belts (BBs) Black Belts (MBBs) Delivery

Green Belt course (two to three days) Req Req N/A* N/A Classroom

DMAIC BB course (20 days) Req Req Classroomprerequisites

MBB course (10 days) Req Classroom

Ford technical education program: Req Req Req Req Online introduces basic DCOV tools with test

Seminars in reliability and transfer Rec Rec Rec Req Classroomfunctions (two days each)

DCOV Team workshops (one-half to one day Req Req Req Classroomat each phase)

Advanced classes** Opt Rec Req Classroom

Shadowing a working DFSS MBB Req On-the-job

Key

Req Required

Rec Recommended

Opt Optional

* Not applicable, BB training supersedes this requirement.** DFSS process (one day), analytical reliability and robustness

(two days), structured inventive thinking (three days)

ford dfss

Documents