Professional Documents
Culture Documents
Ford Team Project Builds Relationships
Ford Team Project Builds Relationships
September 2011
With the Ford Fiesta just months away from arriving at dealerships, pre-launch reviews pinpointed
a concern—the vehicle’s carpet contained visible brush marks. Simply put, the carpet’s appearance
would not be acceptable to the customer. Ford typically uses warranty and customer satisfaction data to
identify top priority projects. In this case, because the Fiesta was a new product, Ford relied on antici-
pated warranty and customer satisfaction impact based on historical benchmarks. Addressing the carpet
quality before manufacturing commenced would alleviate customer concerns and avoid warranty costs.
The card
opens fibers
and lays Flat needling of loose two-dimensionally
them into a oriented fibers. Pad is bonded by
loose two- adding vertical orientation to fibers.
dimensional
felt.
To finished goods,
coating, or surface finishing
Winding
Station
After discussing the current process with HP Pelzer’s carpet Analyzing for Root Cause
engineering experts, the team developed a fishbone diagram and
an f(x) cascade. The y = f(x) cascade is a physics and engineer- Despite careful planning, the team experienced some unexpected
ing-based flowchart that uses why and how questions to identify issues during the DOE. The first four runs took three hours to
potential root causes. The two key questions are asked until spe- finish. Sterbenz says the biggest obstacle was simply coordinat-
cific measurables are identified. If, through the analyze phase, ing the experiment because the carpet manufacturing process is
these measurables are determined to be the key process input complex and uses a continuous roll of material. It would have
been wasteful to conduct each DOE run combination on indi-
variables (KPIVs), then control of these KPIVs automatically
vidual rolls. The only alternative was making changes to the
leads to control of the key process output variables (KPOVs),
machine settings on the fly and marking the roll where those
which is the big Y in the y = f(x) equation. In this case, the
changes were made, which required great coordination and more
y = f(x) cascade brought the team to the needler, which is a
manpower than originally anticipated. “It took us a few hours
machine that has a bed of needles that penetrate the raw material
and a few botched runs to work out the logistics, but we got it.
to produce the pile of the carpet. The needler and the needler set-
This created some stressful moments where I thought the DOE
tings became the factors in the design of experiments (DOE). would be abandoned, but through teamwork and some quick
It is essentially through the DOE that certain needler settings problem solving, we succeeded,” explains Sterbenz.
were determined to be the KPIVs. At the end of the project,
Sterbenz says the team was able to control the KPIVs and there- The next step was analyzing the DOE using an analysis of vari-
fore control the big Y—the carpet quality parameters. ance table, Pareto chart, and plots of the interactions and main
effects. When analyzing the data from the brush marks, it became
He explains that it was clear from the start that the needler was clear why HP Pelzer struggled to achieve the specified quality by
critically important to the carpet’s overall quality. Fortunately, unsystematically adjusting the needler’s settings—there were two
the needler had a limited number of settings to manipulate, offer- significant three-way interactions and several significant two-way
ing an ideal situation for using DOE to optimize the process and interactions. The analysis for softness/plushness was less complex
conduct root cause analysis. Although HP Pelzer had changed with only two significant effects, and this simplified the process of
the needler settings in past studies, Ford’s Black Belts found that simultaneously optimizing both responses.
the trials had been done unscientifically by changing one factor
at a time. Sterbenz explains that the DOE not only provided a list of sig-
nificant variables and interactions, but also a transfer function
Once the DOE planning was complete, the Ford team mem- between the inputs and the response. The transfer function is the
bers traveled to HP Pelzer’s plant in Eudora, Kan., where they mathematical and/or physics relationship between the KPIV and
viewed carpet samples produced when the needler settings were the KPOV. In this case, the transfer function shows the math-
altered. They quickly noticed that while some carpets were quite ematical relationship between the settings in the needler (KPIV)
plush, others were the unappealing texture of felt. At this point, and the carpet quality (KPOV).
the team decided that brush marking and softness/plushness
Using the supplier’s knowledge of the carpet process and the
would both be treated as responses in the DOE.
analysis of the DOE, the team felt certain that the variables that
Next, the team assembled an evaluation jury comprised of Ford control brush marking and softness/plushness were identified. It
and HP Pelzer representatives to conduct a gauge repeatability was clear to all that previous needler setting adjustments were
not successful because of significant interactions among the
and reproducibility (R&R) study on the two responses. Brush
variables. Now armed with mathematical models for each of
markings and softness/plushness were each rated using an ordi-
the responses, the team set targets and, after validation, moved
nal ranking scale, with 1–10 used for brush markings and 1–5
ahead with using optimization solvers to uncover the best set-
used for softness/plushness. The higher the rating, the better the
tings for both brush marks and softness/plushness.
quality. Since both responses were attribute and ordinal in nature,
Kendall’s Co-efficient of Concordance was used as the indicator The data analysis was shared at a full team meeting when the
of success. This formula measures directional agreement with optimum needler settings were presented. Unfortunately, these
a value of greater than 0.7 for both responses, indicating that settings were not acceptable because they could damage the
the team could trust the jury to evaluate the carpets fairly and needler if used for extended periods and one of the factor set-
consistently. In addition, the team knew that shifts in the ratings tings could adversely affect the carpet’s durability. These two
were not because of variation in the jury’s opinion, but instead new constraints meant that additional analysis was needed.
No brush marking
Brush marking
Warranty avoidance Ford $630 per claim No warranty claims (mass production
started in May 2010).
Scrap avoidance HP Pelzer $1,200 per roll No scrapped carpet rolls (mass
production began in April 2010).
Customer satisfaction Ford Proprietary calculation No customer complaints about carpet quality
Loss avoidance _____________________________ _____________________________________ on either internal or external surveys.
External customer Carpets that look and feel as expected
This project also clearly achieved the goals of the One Ford Engineering competed in the 2011 ITEA and was named as one
strategy. Ford’s relationship with the supplier was strengthened of 29 finalists. Team members shared their story with a live audi-
through teamwork and the mutual sharing of expertise. ence at the 2011 World Conference on Quality and Improvement,
where they captured the best OEM/supplier relationship award and
Sterbenz says that team building and understanding the critical earned second place in the team display competition.
KPIVs were the keys to the success of this project. “It was sur-
prising to us that scientific experiments had not been conducted Sterbenz says this project was a good candidate for the ITEA
prior to this project to fully understand the KPIVs. Now that process because the team used quality tools to reduce the defect
the KPIVs are understood, we have much better control over rate to zero, and after one year, there have been no warranty
the process and understand how to control the sensitivity of claims on the carpet and no scrapped carpets. “That is total suc-
certain factors.” cess and the epitome of teambuilding,” he states.
Improvement projects typically encounter resistance during the For More Information
implementation phase. In this case, questions came earlier when
the DOE was conducted and the supplier was concerned about • To learn more about this project, contact Scott Sterbenz at
potential damage to the needler and possible adverse side effects ssterben@ford.com.
on carpet durability. These issues were addressed by providing • Complete details on ASQ’s ITEA process are available at
training and successful examples, along with leveraging the sta- wcqi.asq.org/team-competition.
tistical expertise of Ford’s Six Sigma Black Belts. • Read more case studies on successful quality improvement
projects in the ASQ Knowledge Center at asq.org/
Sharing the Success Story knowledge-center.
Ford, an enterprise member of ASQ, believes in the value of shar- About the Author
ing quality success stories because tools or methods used in one
project can often be useful in another. The results of this improve- Janet Jacobsen is a freelance writer specializing in quality and
ment project were shared with several other teams at Ford and HP compliance topics. A graduate of Drake University, she resides
Pelzer and with a wider audience through ASQ’s International in Cedar Rapids, Iowa.
Team Excellence Award (ITEA) process. Team Ford—Body