Designing for Six Sigma takes on greater significance when you’re involved in the design and manufacture of braking systems for passenger rail vehicles. As Bruce Thomas, manager of quality assurance, Wabtec Passenger Transit, Spartanburg, South Carolina, puts it, “When failures occur in the field and retrofits or modifications must be made, the effect on external failure costs is significant. The cost of commissioning a modification in the field or returning product for a retrofit is the driving contributor to warranty costs. It’s not like you can tell the customer to drive their rail vehicle to the local dealer for servicing.”

So in 2004 when Wabtec embarked on the design of an electronic relay valve (ERV) for use in the braking systems of New York City subways, the company developed its Design for Six Sigma process, which today remains an important part of its new product development efforts. The design team for the ERV, headed by John Crispin, program statistical engineer, performed Design for Six Sigma activities to address high-risk elements of design, environment, inherent failure modes, and issues regarding manufacturability.

“Everything we do is custom-engineered based upon a detailed technical specification,” says Thomas. “With the ERV, we set out to incorporate eight mechanical, pneumatic, electrical, and electronic devices as one, making it a high-risk LLRU (lowest-level replaceable unit) that had to perform in a harsh environment. It was extremely important that the product went through several cycles of improvement before introduction to reduce risk, improve performance, lessen time to an acceptable product launch, and ensure the expected quality and reliability.”

Such a project, says Thomas, required a systematic approach to eliminating design errors, improving performance characteristics, and ensuring capable processes before product launch. For Wabtec’s purposes it called for blending Design for Six Sigma—identify, define, measure, develop, analyze, improve, optimize, verify, and control—with the Shainin approach to eliminate design errors quickly. Using tools like “nominal-nominal” testing ensures that targets are set and proven capable and that all key characteristics are captured and controlled properly. Nominal inputs should produce nominal performance in the product. This approach blends Six Sigma, design, and process improvements to meet customer demand.

Today, the Wabtec ERV, introduced in 2004, has experienced just 11 field failures, for a failure rate of less than 1 percent, and there haven’t been any design modifications since its launch.

“This was a revolutionary, not evolutionary design, as we didn’t have any data points to pull in from previous experience and we required cutting-edge development and manufacturing processes,” Thomas says. In addition to the nominal-nominal test, Wabtec’s design process involves Weibull product life tests. Particular concern was given to the effects on the overall system of components with high-risk functions. Heat-generating, pulsing electric magnet valves positioned near the printed circuit board (PCB) as an element of the development phase of the design generated a focus for product improvement. Overstress probe testing helped “to induce as much heat as we would see in the field and to determine the reliability of isolation principles, eventually resulting in a single-cover, compartmentalized controller board whose position under the subway car isolates the board from the magnet valve while providing air flow to cool the unit and overcome risk of overheating.”

Says Thomas, “Our usage of tools such as nominal-nominal, Weibull analysis, and overstress probe tests gives us the concrete data we need to have the confidence before product launch.”

That’s not to say that tools such as FMEA (failure mode effects analysis) and poka-yoke (mistake proofing), don’t figure prominently into Wabtec’s DFFS template. Rather, as Thomas says, “We are going to continue to apply and refine our DFSS process.”

Next stop for the Wabtec Passenger Transit Division? A much smaller (40 percent) version of the ERV incorporating even more functions due for introduction in 2008. According to Thomas, “All of our high-profile, high-risk product development and manufacturability issues are being driven through the same process that was developed in 2004. And we’re headed toward developing the standard braking system that could be configured for any passenger rail application.”

This article originally appeared in the e-newsletter Six Sigma Quality in Manufacturing. Copyright © 2007 Society of Manufacturing Engineers. Used with permission.

About the author

Joe Jancsurak is editor of Six Sigma Quality in Manufacturing and Lean Directions e-newsletters and a Cleveland-based freelancer who writes on business and manufacturing topics.