So you thought you knew it all when it comes to quality? Just read Dr. Deming on statistical process control (SPC) and apply it to the line. Then relax and it’s Six Sigma, right? Well, what about the $10 billion that industry annually pays on warranties?
Warranty costs? The usual consumer reaction to that is: "Make it right and there won’t be any warranty costs!" True, said Dr. Deming, and true, say today’s experts. But how? What’s missing to the tune of $10 billion in manufacturing?
A new methodology called Lean Quality has captured the attention of the automotive/transportation industry and was the topic of a seminar April 21 in Livonia, Michigan. "Lean Quality: The Coming Revolution in Reducing Warranty Expense" was sponsored by Carnegie Mellon’s Graduate School of Industrial Administration and the American Society for Quality Automotive Division. The seminar brought experts on various aspects of manufacturing quality to the discussion, including the developer of the technology enabling this new methodology, Attensity Corporation of Mountain View, California.
The Lean Quality methodology was developed by the founders of Attensity who were from the University of Utah’s computer science and linguistics departments. The core idea of the methodology is to incorporate text-reading software into failure analysis. Here’s how it works, according to Attensity’s CEO Craig Norris and Bart O’Brien, vice president of business development.
"Instead of relying solely on the codes assigned to technicians to describe failures, Attensity’s software converts the technician’s text problem description into multiple causal failure attributes, which are then analyzed to reveal specific patterns of cause and effect. Leveraging the technician’s observations as to observed causality is the key to implementing Lean Quality," explains O’Brien.
Their software enables headquarters computers to analyze the texts sent in from the field and more quickly locate, for example, a problem back in manufacturing that is really causing the field failures.
"The clear understanding of problems promoted by incorporating technician comments allows manufacturers to describe problems more accurately and respond to problems more efficiently — this is the central concept of Lean Quality," adds O’Brien
"Some 70 percent of reported parts failure in automotive equipment are not matters of failed parts but rather," Norris says, "are connected to the Three L’s: Loose, Leaks and ’Lectrical. In other words, the problem is in the usage or the manufacturing process of the part. The part itself is fine.
"Furthermore, these Three L problems are not adequately described by part numbers — that is why they are missed by the current batch-oriented approach of using a part number Pareto chart to prioritize problems to be resolved. Our software has the ability to understand natural language at the level of diagramming sentences, and allows text to be used as part of the problem-analysis process," Norris continues.
"Our system," adds O’Brien, "solves the problem of ‘statistical dilution’ of failure reports. This statistical dilution is caused by multiple causal part numbers being used by different technicians to describe the same problem, which spreads the problem over so many part numbers that is does not appear statistically significant, but in reality it is."
For example, a report about a field complaint that includes text as well greatly speeds problem-solving, leading to the discovery of some mistakes or omissions in a material handling or production process.
"We transform a company from a code-centric operation to a problem-centric system. The system, known as QRD [Quality, Reliability and Durability] Center, is already being used by John Deere and Whirlpool, and is being studied by several automotive-related companies," adds Norris.
So how much are your warranty costs? Lean Quality can reduce warranty costs by "at least 20 percent," sometimes much more. One company’s time taken to solve failure problems was reduced by 50 percent and costs by 31 percent, Attensity claims.
Government security work? "We’re all over that field. The CIA’s investment arm, In-Q-Tel, is an investor. Our system is being used to find the bad guys all over the world," Norris adds. Text-analysis is crucial, after all, in the attempt to accurately decipher "chatter" among suspected terrorists and in other source material like bomb threats and audio and video messages.
When it comes to material handling and manufacturing, however, "Our system is not different from Dr. Deming in principle," says O’Brien. "But, in cars and trucks, the problems are complex and unpredictable. We’re adding a different kind of measurement — computerized textual analysis — to find solutions. We’re talking about feedback from the field to the line."
In effect, Lean Quality in manufacturing means finding the bad guys in the process rather than merely taking care of casualties. To learn more, go to www.attensity.com.
George Weimer, contributing editor, [email protected].