Quality by Design

According to one expert, engineers save focus. A great deal of time and money when they on preventing fires instead of putting them out.

By Patricia E. Steding

In today’s industry, the focus tends to be on reducing development time and costs-a task made more difficult by increasing expectations for quality. Taguchi says that oftentimes the industry goes about this in the wrong order: by first reducing development time and cost, and then focusing on quality. Instead, he says, designing for robustness-or simply stated, focusing first on quality-will always preclude a true long-term reduction in development time and costs.

Sounds like a simple and straightforward approach? Or does it sound like it’s easier said than done? ACTIONLINE met with Taguchi to find out.

ACTIONLINE:
What do you think the future of engineering is?

Shin Taguchi:
Every company needs to develop a corporate memory. Some people call it an "enterprise design database."

These are used to reduce development time and cost, and to prevent fire fighting in downstream stages. The problem is that when a new boss starts, a new focus begins. Technology robustness is a key element in this technology bookshelf. You need Robust Engineering to accomplish this.

ACTIONLINE:
Please define technology robustness.

Shin Taguchi:
When a system is insensitive against noises, it is said to be "robust." Noise factors are variables we cannot control. For a product, noise factors would include clients’ usage conditions, environment and aging. Manufacturing variability is also noise. For example, you can’t ask your computer not to use a certain brake when temperature is less than 20 degrees Fahrenheit.

There are also many noises for manufacturing processes. For example, incoming material variability, lot to lot, within batch variability, operator, and shift to shift, to name a few.

Noise factors cause variability in the function of products and processes. Noise also causes problems and failures, and that’s what starts fires. And we spend a lot of time fighting fires. It is said that 75 percent of an engineer’s time at any given company is spent on fire fighting. We must reduce this number.

There are three types of countermeasure we can apply to noises. The first is to control or eliminate; the second is to compensate the effects; and the third is to optimize the design that the effect of noise is minimized. That is, desensitize the design against noises.

This minimization of noise effect is called robust design. Nothing is wrong with the first two. But it is always the most cost effective to work on Robust Design first, then on the first two. This is because to control, eliminate and compensate for noises will increase cost.

ACTIONLINE:
How will this approach, reduce cost? Could you give an example?

So Robust Design is fire prevention. Not fire fighting. You can use the same approach to fight fires; if they happen. If you use this approach to fight the fires, then you need only solve the problem once.

But if you want to be a world-class corporation, you need to prevent fires. The impact of reducing fires would be tremendous. The 75 percent of time a typical engineer uses to fight fires would allow for a tremendous cost savings and numerous opportunities for innovations.

ACTIONLINE:
I understand that your father, Genichi Taguchi, the world-renowned engineering expert sometimes referred to as one of the "father of quality," created some methods for preventing fires. How do his Robust Design strategies apply to fire prevention?

Shin Taguchi:
In order to prevent fires, we must optimize technology for robustness. This is much cost-and time-effective than to optimize each product for robustness and it’s probably 100 times more effective than to wait for fire fighting. By optimizing for robustness at the technology level, robust technology can be applied to family of product and future products.

There are several strategies that my father, Dr. Genichi Taguchi, recommends. He has been saying these things since the 1950s, when his methods became known as "Taguchi Methods," and it finally has grown into a system of Robust Engineering during the last 10 years.

In the 1980s, Taguchi Methods became very popular in manufacturing companies. I hear people say they know Taguchi Methods, they use Taguchi Methods, and they practice Robust Engineering. However, what I see is usually a simple statistical design of experiment for fighting fires to develop a model. Nothing is wrong with those activities. When there is a fire, you need to put it out. But Robust Design or Robust Engineering is different. And it is not difficult to understand those strategies.

Problems are symptoms of variability caused by noise factors.

It is much more effective to measure function and to search for a design in which variability of function is minimized.

Function is an energy transformation. (For software, it is transformation of information.) To achieve robustness, we measure the energy transformation or something that represents energy transformation.

Function is a relationship between input (signal) and output (response).

Function is not scalar. It has infinite dimensions. For example, the function of machining is to remove material to generate a shape. A shape has infinite dimensions. The imput to machining is electrical power and the output is the removal of material. Quality problems such as flatness, roundness (runout) or straightness are all symptoms.

ACTIONLINE:
What characteristics will engineers of the future need in order to be successful? Will robust design be an essential element of engineering in the next century?

Shin Taguchi:
Yes and no. A genius engineer does robust design. Unfortunately, not every engineer is a genius. I have met a lot of engineers, and I was impressed with many of them. All impressive engineers have common characteristics. These engineers:

	understand the voice of client.
	are innovative.
	understand that variability exists and know how to deal with it effectively.
	recognize a need for Robust Design.
	are team players.
	use facts and data effectively.

ACTIONLINE:
Doesn’t your father’s present research involve Robust Design? How is it going to impact the automotive engineering society?

Shin Taguchi:
He is working on something called "Mahalanobis-Taguchi System." We named it "Robust Decision Making" (RDM) so it does not sound like a medicine. The Japanese government is funding close to $1 million to develop case studies using RDM. The major study is called " Automotive Accident Avoidance System." The study, conducted with Nissan, discusses optimizing a sensor function. RDM optimizes decision-making based on a great deal of information. Optimizing a sensor function is a Robust Design issue. How to make a correct decision based on many chunks of information is an RDM issue.

ACTIONLINE:
How will engineers gain a competitive advantage in the 21st century?

Shin Taguchi:
I think it will still be achieving high-quality, low cost and reduced product development time. However, development time is becoming a more critical issue.

Another challenge will be providing new products and services that society really needs or wants. These products must also function as intended for their target-market clients’ lifestyles and conditions.

We must also provide products and services in which the value is maximized. Value is defined by value function/ cost. As would be expected, the cost must be minimized and the function must be robust.

ACTIONLINE:
How wold you rate the level of quality in the U.S. auto industry?

Shin Taguchi:
Quality in the U.S. industry has progressed a great deal. In the early 1980s, the quality movement started. Deming became a major figure in quality management. Companies started to implement statistical process control (SPC), Deming’s 14 points, design of experiment, Quality Function Deployment (QFD), Taguchi Methods, Design Failure Mode and Effects Analysis (DFMEA), concurrent engineering, Failure Modes and Effect Analysis (FMEA) and Fault Tree Analysis (FTA), to name a few. But most of these activities were geared toward solving their current quality problems and/or reducing defects and failures. As a result, the number of things that go wrong during the first three months of usage per 1000 vehicles (abbreviated "TGW") went down 10 times. It varies from 40 to 300, depending on the vehicle. The TGW of the Big Three automakers is as good as those of other automotive makers. However, reliability is still a big problem. It is said that warranty cost per vehicles almost $1,000, and the Big Three automakers produce 10 million cars and trucks per year.

For the road . . .

On the subject of future improvements, according to Taguchi, the issues are robustness over time and continuity. What it all boils down to is that the industry must change its culture to improve quality. New bosses enter the scene, and a new focus begins; people leave companies, and take with them what they’ve learned about solving company or product problems. Robust Design provides the bridge-an avenue for achieving today and tomorrow’s demands for quality, cost and delivery.

Nissan Case Study A case study from Nissan provides a good example of robust engineering. Nissan engineers reduced noise from the inter-cooler. They did not measure noise to reduce. They measured the function. The function of inter-cooler is to cool exhaust gases, also known as heat exchanging. It is done by letting hot exhaust gas go through many tubes with fins. Input is input gas flow and output is the gas-flow velocity in tube. For the study, engineers measured the relationship between input gas flow and gas-flow velocity. The ideal relationship between input and output is called ideal function and it is usually based on physics. They tried several design parameters such as intake angle, fin height, height-to-width aspect ratio and fin thickness. They found design parameters that can reduce the variability of the relationship between input and output based on the ideal function. As a result, they were able to achieve the following: Variability of flow velocity was reduced by 66 % The inter-cooler became much quieter. (6dB less in sound pressure, which is equivalent to reducing the energy to generate noise to 25 percent.) The heat exchanging efficiency was improved by 20 percent. (By reducing variability of energy transformation, the efficiency in energy transformation typically improves because it is smoother, meaning that more energy goes to perform the intended function.) One of the design parameters that contributed to the improvement was intake angle. There is no cost difference in changing these design parameter levels. As a result of a 20-percent improvement in efficiency, they were able to reduce the size of the inter-cooler by 20 percent. It is now cheaper, lighter, quieter and performs better. Nissan says the cost should be $4.00 less per unit for all inter-coolers in the future. They were able to kill many birds (requirements) with one stone. It is a study at the technology level in which the result is applicable to all inter-coolers today and in the future, which translates into significant reductions in product development time. They have have achieved better quality faster and at a lower cost.