Pain-Free SPC

Implementing SPC in your shop doesn't have to be 
expensive, complicated, or confusing.

Are your customers asking—or even demanding—that you implement statistical process control (SPC) at your shop? If you’re like many in the metalworking industry, you might be a little apprehensive about SPC. You may be discouraged at the thought of spending hundreds or thousands of dollars implementing it. Or you might not know the difference between a control chart and a Pareto chart. Well, you can relax, because SPC doesn’t have to be expensive—or complicated. In fact, implementing SPC at your shop can be relatively painless. It may even be one of the easiest quality-control (QC) methods that you’ve ever used. And it pays off in the long run, because it’s also the most accurate.

What Is SPC?

SPC is a QC method that was developed in the 1920s by W.A. Shewhart of Bell Laboratories. Instead of attempting to control the product, Bell monitored, analyzed, and controlled the process itself. SPC allowed operators to catch problems in a manufacturing process before the process produced bad parts.

Rather than culling parts that have been imperfectly molded from a batch of manufactured pieces, for instance, an operator might monitor mold temperature to avoid molding bad parts in the first place. This was seen as an advancement in QC because it eliminated wide variations in part quality.

During World War II, the government encouraged American manufacturers to use statistical QC methods to ensure quality. However, SPC—and quality itself—was not widely embraced by American businesses at the time.

Industry and government leaders in Japan, however, made an important decision to make quality the solution to its economic problems shortly after World War II. The country’s goods had a reputation for extremely poor quality and its economy was at an all-time low. The United States agreed to help by sending over many of its experts, including Dr. W. Deming and Dr. Joseph Juran, to teach Japan about quality, management, new manufacturing skills, and the use of statistical QC methods. The Japanese proved to be astute learners. By 1980, it was obvious that superior Japanese products were having a major effect on the U.S. economy.

Since then, the United States has been trying to catch up by increasing the use of quality systems such as total quality management and statistical QC methods such as SPC. As quality has become a more important issue to American businesses, even small and midsize shops in the metalworking industry have been faced with the task of implementing SPC into the workplace.

SPC and Metalworking
Manufacturers in the metalworking industry typically use SPC for high-volume production, particularly for the manufacture of tight-tolerance parts. According to Robin McDermott, director of training at Resource Engineering, a manufacturing consulting firm in Tolland, CT, the initial step is to determine what element in the manufacturing process is producing the biggest variation.

“Preferably with SPC, you look at the variables in the process rather than the characteristics of the part,” McDermott says. This means that an operator would be looking at the speed and feed rate rather than the part ID and OD. However, in metalworking, the speed and feed rate tend to remain constant.

One process characteristic that does change over time is tool wear, but tool wear cannot be monitored directly in real time. Instead, the operator examines characteristics of the cut that indicate how worn the tool is. Gaging the part, in this case, is the equivalent of monitoring the process. In fact, most machining-process variables are monitored by measuring features on the parts produced. “I find that we tend to work more with part characteristics in metalworking,” McDermott says.

After the key process variables are identified, the operator begins to take measurements at specific intervals, such as once every hour. Often, the operator is surprised at how few measurements are taken. For example, it’s possible for an operator who, using another method, took 48 measurements per hour to prevent one bad part from slipping through to take only one measurement in that same period of time using SPC. As long as enough measurements are taken to indicate variations in the monitored process, the operator does not have to gage every part.

Measurements that are taken may be plotted on several different charts, such as control charts, frequency histograms, and Pareto charts. These tools may seem confusing at first, but most users tend to pick them up quickly after learning basic statistical terms and theories. All of these can be produced by hand or computers and are used to organize and analyze data.

A control chart, for example, depicts measurements of part characteristics over time. This type of chart shows a midpoint that corresponds to the process average, an upper control limit that corresponds to three standard deviations above the average, and a lower control limit that corresponds to three standard deviations below the average. If all of the points on the chart fall within the upper and lower control limits, then the process is said to be in statistical control.

A frequency histogram, another valuable tool for SPC, is easy to generate by hand and provides powerful information about process data. This simple bar chart displays averages of small samples to show the variation in measurements.

Pareto charts are similar to histograms, except that these bar charts are always arranged from high to low. These charts illustrate the Pareto principle, which states that not all of the causes that are responsible for some phenomena occur with the same frequency or the same impact. Further, it says that only a few of the causes may be largely responsible for a phenomenon. Therefore, controlling those relatively few causes or characteristics can significantly influence a phenomenon.

By providing a window on the machining process, SPC allows operators to predict tool wear and change the tool before it becomes damaged or worn out. “I try to help them understand the value of predictability in the process,” McDermott says. “I can show that if they can predict what will happen with tools, they can optimize tool use.”

Often, McDermott finds that machinists are reluctant to make the switch to SPC. Some find it difficult to believe that taking so few measurements can provide accurate information. Her response is usually, “Do you ever get returns?” If so, she says, then the company’s 100% inspections aren’t 100% effective. And, she insists, SPC will be this effective, even though fewer measurements are taken less frequently.

Implemention Issues
There are several reasons that the metalworking industry has been slow to adopt SPC. For example, many shop owners and managers feel that they do not have the time, staff, or money to implement it. However, SPC doesn’t have to be time-consuming or expensive, according to McDermott.

“My preference is always to start out with hand charts,” she says. “If we can identify the key variables and only monitor those key variables, then hand charting isn’t that big of a deal.” McDermott does not feel that spending $500 to $20,000 on software packages or computer networks is necessary for small to midsize shops, since the basis of SPC is the measurements.

Some of her clients do use expensive software to produce more professional charts to give to customers. McDermott cautions that shops must take the time to analyze and learn from the charts instead of just producing them to satisfy customer requirements and then filing them in a drawer.

One problem that many shops face is whether to use SPC for shorter runs. Jeff Sutton, president of Maine Machine Products Co., a precision manufacturing company in South Paris, ME, uses other QC methods for short runs.

“For six pieces, we probably won’t have any SPC charts out there,” he says. “But we will have a quality plan. We’ll 100% inspect every dimension the first time, because the customer wants to see something.” Sutton says customers ask Maine Machine Products to supply SPC data 30% to 40% of the time. However, he finds that SPC is useful even if the customer does not request it. “SPC becomes automatic whether it’s mandated or not,” he says. “In some cases, we check things that the customer doesn’t care about. We might monitor metallurgical properties or check things that will help us control tooling dimensions.”

Getting Started
If you plan to utilize SPC in your shop, you may be wondering exactly how to go about it. For starters, designate at least one person to be responsible for SPC’s implementation. This person should become very familiar with SPC through books or seminars.

Many people worry that SPC is complicated and that the statistics involved are confusing. However, it’s not difficult to learn, according to McDermott. “You need to understand statistics, but the basics can be easily learned,” she says. “The objective with SPC is to get you to use it as a tool, and you have to understand the statistics underlying it to have confidence in the charts.”

During the implementation period, many companies offer their employees SPC training classes or seminars. Maine Machine Products, for example, offered a three-day SPC seminar to its employees when the company started using SPC 15 years ago. The company’s quality-assurance engineer runs classes every few years for new employees or anyone who needs additional training. The company also has offered additional seminars to inform employees about improvements to SPC.

The Bottom Line
SPC is a valuable tool for improving the quality of your products. But it must be used effectively.

“It’s frustrating when you see what companies can get out of variation reduction, but they’re doing it just to satisfy the customer,” says McDermott. “Many companies are taking more time but not getting any benefits out of it.”

When you decide to implement this statistical QC method, it’s important to execute it correctly. Try to remember that you will overcome the initial learning curve. And when you do, your company—and your customers—will bear witness to the true power of SPC.