Adding building blocks toward an ideal system in any manufacturing or service arena is predicated on an appropriate and sound foundation. The quest for an improved competitive position for the longterm is greatly enhanced by focusing on the process rather than the product.
The use of statistical methods is not an immediate remedy for all chronic ills nor a solution for every corporate problem. For effective improvements in quality and productivity, however, statistical methods provide efficient and effective guidelines.
What is statistical process control (SPC) ? To appreciate the essence of SPC, one must think of employing the language of statistics, applied to a process for the purpose of control. Basically, by using SPC we can:
Reduce scrap
Improve overall quality
Improve productivity
Compete in today's world market
Perhaps the most important message of the SPC is that as quality continues to improve, the cost of producing the product/service will decrease. As a result, an organization will see results in:
Increased profits and profit sharing
Increased market share
Steady work, rather than fluctuating
Present and future job security
New jobs for the future
An overview of what each aspect of SPC contributes to the total method is appropriate.
Statistics has often been described in a broad sense as a universal language which is most useful in describing variability. Applied to a process, this universal language may be employed to describe and analyze the physical variability of a process. Another way to say this is to allow the statistics to explain the behavior of the process. In essence, then, SPC is a methodology (not a specific tool) of collecting data, arranging the data in a chart or graph form, and interpreting the data to reduce variations in the process. A typical chart is shown in Figure I.1.
Statistical process control is not a quality improvement fad, but a continuing, ongoing, vital part of the process. When SPC is used on the actual work environment as a tool (not as a weapon for retaliation and assigning blame to the worker), process improvements will occur. We must understand that SPC is not a means to check the worker, but a means of monitoring the behavior of the process. We allow the process to do the talking and we respond accordingly . That is why some people refer to SPC as the "voice of the process," especially when the data are transferred into a control chart like the one in Figure I.1.
The main purpose of SPC is to prevent defects by monitoring the process while parts or a service is being employed, instead of detecting defects after the process is finished.
A familiar analogy involves an engineering drawing as a universal language used in describing the physical shape of a product. Many readers, including manufacturing personnel, are more familiar and comfortable with an engineering drawing than statistical methods. To gain the maximum improvement potential, proficiency in the statistical language is encouraged.
A process may be defined as a combination of inputs including personnel, machinery, materials, methods, measurement, and environment to attain desired quality outputs. Using this definition, a process may be thought of in global terms as all the operations of a business collectively or in a narrow sense as a specific spindle from a specific machine. It is imperative for the reader to understand that every organization, without a single exception, has processes. Every individual in any organization works with a process ”no exception. It is of paramount importance not only to identify, but also to understand "that" process.
Control, the final concept of SPC, is frequently misconstrued as a misnomer. Does statistical process control anything? The classical control cycle involves at least four actions: observing or measuring, comparing, diagnosing, and manipulating. Process control orients these four actions as a feedback loop depicted in Figure I.2.
Product control, characterized by traditional quality control, orients the four key actions in a feed-forward loop. As a feed-forward control system, the result is filtered outputs as shown in Figure I.3.
Product control fosters detection approaches and primarily focuses on containment of problematic process outputs. Process control nurtures a prevention philosophy and primarily focuses on process improvement.
SPC, then, is an appropriate term for a collection of sound techniques, statistically based, to guide process improvements in an efficient and effective manner. With the process control loop in place, emphasis should be beneficially applied to closing the loop appropriately through employment of the various techniques available.
To have an outstanding SPC program we must recognize individual responsibilities, some of which are to:
Be honest and truthful of what you see.
Record accurate readings and plot points.
Follow procedures for gauging frequency.
Circle out-of-control signals without fear of any kind.
Record explanation for out-of-control signals.
Follow action plan for out-of-control signals.
Use correct plotting codes if necessary.
Explain any gaps in data truthfully.
Record significant process changes.
Ask for help when needed.
To appreciate SPC, we must recognize the importance of data characterization and their application. The four items of concern in SPC are:
Appropriate and applicable data. Data characterization will define the accuracy of the behavior by looking at:
Sampling
Frequency
Variation:
Central tendency
Dispersion
Shape of distribution
These elements of variation will present themselves with some concerns in dealing with populations as opposed to samples. A cursory review is shown in Table I.1.
Data Characterization | ||
---|---|---|
Central Tendencies | Population | Sample |
Mean | ¼ = & pound ; X/N | X -bar = X/N |
Median | Middle value | Middle value |
Mode | Most frequent value | Most frequent value |
Dispersion or Spread | ||
| ||
Shape of Frequency Distribution | ||
Normal, skewed, uniform, exponential, etc. |
Control charts :
Variable charts
Attribute charts
Special charts
The problem: To determine from the variability pattern in the data which deviations from the norm or target have likely been produced by special causes and which have been produced by common causes.
Why ? The responsibility for corrective action rests with different people.
Specials cause: | Local fault which frequently can be corrected at the process by the operator and/or the supervisor |
Common cause: | System fault that requires the attention of management |
How ? Statistical techniques have the ability to separate the presence of special and common causes. It is for this reason that we use statistical control charts!
Capability:
Stability
Consistency
Predictability
The fundamental question here is: Is the process capable of producing acceptable parts/service consistently? This is very important and it must not be confused with whether the process is in control or not. The two are not the same. Control indicates the "voice of the process," where capability addresses "customer's requirements." The first are calculated numbers , whereas the second are given ”you have no choice in the matter.
In control you compare averages and ranges that are statistical configurations of the actual data, whereas in capability you compare individual data with single specifications. This is why control limits and specifications are not ” ever ”compared. To further make this point we provide a cursory summary of specifications and control limits in Table I.2.
Oranges Control Limits | Apples Specifications | |
---|---|---|
Characteristic of Basis Depended upon Utility | Process Process variability Sample Monitor stability of process | Product Function Part measurement Establish conformance or nonconformance |
What does all this have to do with quality and improvement? Plenty! For starters, the implication is that we are all concerned about prevention and not correction. If that is the case, then we all should be committed to early planning and execution of products and services that satisfy the customer(s). ONLY and ONLY THEN can we talk about the problems that may be prevented; otherwise the leverage of prevention is reduced as correction of problems ”a more costly approach ”becomes the dominant mode. A key aspect of this concept is designing products and services that can be produced with high yield the first time through the process within the capability of the manufacturing or service process. Designs that are immune to manufacturing and operational use variability are said to be robust (more about this concept in Volume V).
A second issue is the notion of setting "true" customer requirements. There is no way one can satisfy the customers' requirements if the provider does not know (1) his own process and (2) what the customer is really after. The first issue is addressed with SPC. In SPC we aim to understand and explain variation. But variation, as we will see in Chapter 1, is waste. The intent of SPC, under the best of environments, is to minimize or eliminate variation (waste) and ultimately itself. The waste may be a product of material, people's time, lost sales, capital, machinery, and so on.
Experts have estimated that the cost of waste in many large organizations is significant. Whatever the exact numbers, the waste in whatever form presents itself is an opportunity for reducing costs through improvement in variation. Much of the high cost of poor quality comes from processes that are allowed to be wasteful . This waste is often chronic and is accepted as the normal cost of doing business. (A top 100 Fortune company has been accepting a warranty cost of over $5 billion on a per-year basis as normal. Yet the same company is pushing for programs that cut other costs including head count, travel expenses, training, and the like.) The conventional way is not to get rid of chronic waste but to prevent things from getting worse by putting out fires. How sad indeed. Chronic waste-of-time material and other resources can be driven out ”or at least reduced considerably ”by implementing continual process improvement. The beginning of such a journey is SPC.
The second point of what the customer is really after is examined by looking at what has been referred to in the literature as the "insight" of the customer. Terminology does not matter here. What is important is the fact that knowledge of the user 's needs and expectations (internal and external) is prerequisite to satisfying them. Of course, it is very critical that these requirements be understood and reflected accurately in the specifications for products, services, and processes. Tools used in this characterization are quality function deployment, the Kano model, marketing surveys, focus groups, and so on.