Applying Statistical Process Control Techniques

Management’s goal is to reduce causes of process variation and to increase process capabilities to meet customer expectations.

Process Variation 

Variation is present in every process as a result of a combination of four variables:

(1) operator variation (due to physical and emotional conditions),
(2) equipment variation (due to wear and tear),
(3) materials variation (due to thickness, moisture content, and old and new materials), and
(4) environmental variation (due to changes in temperature,
light, and humidity).

Variation is either expected or unexpected, and it results from common causes, special causes, or structural causes.

Variation affects proper functioning of a process in that process output deviates from the established target.

Common causes of variation affect the standard deviation of a process and are caused by factors internal to a process. These causes, which are present in all processes, are also called chance (random) causes.

Chance causes are small in magnitude and are difficult to identify. Examples of common random causes include worker availability, number and complexity of orders, job schedules, equipment testing, work-center schedules, changes in raw materials, truck schedules, and worker performance.

Special causes affect the standard deviation of a process and are factors external to a process. Special causes, also known as assignable causes, are large in magnitude and are not so difficult to identify. They may or may not be present in a process.

Examples of special (assignable) causes include equipment break-downs, operator changes, new raw materials, new products, new competition, and new customers.

Structural causes affect the standard deviation of a process; they are factors both internal and external to a process. They may or may not be present in a process; they are a blend of common and special causes.

Examples of structural causes include sudden sales / production volume increase due to a new product or a new customer, seasonal sales, and sudden increase in profits.

A control chart is a statistical tool that distinguishes between natural (common) and unnatural (special) variations. The control chart method is used to measure variations in quality. The control chart is a picture of the process over time.

It shows whether a process is in a stable state and is used to improve the process quality.

Natural variation is the result of random causes. Management intervention is required to achieve quality improvement or quality system.

It has been stated that 80 to 85 percent of quality problems are due to management or the quality system and that 15 to 20 percent of problems are due to operators or workers. Supervisors, operators, and technicians can correct the unnatural variation. Control charts can be drawn for variables and attributes.

The control chart method for variables is a means of visualizing the variations that occur in the central tendency and dispersion of a set of observations.

It measures the quality of a particular characteristic, such as length, time, or temperature.

To improve the process continuously, variable control chartscan be used to overcome the limitations of the attribute control charts. Continuous process improvement is the highest level of quality consciousness. Control charts based on variable data reduce unit-to-unit variation, even within specification limits.

Variable data consist of measurements such as weight, length, width, height, time, and temperature. Variable data contain more information than attribute data. Variable control charts can decrease the difference between customer needs and process
performance.

The attribute chart refers to those quality characteristics that conform to specifications (specs) or do not conform to specifications.

It is used where measurements are not possible, such as for color, missing parts, scratches, or damage. Quality characteristics for a product can be translated into a go/no-go decision, as follows.

Go/No-Go Decision for Product Quality 
– Go conforms to specifications.
– No-go does not conform to specifications.

Two types of attribute control charts exist:
(1) the chart for nonconforming units
(2) the chart for nonconformities

A nonconforming unit is a product or service containing at least one nonconformity.

A nonconformity is a departure of quality characteristic from its intended level that is not meeting a specification requirement.

Two types of statistical errors in quality can occur, leading to incorrect decisions, as follows:

1) Type I error is called producer’s risk (alpha risk), which is the probability that a conforming (good quality) product will be rejected as poor quality product and not sold to customers.

This product meets the established acceptable quality level. This results in an incorrect decision to reject something when it is acceptable.

2) Type II error is called consumer’s risk (beta risk), which is the probability that a nonconforming (poor quality) product will be accepted as good quality product and sold to customers.

This results in an incorrect decision to accept something when it is unacceptable. Type II error occurs when statistical quality data fails to result in the scrapping or reworking of a defective product.

Stable and unstable processes are defined as follows. When only chance causes of variation are present in a process, the process is considered to be in a state of statistical control (i.e., the process is stable and predictable).

When a process is in control (stable), there is a natural pattern of variation, and only chance causes of variation are present. Small variations in operator performance, equipment performance, materials, and environmental characteristics are expected and are considered to be part of a stable process.

Further improvements in the process can be achieved only by changing the input factors, that is, operator, equipment, materials, and environment. These changes require action by management through quality improvement ideas.

When an assignable cause of variation is present in a process, it is considered to be out of statistical control (i.e., the process is unstable and unpredictable).

When an observed measurement falls outside its control limits, the process is said to be out of control (unstable). This means that an assignable cause of variation is present. The unnatural, unstable variation makes it impossible to predict future variation. The assignable causes must be found and corrected before a natural, stable process can continue.

Process Capability

Process capability is a production process’s ability to manufacture a product within the desired expectations of customers.

The process capability index (PCI), which indicates whether a process is capable of meeting customer expectations, must be equal to or greater than 1 to meet customer expectations.

A PCI of less than 1 means that the process does not meet customer expectations.

Process Capability Index (PCI)

The PCI is computed in several ways:
PCI = (UL – LL) / (6 x Standard Deviation of a Process) 

where UL is upper specification limit and LL is lower specification limit.

Upper specification limit means some data points will be above the central line in a control chart. Lower specification limit means some data points will be below the central line in a control chart.

Capability Ratio (Cp) 

Capability ratio (Cp) is specification tolerance width divided by the process capability.

Specification tolerance width refers to variability of a parameter permitted above or below a nominal value. Cp is a widely used PCI.