Math In Context: Statistical Process Control

by Carol Sikes

Consumer Product Manufacturing and FPLA Regulations


All manufacturers whose products are sold in the United States are required by law to abide by FPLA (The Fair Packaging and Labeling Act) regulations. The fundamental requirements of FPLA are that any amount specified on the packaging of a consumer product must be statistically accurate. For example, if a package states that it contains 2.05 ounces of candy bar, then the manufacturer must be able to prove that this packaging claim is statistically valid.

The affects of not abiding by FPLA can be seen in lawsuits against companies that are found to distribute product that is mislabled. For example, in a lawsuit that was settled late in the year 2000, H.J. Heinz Co. was required to overfill its Ketchup containers in certain sizes by at least 1% for the next year throughout California as a result of an investigation that found repeated underfilled ketchup containers ranging around .5% to 2% less than the amount stated on the package.

Click here to see an article about the settlement of the Heniz lawsuit.



This essay will explore some of the aspects of statistical process and quality control in manufacturing environments that stem from the FPLA regulations as well as the need to meet customer needs for certain product attributes.

Statistical Process Control

While consumer manufacturers must be concerned with meeting FPLA requirements, manufacturers that produce raw materials for other manufacturers must be concerned with consistently meeting their customers' needs. As an intern throughout engineering school, I developed an understanding of statistical process control and quality control. I worked for a company called Photocircuits Corporation, which has a branch in Peachtree City, Georgia. Photocircuits is the largest printed circuit board manufacturer in the United States, and they produce boards for computer companies, automobile manufacturers, and other companies that manufacturer products requiring electrical circuitry.


In this manufacturing environment, statistical process control was used to make routine adjustments to process parameters in order to maintain appropriate finished product quality. Operators within the plant regularly measured critical variables and plotted values on control charts. The control chart limits were used to dictate when the operators should make changes to the process. For instance, the brush fingerprint of a cleaning brush, which removed surface defects from the copper covered surface of the circuit board laminate, was checked periodically throughout the shift. When the width of the fingerprint reached a level that was too low on the control chart, then the height of the brush was lowered or the brush was replaced. Similarly if the fingerprint was too wide, the height of the brush was raised. Throughout different areas of the plant key process variables were identified and measured to maintain good process control. Often the control strategy triggered by the control chart involved different degrees of reaction to an out-of-control condition.

The control limits for all of the process variables were checked on a monthly basis by the process engineers in the plant. The engineers would determine if the control limits needed to be changed, and if so they recalculated the limits and created new control charts. While the operators did not do calculations for the control charts, they did have to understand the correct graphing procedures and the logic involved in following the control strategy.

Statistics were also used in determining quality. One of my major projects as an engineering intern was to set up a new coating line and determine the quality capability of the line once it was operating stably. The method of determining the process capability required a great deal of quality testing and evaluating the quality data using statistics. The measure of process capability is related to the statistical confidence that the process can deliver a specific quality target.

Quality control was also important for established lines. Finished product quality was checked to ensure that each lot of circuit boards produced met the product quality criteria. If product was found that did not meet quality specifications, then the lot was placed on hold and then sampled more thoroughly to determine if it could be shipped or if it had to be reworked. Operators in the quality control department had to be aware of these limits and the proper responses.

Quality Control and FPLA

While in-process product manufacturers must be conscious of meeting quality requirements for their customers, consumer product manufacturers are required to meet certain quality targets by federal law. FPLA requirements do not apply to certain types of products, but the majority of consumer products that are found in retail stores and have labels specifying certain quantities for the product inside are covered by this law.

The National Institute of Sandards & Technology has a handbook that outlines the sampling procedures for determining adherance to FPLA. These procedures dictate how samples should be collected and how compliance for certain packaging aspects should be determined. Click here to link to the NIST handbook.

Following my graduation from engineering school, I started working for Procter & Gamble in Albany, GA as a process engineer. Unlike Photocircuits, P&G is a consumer products manufacturer, and the products produced at the Albany plant (Charmin and Bounty) are covered by FPLA. For all of my time at P&G I worked with Charmin toilet tissue.

Similar to the Photocircuits environment, certain aspects of the process at P&G are controlled using statistical control charts. While the engineers are involved in determining the limits for these charts, the technicians at P&G are heavily involved with these control charts. The control strategies for the charts involve making calculations to determine appropriate process moves. While they often don't think about the math while they're doing it, a lot of graphical interpretation and arithmetic is required by the technicians.

All of the engineers and many of the technicians at P&G are trained in basic statistics because it is such an important aspect of the quality control procedures. There are many aspects of product quality that are tested on an on-going basis, and two of these are governed by FPLA. Both Charmin and Bounty are required to deliver the sheet count and the sheet length listed on their packages. The managers and engineers involved in quality take the necessary steps to ensure that FPLA requirements are always met. The procedure for determining the in-house quality control limits is proprietary, but it involves a great deal of statistics.

More FPLA and the FDA

While I've had experience with statistical process and quality control as well as FPLA regulations, I was interested in the requirements for products that have much more information on the label. To investigate further I talked with a colleague from Georgia Tech, Mrs. Ellie Griffis, who is a process engineer for General Mills in Atlanta, GA. Ellie and Dave, the quality engineer for General Mills, confirmed that their products are definitely required to meet FPLA requirements. In fact, the Food and Drug Administration (FDA) has the authority to enforce these regulations for their products, the biggest of which is Honey Nut Cheerios.

It seems logical that the FDA would be very concerned with accurate labeling for any food or drug products. There is much more information on the packaging label for food and drug products than many other consumer products that typically must only meet weight or content specifications. I was curious about whether General Mills had to control each of the specifications listed on the nutrition label along with the weight requirements, and they assured me that was the case. Here are the specifications they have to meet to adhere to FPLA:

Nutrition Information 

 Calories 120

    Calories from Fat 15

 Total Fat 1.5g

    Saturated Fat 0g

 Cholesterol 0mg

 Sodium 270mg

 Potassium 90mg

 Total Carbohydrates 24g

    Dietary Fiber 2g

    Soluble Fiber 0.75g

 Protein 3g

 % Daily Value

 Calcium 10%

 Iron 25%

 Thiamin 25%

 Riboflavin 25%

 Niacin 25%

 Vitamin B6 25%

 Folic Acid 50%

 Vitamin B12 25%

 Zinc 25%

To see a picture of the side label of the box, click here.

The different nutrition variables are tested on different frequencies depending on their usual variability. While General Mills sampling frequencies and procedures for determining in-house process control limits are also proprietary, they do use statistical process control to ensure that their product meets FPLA requirements. The technicians in the Atlanta plant are aware of process control strategies and make process moves according to periodic quality testing.

Many thanks to Ellie and Dave for talking with me about their process!

An Interesting Classroom Activity

For anyone who has ever seen the movie Summer School, where the teacher motivates his students to learn to write by having them make complaints to various companies concerning problems with their products, it is clear that the possibility of benefiting from corporate wrong-doing is a potential motivator for students. The Heinz lawsuit provides an interesting example of a company that had to give away some of its product as a penalty for not meeting FPLA requirements.

Teachers might consider having their students pick a product or products to investigate in groups. The groups would experiment to determine if the product meets FPLA regulations. Although most manufacturers will take great precautions to ensure they are meeting these requirements, the prospect of uncovering something illegal is intriguing. The link to the NIST handbook for sampling can provide some guidelines for this type of experiment, or the teacher could set an arbitrary statistical limit for the investigation.

A sample unit for applied statistics based on problems involving statistical process control and quality control can be found by clicking here.

Happy Hunting!



Return to Carol's Home Page