Statistical process control offers huge potential for bottom-line savings by reducing scrap, increasing yields and improving your company's ability to meet customer specifications. But few companies use SPC to its maximum advantage. Why not, with thousands, even millions of dollars in potential savings at stake? Are the statistics just too hard?

Not at all. The real difficulty in implementing SPC throughout a plant is in overcoming resistance from people at all levels of the facility. From operators to engineers to purchasing to finance to senior management, everyone has their own brand of resistance.

But it's not hopeless. Resistance can be overcome and the benefits of SPC gained to the bottom line. It just takes planning, commitment and easy-to-use software tools that automate and simplify analysis.

How SPC Works

Fundamentally, SPC enables you to reduce variation in a process whether it's an extruding or filling operation, heating or even calibrating gauges and scales.

All processes -- whether they are sampling techniques or product production -- have a certain amount of natural variation that is statistically predictable. The key to achieving the benefits mentioned above is to identify which variation is within that predictable range and which variation is unusual, or outside of it.

If your process is out of control (showing unusual variation), SPC allows you to investigate and work to eliminate the out-of-control conditions. Then, once your process is in control (statistically predictable), SPC allows you to perform a capability study to see whether the amount of natural variation in the process is acceptable. Further statistical analysis can help evaluate the ability of various process changes to reduce variability.

How SPC Works: A Case Study

The case of a commercial baking operation that needs to control product fill weights and portions illustrates how SPC works. To produce rolls, each of the bakery's three mixers prepares 1,200 pounds of dough, then dumps its batch into a pouring machine which then pours dough onto baking sheets. Dough for six rolls is poured onto a baking sheet lane with each sheet holding four lanes of rolls. The entire sheet is then baked, after which the rolls are cut and packaged in six-roll packs.

Each pouring machine dispenses dough onto baking sheets until it is empty. It then receives another 1,200 pounds of dough from the mixer and begins the pouring process again.

To meet FDA requirements, 98 percent of the six-roll packs have to weigh at least 380 grams. To maximize profits, the bakery needs to reduce fill weights as much as possible without violating its standard.

This is a perfect situation for SPC. The first step is to determine if the process is in control. Samples of baked sweet rolls from three lanes are taken every hour on both shifts and weighed. The data is then analyzed using an X-Bar/Range chart. As shown in Figure 1, the process is not in control (note all of the points above or below the control limits). These results indicate that external factors (i.e. factors which are not part of the pouring/filling process itself) are influencing the process.

Figure 1

Additional investigation reveals that there is little variation between individual rolls in a lane or between lanes on a baking sheet. The engineers also find little variation between shifts. To understand the problem more clearly, the engineers take samples every minute. This analysis reveals a pattern that the excessive variation occurs approximately every 15 minutes, the time between new dough batches.

Since each dough batch comes from a different mixing machine, the engineers analyze the dough coming from each mixing machine. One machine exhibits significantly higher variation in the texture of the dough it produces. Further examination reveals that the machine is not mixing flour and water according to the bakery's established formula. They adjust the machine and take more samples. As shown in Figure 2, the X-Bar/Range chart for the pouring/filling process reveals that, although the process is still not in perfect control, it exhibits much less variability.

Figure 2

At this point, the engineers examine the capability of the process -- i.e. they measure how predictably the process is producing within specifications. To do this they use Cpk, the ratio of the spread of the specifications and the spread of the process (the distance between the +/- 3 Standard Deviation lines). A Cpk of 1.3 is usually considered adequate. This process, as shown in Figure 3, shows a Cpk of 1.148, indicating that it will not be capable of consistently producing within these specifications. However, this capability chart also shows that the process is capable of producing rolls that meet the FDA weight requirement predictably as long as the overfill averages 10 grams (note virtually all of the distribution is above the specification).

Figure 3

Since the process is not in control, the bakery must first find other external factors that are influencing the pouring/filling process. Once the process is in control, they can continue making improvements to reduce the overfill.

This is a real-world example from a commercial bakery. Despite the fact that the process was not in control at this point, the process information they learned from their SPC study added up to nearly $200,000 per year in savings.

The People Part of the Equation

If SPC can uncover $200,000 worth of savings before the process is even in control, imagine it being applied throughout a manufacturing operation -- up and down manufacturing lines, across shifts, between plants, in the laboratory, among suppliers, even in the warehouse. With this kind of bottom-line impact, you'd expect to find SPC programs throughout every manufacturing enterprise.

Yet, most of the time, SPC is relegated to a quality department and/or the laboratory. Why? Because the people who can use it to the best advantage think it's too hard and/or too time-consuming. Plant-floor operators typically have little advanced math education and are phobic about statistics. Purchasing and Accounting don't see how SPC applies to them. And managers, while they may understand and even value SPC, don't have the time and attention to devote to analyzing data.

The solution to overcoming resistance in all of these cases is to make SPC easy.

Yes, statistical analysis can be easy if you combine the proper tools with encouragement and training. Companies that have successfully implemented SPC say the keys to success are:

• Easy-to-use software
• A team approach to SPC
• Plenty of training
• A cost-benefit mind-set
• Commitment and encouragement

Let's look at those factors one by one to see how you can successfully implement and use SPC throughout your organization:

Choose the Right Software

SPC software is the first tool to consider. Some organizations build a huge hurdle from the get-go by choosing statistical software that is too complicated for anyone but a statistician to use and requires significant IT resources to install. These packages often spend more time on the shelf than they do in use.

Other organizations don't invest in software at all, mistakenly thinking that a good spreadsheet program like Excel will work. The fact is, you can do many types of SPC analysis with Excel, but you will waste lots of time creating and setting up tables that you could be spending on data analysis. And if someone else wants to use your Excel tables, there's no guarantee they'll be able to follow your train of thought.

Invest instead in SPC software that is easy to use, yet technically stringent enough to do the types of SPC analysis you need to do. Be sure also, that it is easy to install and will be able to pull data from your corporate databases (such as MES or SCADA).

You need software that makes it easy to automate data analysis, allowing everyone from operators to engineers to management to spend more time reviewing the charts and less (or no) time learning to use the software. The software should provide clear, high quality graphics that allow a user to quickly and easily grasp the process situation without having to do any programming. It should allow SPC managers to easily automate routine tasks. It also should enable users to drill down into the data if desired but should also allow analysis without needing to deal with the data.

Consider also tools that can help automate data entry to minimize measuring and transcription errors. The ideal situation is to have the measurement device send the data directly to a database that can be accessed by the SPC software (or directly to the SPC software itself if you don't need the data for other purposes).

Finally, to get and keep everyone involved, put your quality data on a Web server and make the charts and reports available on your corporate intranet or extranet.

A Team Approach to SPC

We're not talking about the ubiquitous quality teams that companies used to create in every department. To make SPC work, you need a much more focused approach.

Start by identifying one or more areas in your operation where defect rates are high, yields are low, or where other conditions, such as customer specifications, will give SPC the most impact on your bottom line. Concentrate your resources at those points, create success stories, and only then consider expanding SPC to other areas of your operation.

Designate an SPC champion, someone with both a command of SPC techniques and enough authority to make the program work. Then, before you even think about SPC analysis, talk with the operators, engineers, maintenance supervisors and anyone else who affects the process. Ask them how they currently control the process. Collect data: are there differences by shift, time of day, season?

Then summarize the current performance of your process, converting the information into dollars (for example: the bakery's over-filling is costing at least $200,000 per year). Get someone from Accounting to help with the calculations. If possible, get your CFO to bless those numbers, taking into account operator overtime and machine costs as well.

Next, assemble your SPC team: operators, supervisors, engineers, maintenance technicians, lab workers, as well as people from accounting and purchasing, if possible. You want to involve whoever can have an impact on the process.

Show the team what improvements need to be made (for example, we need to reduce fill weights by x amount) and tell them that SPC is one of the tools you're going to use to make them. Show them what the process problem is costing and what the impact will be on the bottom line when improvements are made.

Only now are you ready to teach your team the theory, practice and implementation of SPC.

Training the Team

There are two common mistakes in SPC training: 1) Making it an esoteric classroom experience and 2) not training the whole team together. A lot of operators didn't go on to higher education because traditional classroom training methods didn't work for them. Making these people flash back to high school math classes will only increase their resistance to SPC. Instead, make the training part of a real-life, real-world experience with something they already know well: your process. Also, resist the temptation to train people by job category (for example, training only the operators as a group). To work effectively together, people need to see each other's points of view. Training them together, hearing the kinds of questions others ask, will greatly improve their ability to implement SPC methods together.

Start with action, not theory. Begin by teaching people how to use measurement tools and collect data correctly. Make sure everyone is measuring the process the same way. Have people collect data and put it into the SPC software.

When the team has properly collected data over a period of time, reconvene the group and teach them how to analyze it. Now is the time to talk about run charts, means, range, standard deviations, histograms, how to plot data. Draw on the entire experience of the group as you talk about interpreting data. And remember to keep the discussion grounded in what's familiar to them. Use the SPC software throughout your discussion so people become familiar with it naturally as they learn the concepts.

Allocate enough resources to give people as much training as they need to understand SPC. Be very careful -- especially if you're going to certify people -- not to make passing a test a source of anxiety for them. Let them take the test as many times as they need to, and give them plenty of help in between to understand the concepts.

Cultivate a Cost-Benefit Mind-set

Senior management's leadership -- through encouragement, support and resources -- is critical to successful SPC. There are two keys to getting it: 1) showing them how SPC impacts the bottom line and 2) keeping them involved and interested by showing them problems, progress and results.

The first key is partly why you should involve your accounting department (and/or your CFO, if possible) as part of your SPC team. This is not as hard as it seems. Your finance people know what scrap, rework and customer returns cost. They often feel frustrated because they don't know how to help operations improve those numbers. Most will jump at the chance to participate when they see how dramatically SPC affects the bottom line.

The second key is to make it very easy for senior management (as well as management at all levels) to follow the SPC team's progress. While they usually possess the math and computer skills, time-pressured executives rarely look at SPC charts unless they're extremely easy to access and provide quick information for decision making. The solution is to put your SPC charts on your corporate intranet where they can punch them up and see what's going on with the click of a browser. You can quickly and easily do this if you have web-based SPC software.

Commitment and Encouragement

A successful SPC program is well-focused, well-resourced and not distracted. It's patient. It has a clear vision of what the goals are and what the process will look like when the goals have been achieved.

When your SPC team achieves a goal, by all means, take time to celebrate. Throw a party. Make sure all involved know their efforts contributed to the success and are appreciated. Let the celebration spur all of them to renew their enthusiasm for the SPC program.

The ongoing challenge of implementing SPC is keeping your program going. There is a natural human tendency to celebrate early success and then abandon SPC as you move on to other newer projects. Remember this, and know that the time to redouble your efforts is when you have succeeded.

About the Author

Jeffery L. Cawley is vice-president of Northwest Analytical, Inc., maker of NWA Quality Analyst SPC software and NWA Quality Monitor plant-floor test station software. He may be contacted at jcawley@nwasoft.com or 503-224-7727.

back to top