The consistent, reliable production of products that meet the customer requirements is important to all businesses. 

This goal is supported with a manufacturing process that functions in a consistent and predictable manner, a process that is not out of control or unpredictable. Understanding the variability in your process and being able to identify out of the ordinary events and what is causing them is the key to managing the process, reducing variability, improving your process, and, as a result, saving time and money.

Why are Control Limits Important in Statistical Process Control?

Using statistical process control methodologies, you can assess current process data to identify the range in which each measured parameter of the process normally functions. This analysis allows you to set boundaries or limits that the process should not cross. These are your Control Limits. If a process crosses these limits, it is no longer functioning the way you expect. It is out-of-control because it is not within the limits you have established. Known control limits allow you to identify out-of-control events and their causes so you can eliminate problems that cause waste and rework. Over time, charting process data against control limits on a control chart will give you the power to visualize your process. You can identify and solve problems to make continuous improvements.

Unlike customer specifications which help determine only if products will be acceptable, control limits give you an understanding about the stability of your process allowing you to manage and refine your process.

The Meaning of Control Limits

The text below is an excerpt from The Book of Statistical Process Control by Zontec. This second edition book, written for the non-statistician, serves as a guidebook for statistical process control. Request a copy of the free Zontec eBook to find out more about statistical process control techniques and tools.

Statistics show us just how good and how bad a process or output can be and still be “normal.” How we define normal, in this case, must be based on the actual history of the operation. If we ask someone to define how the operation should work, they will most likely describe a perfect operation. However, for an accurate picture of the actual situation we need to ask how they expect the process to behave.

It is easy to predict the operations level with historical data. If we measure “enough” parts, we can calculate how much they “typically” vary from the target. This typical value, the average, is our prediction for how close future parts will be to their target. Remember, we can’t predict the behavior of the individual parts, but we can predict the distribution for a group.
We develop estimates of how far the parts can be from the prediction and still have variance caused by sampling rather than changes in the process. Each sample will have a different average value, but it will be within these limits.

For example, let’s say I give you a deck of cards. I tell you that I may have substituted red cards for none, some, or all of the black cards in the deck. You have no idea how many red cards are in the deck.

If you draw one card and it is red, you have sampled with n = l. Your statistic, the count of red cards, implies a distribution of 100% red in the deck. If a fair deck is 50% red, how much would you be willing to bet that this is not a fair deck?

You draw another card. It, too, is red. Now how much would you want to bet the deck is fair? You have more information, but it still is not quite enough. You draw 20 cards and all of them are red. Do think the deck is half red? Probably not. The odds of drawing 20 out of 20 red cards from a fair deck are so small there is not much risk in saying the deck is not fair.
With 20 cards, you can see that the odds are more than enough to make a safe bet. You could, however, make a safe bet with fewer cards. If you want to be correct 996 times out of 1,000, how many red cards must you draw without any black cards appearing?

The answer is eight. If eight cards of one color appear, there are less than four chances in 1,000 that the deck is fair. We have just developed control limits for betting the deck has not changed with the risk of error being roughly 0.4%.

Now suppose I hand you a series of card decks. You draw 20 cards from each and plot the number of red cards. If there are less than eight red or more than zero red, you call the deck fair. If the count exceeds either limit, you’ll say the deck was changed.

Now let’s see how the plot is drawn. Figure 5.1 shows a portion of a control chart. The center line is our average value, or prediction for the most likely outcome. The upper control limit is the highest number of red cards we can count and say nothing has changed. The lower control limit is the lowest number of red cards we can count and still say the process hasn’t changed. We plot the actual counts on this chart.

To take this idea a step further, let’s call the red cards bad and the black ones good. If we sample a series of decks and see no sign that they are unfair, we won’t change the process. If we start to produce decks with more red cards, we should stop the process and find out why we are getting too many red cards. Once we know the cause of the unfair decks, we can change the system to prevent this cause from recurring.

Now suppose we find eight black cards in a sample from our process. We conclude our process has changed so that we now produce more black cards than we did when we set up the limits. In this case, we would again stop the process to find the cause of the increase in black cards. Since, for this example, it’s better to have more than 50% black cards, we would want to change the process so it consistently produces decks that are more than 50% black.

Once we change the system of producing card decks, our old rules for testing the decks are outdated. We must recalculate the new average number of red cards in the decks. Then we need to create new limits for how many red we will allow before we say the process has changed to a new level.

Sample averages or ranges reflect any change in the population and give us a statistical signal similar to getting eight cards of one color. This signal only tells us that something has changed, not why it changed. We have to use our engineering and operating knowledge to find the cause or causes and act on them.

If there aren’t any statistical signals from the process, we say the process is in a state of statistical control. This means there are no signs of change in the process. It doesn’t mean we are happy with it, just that it is stable and unchanging.

If a change is signaled, and we have evidence that the process has changed, we say the process is “out of control.” Out-of-control in the statistical sense can be a good thing. If the process is never out of statistical control, there will never be a reason to make changes in it, and your quality and productivity won’t change. If we remove the causes of undesirable changes and maintain the causes of good ones, our quality, productivity, and costs will improve.

SynergySPC Control Charts Allow You to Visualize Control Limits and Improve Your Process

Using control charts to compare quality data against established control limits provides a picture of your process that can be easily understood by everyone in your organization. Control charts are a way to map the manufacturing process and show when and where variability occurs. They allow you to see even subtle shifts in the process that might be missed if you just looked at raw data. You can evaluate and adjust the process and then monitor the results against the adjustments. This is how control charts help you achieve continuous improvement.

The Zontec commitment to your process improvement is the driving force behind our innovative SPC tools. Our control limit and control chart features provide unsurpassed insights and visibility at-a-glance.

SynergySPC Gives You Flexibility in Setting Control Limits

SynergySPC products are so uniquely designed, that you have to experience them to fully understand their power. This is especially true when it comes to control limits. Setting control limits can be done in a variety of ways with just a few clicks. These flexible features include:

  • Multiple methods to calculate control limits
  • Fixed control limits – that you choose
  • Historic control limits – preserving the data in its original form

The Unique Control Charts in SynergySPC Quickly Identify Opportunities for Process Improvement

The unique control charts in SynergySPC are information-rich and monitor data in real-time. Saving you time, our exclusive color and symbol indicators give you a quick understanding of your process. These indicators help you determine, at-a-glance, if further action is required. You can quickly access a variety of SPC tools to analyze data, at any level of detail, letting you identify problems and take corrective action.

SynergySPC goes beyond charting and automatically monitors quality data triggering real-time alerts when out-of-control data is entered. Operators are alerted and automatically prompted to document cause and corrective action so details are not lost. Additionally, emails and text messages are sent to managers, so they can take appropriate action. Supervisors are also alerted as out-of-control indicators on the manager dashboard change in real-time. These important alerts are critical communications for the entire organization.

We understand the pressure and time constraints under which you work.

Our manufacturing background gives us the ability to truly understand how important the right tools are to take your quality to the next level. That’s why we include control chart and control limit features that save you time while giving you critical and actionable information. This will ultimately help you drive continuous process improvement that will benefit your bottom line.

As you use control charts to manage the stability of your process, it is important to also focus on your capability. Zontec also has a unique way to visualize your process capability. See this and other powerful, easy-to-use SynergySPC features in a demo.

Three Simple Steps to get Started:

  1. Zontec will Assess Your Needs: Schedule a 15 to 30-minute phone call to determine which Zontec product best fits your needs.
  2. Attend a SynergySPC Webcast: We will demonstrate the features, benefits and ease-of-use built into SynergySPC software.  
  3. We Ship Your Software: Congratulations, you purchased SynergySPC software and it shipped to you. After a few minutes to install, you begin collecting and monitoring real-time data.

 


Remove Uncertainty from Your Process

Request a Demo to Experience the Power and Ease-of-Use of Synergy SPC Software

 


 

 

Zontec and SynergySPC are trademarks of Zontec, Inc.

© 2018 Zontec, Inc.