Locked lesson.

Upgrade

About this lesson

This class reviews the appropriate and inappropriate response to be taken based upon the data found on a control chart. It reviews what steps to follow when there is special cause variation and what steps to follow with common cause variation.

Exercise files

Download this lesson’s related exercise files.

SPC Corrective Actions.docx
209.9 KB SPC Corrective Actions - Solution.docx
60.1 KB

Quick reference

SPC Corrective Action

There are appropriate and inappropriate responses that can be taken based upon the data found on a control chart. Actions for special cause variation is different than that for common cause variation. Some actions are preventive in nature and some are corrective.

When to use

The SPC control chart should be updated in real-time with the current process performance.  Whenever the chart indicates the need for action, that action should begin immediately.

Instructions

SPC Control Charts are to be used for process control.  That means the information needs to be available to those who are making decisions concerning process operations.  In most cases, this means process operators and process managers.  In addition, any good control system provides immediate feedback on performance.  Based upon these principles there are three attributes that should be associated with using control charts for monitoring and tracking.

  • The control charts should be updated in real-time, or as near real-time as is practical. This will allow the operators to take immediate action if a problem is identified.
  • The control charts should be maintained where the process operators and managers are located.  This normally means that they are located with the process.  (Not sequestered in a location that is removed from the process)
  • The process operators and process managers must be trained in how to update the control charts and how to read the charts to recognize when a special cause variation condition has occurred.

A common mistake made by those who have not been trained or do not understand SPC, is to treat the control limits as inspections criteria for conforming or non-conforming process results.  The control limits are based solely on the voice of the process.  The customer spec limits could be wider, narrower, higher or lower than the control limits.  The control limits should never be used in that fashion.  Rather they indicate process stability, not conformity.

When a process that had been stable and under control shows the indications of special cause variation, the process operator should immediately stop the process to investigate the special cause.  In some cases, the process operation may be continued, but only when the special cause condition leads to conforming results.  This should not be the normal practice, but only an exception. 

When the process is stopped for a special cause, the operator should investigate the process operation to isolate the special cause conditions.   Once the special cause is understood, the business can make a decision how best to control that condition and restart the process.

Even though there is no special cause variation present in a process, action may be needed.  However, the action will be based upon the conformance to customer spec limits.  If the process is stable but creating non-conforming results, the process should be managed so as to center the results in the middle of the customer spec limits.  This will take maximum advantage of the available spec limits.  It is important to note, that the control chart does not include those spec limits, so this must be done based upon inspection data.  Also, when the process is centered, it may violate the special cause rule for a shift of the mean, depending upon the magnitude of the shift.   This is an intentional special cause and is not a reason to stop the process and investigate.  If the common cause variation is wider than the spec limits, the business should consider changing to a different process with less variation.

There is an additional means for categorizing SPC process actions.

  • Type 1 actions result in a process change that prevents the particular problem from ever occurring again.  The SPC chart highlighted the issue and a permanent preventive action is taken.  This does not correct any existing defective items, but it ensures no new ones will occur.
  • Type 2 action is a modification to the process so that when a defective condition occurs or is about to occur, the process automatically notifies the operator with a warning signal.  The operator can then take the appropriate action based upon the nature of the defect.  This is often a combination corrective and preventive action.  The operator is able to correct the problem as it is occurring, and the warning prevents the problem from continually repeating until it would be noticed at a later step.
  • Type 3 actions consist of training the process operator in SPC.  The operator is then able to read the control chart and recognize when the process goes out of control and an action is required on their part.  In this case, the process is not changed, rather the operator is changed through training.  This can have both corrective and preventive qualities depending upon the action taken by the operator.
  • Type 4 actions are to setup a special temporary off-line process to rework the items that were produced while the process was out of control.  This is purely a corrective action since there is no change to the process or operator to prevent the situation from recurring.

Hints & tips

  • Some organization use their SPC charts as a mark of excellence in their management practices and use them for displays.  If they are going to be at all useful, they need to be at the workstations and maintained with real-time data.
  • Do not include the spec limits on your control charts as an additional threshold line.  This leads to confusion on the purpose and use of control charts.  Keep inspection separate from process management.
Login to download
  • 00:03 Hi, I'm Ray Sheen.
  • 00:05 Your control charts are great, but only if they lead to action.
  • 00:09 Let's take a few moments and talk about the types of corrective actions
  • 00:13 that are indicated by the performance that we see on the control chart.
  • 00:17 Control charts are used to monitor and track process performance.
  • 00:20 And when used effectively,
  • 00:22 they will lead to improved appropriate corrective action when required.
  • 00:26 Now, to do that, the control charts need to collect and
  • 00:29 record data in real-time, or nearly real-time.
  • 00:33 It doesn't do any good to collect the data back in an office somewhere and
  • 00:36 then issue a control chart once a month.
  • 00:39 Yet I have seen that exact process being used.
  • 00:42 The process operator or manager needs data in real-time so that they can understand
  • 00:47 what is happening on the process and take action when it is unstable.
  • 00:51 Add to that, recognizing a special cause is occurring will make it much easier to
  • 00:56 immediately gather information about that cause,
  • 00:59 rather than trying to recover it days or weeks later.
  • 01:02 Not only is real-time important, but control charts need to
  • 01:05 be where the process operators and managers are located.
  • 01:09 By that I mean, right at the workstation or equipment.
  • 01:13 Some companies centralize all their control charts on a measurement board
  • 01:17 in a prominent area of the workplace, like the cafeteria or the lobby.
  • 01:21 Well, that's a great practice if your only goal is to brag about all those SPC
  • 01:25 control charts that you generate.
  • 01:27 However, if the goal is to monitor process performance,
  • 01:30 they need to be with those who are operating the process.
  • 01:34 That way, they can immediately recognize a problem and stop a process that has gone
  • 01:38 unstable, and communicate with others in the process about what is happening.
  • 01:42 And of course, if you want process operators to take
  • 01:45 action based upon the control charts, they need to be trained on the use of them,
  • 01:50 both the data entry and the interpretation of the results.
  • 01:53 Otherwise, you may get data plotted in strange ways, and
  • 01:57 you will likely find that operators start process tampering.
  • 02:01 Either that, or the operators will do nothing at all with the control charts,
  • 02:05 because they don't understand them or their significance.
  • 02:09 Ignorance leads to inaction.
  • 02:11 A major problem that occurs when control charts are not understood is
  • 02:15 that the operators and
  • 02:16 managers incorrectly interpret the meaning of the control limits.
  • 02:21 Control limits tell us if the process is stable.
  • 02:24 They say nothing about whether the process result is acceptable to the customer.
  • 02:28 Therefore, control charts should not be used to inspect the product.
  • 02:32 Inspections determine conformance, and control charts cannot do that.
  • 02:37 I've seen out of control,
  • 02:38 unstable processes consistently deliver conforming results.
  • 02:43 In some cases, it was because of very loose customer specs.
  • 02:46 In another case, it was a special cause Hawthorne effect.
  • 02:50 The company put in the extra inspectors and auditors to find defects and
  • 02:53 screen parts.
  • 02:54 The process performance well when the oversight was there, but
  • 02:58 as soon as it left,
  • 02:59 the performance returned to the previous levels of normal common cause variation.
  • 03:04 I've also seen many instances where stable,
  • 03:06 well-controlled processes were not producing conforming results.
  • 03:10 A common problem is a process not centered in the spec limit band.
  • 03:14 So the process consistently yielded results that were too high or too low.
  • 03:19 But I've also seen the case where the tolerances or acceptable performance
  • 03:22 limits were ratcheted down so tightly that normal variation exceeded the spec limits.
  • 03:28 In some of those cases, we just accepted high scrap, and in others,
  • 03:32 we put permanent rework into process.
  • 03:34 Both approaches added significant cost, but
  • 03:37 that was still less expensive than switching to a totally different process
  • 03:41 approach that could hold those very tight tolerances.
  • 03:44 So let's look at some specific corrective actions to take
  • 03:47 based upon the process performance.
  • 03:50 First, let's discuss special cause.
  • 03:52 In this case, one of the special cause rules has been broken.
  • 03:55 There's an extreme data point, a process shift, or a trend has developed.
  • 03:59 The first thing is to stop the process if possible.
  • 04:03 In some cases, you may continue to run the process with tight monitoring.
  • 04:07 Or if, for instance, a shift occurs that puts the process performance well within
  • 04:11 spec limits, but that is unusual.
  • 04:14 The standard practice should be to stop the process and investigate the cause.
  • 04:18 By immediately stopping, it's easier to collect context data about the problem.
  • 04:23 Often the root cause is immediately evident, but if we delayed,
  • 04:26 we'll lose some of that information.
  • 04:28 Once a root cause has been isolated, appropriate action
  • 04:31 should be taken to prevent the cause or to put a control in place.
  • 04:35 So if the cause does reoccur, the process is immediately notified and
  • 04:38 the operator can take action to avoid the impact from that root cause.
  • 04:42 Let me give you an example.
  • 04:44 I was working with an injection molding company making plastic parts.
  • 04:47 They had a control chart on some of the material characteristics of the part,
  • 04:51 including voids in the part.
  • 04:53 On one occasion, a control chart indicated a problem, and
  • 04:56 the process immediately stopped.
  • 04:58 In the ensuing investigation,
  • 04:59 it was found that some material in the stockroom had been mislabeled.
  • 05:03 That problem was immediately corrected before additional bad parts were made.
  • 05:07 And the stockroom procedures were changed to eliminate the type of error that led to
  • 05:11 that problem.
  • 05:12 Now for common cause variation,
  • 05:14 I know you're probably saying it's only common cause variation present.
  • 05:17 The process is stable, what could be the problem?
  • 05:20 Well, it may still be making non-conforming results.
  • 05:23 It could be threshold performance.
  • 05:26 Of course, you won't know that by looking at the control chart,
  • 05:28 since the spec limits are not on the control chart.
  • 05:31 When there's non-conforming results, even though the process is stable,
  • 05:34 you don't need to stop the process and conduct an investigation.
  • 05:38 However, you should center the process to take maximum advantage of the allowable
  • 05:41 spec limits.
  • 05:43 In addition, you may want to consider changing to a totally different process
  • 05:47 with lower levels of common cause variation.
  • 05:50 This often will be a more expensive process, but
  • 05:53 the total cost of quality may show that the change is appropriate.
  • 05:57 Another way to think about actions based upon the SPC charts
  • 06:01 is to use these four categories.
  • 06:03 Type 1 actions are those that make sure the problem can never occur again.
  • 06:08 This usually requires a process change.
  • 06:10 And while it doesn't fix the bad part, it prevents bad parts from ever reoccurring.
  • 06:16 Type 2 actions are those that provide a warning to the operator
  • 06:20 whenever a similar problem is about to occur.
  • 06:23 Again, it doesn't fix the current problem, but the warning allows the operator to
  • 06:27 take action to prevent a significant occurrence.
  • 06:31 Type 3 actions are a training program, so that operators are able to
  • 06:35 recognize problems based upon reading the control chart.
  • 06:39 The process is not changed in any other manner, but
  • 06:42 the operator is now responsible for identifying a problem and
  • 06:45 taking appropriate action based upon how they read the control chart.
  • 06:48 And the Type 4 actions are purely corrective actions, not preventive.
  • 06:53 With Type 4, a short-term containment process is put in place to capture
  • 06:58 the defective items, and normally to rework them until they are conforming.
  • 07:03 However, the process is not changed, and
  • 07:05 nothing would prevent the problem from occurring again.
  • 07:09 Control charts are providing valuable information for process operators and
  • 07:15 process managers to maintain control of their processes and
  • 07:19 take appropriate corrective action.

Lesson notes are only available for subscribers.

PMI, PMP, CAPM and PMBOK are registered marks of the Project Management Institute, Inc.

Gift this course