Locked lesson.
About this lesson
Process Capability consists of comparing the Voice of the Customer to the Voice of the Process. SPC Control Charts provide a quantitative measure of the Voice of the Process. This lesson explains the principle of process capability and the role of SPC in achieving and maintaining it.
Exercise files
Download this lesson’s related exercise files.
Process Capability.docx61 KB Process Capability - Solution.docx
61.3 KB
Quick reference
Process Capability
Process Capability consists of comparing the Voice of the Customer to the Voice of the Process. SPC Control Charts provide a quantitative measure of the Voice of the Process giving us insight into the process capability.
When to use
Process capability is used to describe whether a process is consistently able to meet the customer expectations for performance. It is often used in industry as a surrogate for process quality. The SPC control chart information is often used to visualize process capability.
Instructions
Process capability is the process metric that compares the voice of the customer to the voice of the process. The voice of the customer is expressed in terms of the process specification limits which are derived from the design of the process and the customer requirements. The voice of the process is derived from the process descriptive statistics. SPC control charts are linked to process capability. They are also derived from the process descriptive statistics and thy are used to maintain process control. Once a process capability is established, the SPC control charts become a primary management technique for maintaining that process capability.
A very important distinction to make is that the control limits on control charts are not the same as the spec limits for the process. They may occasionally be the same value, but they are still very different concepts and derived totally differently. The control limits are derived from the process and are based upon the process descriptive statistics; in particular, the mean and standard deviation. When the process performance changes, the control limits change. In contrast the spec limits are derived from the customer requirements and the design of the product or process to meet those requirements. It doesn’t matter what happens in the process, the spec limits are unchanged unless the customer requirements change. Ideally, the control limits are within the spec limits. In that case, as long as the process remains in control, the process should always deliver performance that meets the customer specification. However, a process can be in control but if the control limits are outside of the spec limits, the process will not be consistently meeting the customer requirements.
The method for calculating process capability varies depending upon whether the data is variable data or attribute data. Variable data is that which is measured on a scale. The value is always a number. With variable data, process capability is expressed using the indices Cp and Cpk for near term process performance and Pp and Ppk for long term process performance. The calculations for these will be discussed in other modules. The attribute data is category data such as pass/fail, yes/no, or a status indicator. This data is usually expressed in terms of a count or percentage of the data that is in the expected status and a count or percentage that is not. The process capability for attribute data will be derived from the yield values such as defects per unit, defects per opportunity, defects per million opportunity or parts per million. Again, these will be discussed in more detail in another module.
Process sigma is often used as a surrogate for process capability. Process sigma became popular with the widespread adoption of the six sigma quality improvement methodology. Process sigma also compares the customer requirements to the actual process performance. A six sigma process will have a process capability Cpk of 2, and a three sigma process will have a Cpk of 1. In many companies, the process capability or process sigma of a process is used as a measure of process quality.
Hints & tips
- Don’t fall into the trap of thinking that control limits are related to spec limits. The spec limits can be broader, narrower, or significantly displaced from the SPC control limits. That is why you need to look at process capability to determine if the process is really able to consistently meet the customer requirements. Process capability will combine the voice of the customer and the voice of the process.
- Attribute data process capability and variable data process capability are calculated differently. The literature loves to focus on the variable data capability indices – in part because these are much richer from a statistical perspective. However, in my experience a significant portion, possibly the majority, of data is attribute data. So you need to understand the process capability and control charts for that type of data also.
- 00:04 Hi, I'm Ray Sheen.
- 00:06 Having briefly discussed common cause and special cause variation, I wanna now go
- 00:11 deeper into the theory associated with SPC control, and best process capability.
- 00:17 As we start to discuss process capability,
- 00:20 let me first emphasize that this only applies to stable processes.
- 00:24 If the process has special cause variation,
- 00:27 it must first be removed before we can calculate process capability.
- 00:32 Process capability is a calculation, a metric.
- 00:35 It is a metric that compares the voice of the customer to the voice of the process.
- 00:40 Now, you may be thinking, what does that mean?
- 00:43 Well, first, let's explain what we mean by Voice of the Customer.
- 00:46 It is a specification for
- 00:47 process performance as derived from the customer's expectation.
- 00:52 Meaning, that this is what the customer wants for performance.
- 00:55 In particular, that specification will normally be a target value
- 00:59 with some tolerance limits around the target.
- 01:02 We call the upper tolerance limit the USL, or Upper Spec Limit, and
- 01:06 the lower tolerance limit, the LSL or Lower Spec Limit.
- 01:10 The voice of the process is based upon some of the statistical measures
- 01:13 that we have of the actual process performance.
- 01:16 In particular, we'll be using our old standbys of the mean and
- 01:19 the standard deviation.
- 01:20 These two values are calculated based upon the current process performance.
- 01:25 I'll explain how the ratios that are created from these
- 01:28 measures are calculated in other modules.
- 01:31 However, the important point for us right now is to realize that control charts
- 01:35 are designed to assist in visualizing the voice of the process.
- 01:39 With that visualization in mind, it's relatively
- 01:42 easy to understand whether the process is able to meet the costumer expectations.
- 01:47 Let me take a few minutes to clarify the difference between Control Limits and
- 01:51 Spec Limits.
- 01:52 This is often a point of confusion for those who are new to SPC.
- 01:56 Let me start by saying that they are not the same.
- 02:00 One comes form the customer and the other comes from the process.
- 02:03 They might coincidentally be the same but that is just a lucky coincidence.
- 02:08 The control limits come from the voice of the process,
- 02:11 they're the bounds of normal operation for the process.
- 02:14 The control limits show the range within which the process is stable and
- 02:19 predictable.
- 02:20 While not precisely, they are generally near the plus or minus three sigma points.
- 02:26 Specification limits are derived from the customer's definition of acceptable
- 02:30 performance.
- 02:31 The spec limits could be different for a different customer or
- 02:34 application even though the process and control limits are unchanged.
- 02:38 These spec limits are derived from the design of the customer application
- 02:42 via product or process.
- 02:44 And the only way to change these is to change the design or application.
- 02:48 Therefore, it is totally independent from what is actually happening in process
- 02:52 performance.
- 02:54 Now the ideal case, and this is the one that Shuhart was considering when he
- 02:58 designed the process capability ratios.
- 03:00 Is a condition in which the normal variation within the process,
- 03:04 which is shown by the control limits, is within the boundaries of the spec limits
- 03:08 that are derived from the customer application.
- 03:11 This would indicate that the process is predictable, and
- 03:14 that predictable zone is within the customer defined zone of acceptability.
- 03:18 In other words, the process is able to always deliver performance that
- 03:22 the customer will find acceptable.
- 03:24 So enough for the tease, let's talk about the process capability ratios.
- 03:29 The ratios are calculated differently depending upon whether
- 03:31 the data is variable data, that means that it can be measured on numerical scale.
- 03:36 Or attribute data, meaning that it is measured with categories like pass/fail or
- 03:41 on/off.
- 03:42 I'll start with variable data.
- 03:44 In this case, Shuhart created a ratio of customer requirements divided by process
- 03:49 performance.
- 03:50 There are actually four ratios.
- 03:51 These are the best case and
- 03:53 actual case ratios calculated with recent data known as Cp and Cpk.
- 03:58 Then there are the best case and actual case ratios calculated with data that has
- 04:02 been collected over a long time period,
- 04:05 historic data, which creates the ratios of Pp and Ppk.
- 04:08 The attribute data process capability uses ratios of successful instances of
- 04:13 the data in the correct categories.
- 04:16 The number of incorrect results are compared to the total number of
- 04:19 opportunities and ratios such as defects per unit, defects per opportunity,
- 04:24 defects per million opportunities and parts per million are calculated.
- 04:28 I have an entire module dedicated to both variable data and
- 04:32 attribute data process capability.
- 04:34 So we'll work through those at that time.
- 04:36 However, one other comment, these assume the data sets are normal data.
- 04:41 If it is not normal data, first try to normalize it
- 04:44 using the central limit theorem or a normalization transformation.
- 04:48 If that doesn't work, there are some special case approaches that can be used
- 04:51 with several types of non-normal data.
- 04:54 Let's wrap up this module with a discussion of Process Sigma.
- 04:59 Process Sigma is often used as a surrogate for process capability.
- 05:03 This came into vogue with the rise and use of the Six Sigma analysis and methodology.
- 05:09 Process sigma directly correlates with process capability.
- 05:13 In fact, the definition of a Six Sigma process is one
- 05:16 in which the process spread of + or- 6 standard deviations from the mean.
- 05:21 That is Six Sigma, falls within the customer's upper spec limit and
- 05:25 lower spec limit.
- 05:26 As you can see, that's the same ratio approach that we're using for
- 05:30 process capability.
- 05:31 So a Three Sigma process will have a process capability CPK of 1.
- 05:35 A Four Sigma process will have a process capability CPK of 1.33,
- 05:40 and a Six Sigma process will have a process capability CPK of 2.
- 05:45 Now it's no surprise the lean six sigma process often strives
- 05:49 to improve process capability.
- 05:51 They wanna achieve a process that has a Sgma greater than or equal to 6.
- 05:57 So that's our introduction to Process Capability.
- 06:00 Now, we're gonna have to go into some more detail to show how to calculate both
- 06:04 variable data process capability and attribute data process capability.
Lesson notes are only available for subscribers.
PMI, PMP, CAPM and PMBOK are registered marks of the Project Management Institute, Inc.