Process Capability - Attribute Data

Locked lesson.

Upgrade

About this lesson

A stable process is one in which only random variation exists. A Lean Six Sigma team must eliminate sources of instability before attempting to improve the normal process performance. To determine stability with attribute data, lookup tables and yield rates must be used.

Quick reference

Process Capability – Attribute Data

Process capability is a measure of the ability of the process to meet the expectations of customers without additional effort. Processes controlled with attribute data rely on process yield to determine process capability.

When to use

Process Capability is normally used during the Analyze phase to determine whether an existing process can meet the customer expectations even under the best of conditions.  It is also used during the Control phase to assist the team in determining whether the process operators and managers are maintaining adequate process control.

Instructions

When determining process capability for attribute data, the process yield is first determined, typically using DPU, DPMO, or PPM.  The yield value is used with the right tail Z lookup table to determine an appropriate Z level. This is sometimes referred to as the sigma level of the process. This sigma level is converted to a process capability level that would be comparable to the process capability values calculated for variable data by dividing the sigma level by 3. For example, a PPM of 1,350 leads to a Z score of 3.00, which would be considered the sigma level for the process. Dividing that by 3 results in a process capability for the process of 1.0. The conversion of attribute yield into process capability is gaining acceptance in the quality community but is not yet a universally accepted technique.

Attribute process capability is not expressed using the Cpk or Ppk terminology. Attribute data process capability has not determined a center for the data and therefore it is not an “apples to oranges” comparison. The term process capability is still used with attribute data but the indices are not.

Process performance is often described with a combination of stability, capacity, and capability metrics. Stability is associated with consistency in the process performance. A lack of stability is often associated with drift in performance. Capacity is associated with the quantity of output. Poor capacity is often due to constraints and bottlenecks. Capability is associated with the ability of the process to create outputs that meet customer expectations. Poor capability is typically associated with poor process control.

Hints & tips

  • The attribute data process yield does differentiate whether the defective items were above spec or below spec. However, since the yield is a positive number we use the right tail z score table.
  • Lean Six Sigma organizations often talk in terms of “process sigma.” They are referring to their process capability.
  • Stability, capacity, and capability are often related. Problems with one of these will usually become a cause for problems in the other two.
Login to download
  • 00:04 Hi, I'm Ray Sheen.
  • 00:06 So now let's look at how we determine process capability
  • 00:09 when managing with attribute data.
  • 00:12 Process capability can also be calculated with attribute data,
  • 00:16 but we need to use a different approach than we had with the variable data.
  • 00:20 When organizations talk about process capability,
  • 00:24 they often do it by referring to as sigma level for the process.
  • 00:28 The sigma level indicates the number of standard deviations between the mean
  • 00:33 of the distribution and the closest spec limit.
  • 00:36 So for the Cpk of 1, we say that it is a 3 sigma process, and
  • 00:41 if the Cpk were 2, we would say it is a 6 sigma process.
  • 00:46 Where variable data is concerned, that means that the mean value of
  • 00:49 the distribution is 6 standard deviations away from the closest spec limit.
  • 00:54 Now, remember that we can express Z scores in the units of standard deviations.
  • 00:59 So if the Cpk is 1, the closest spec limit is 3 standard
  • 01:03 deviations away from the mean, which is a Z score of 3.
  • 01:08 So that brings us to the concept of process capability with attribute data.
  • 01:12 The attribute data can give us a percentage for defective units.
  • 01:16 You may recall that we can translate percentages into Z scores using the Z
  • 01:20 transformation table.
  • 01:21 So that's the approach we will use.
  • 01:24 One thing I want to be clear about, most of the quality community does
  • 01:28 not recognize the concept of attribute process capability.
  • 01:32 However, it's starting to become more widely accepted, so
  • 01:35 we're going to include it here.
  • 01:37 Let's look at how we go about finding the attribute process capability.
  • 01:40 We start with the yield for the process, that is the unit yield.
  • 01:44 So we are looking at defectives, not defects.
  • 01:48 Also, we want the yield rate, not the defective rate.
  • 01:51 So if the defective rate is 3%, the yield rate is 97%.
  • 01:56 It's a straightforward process then.
  • 01:59 Take the percentage defective, which you can get from the parts per million for
  • 02:03 the process, and then convert that to a yield rate.
  • 02:05 Now, using that rate, go into the right tail Z transformation table and
  • 02:10 determine the Z score.
  • 02:11 Now you can stop at that point and treat the Z score as your process capability,
  • 02:16 or you can further translate it to a sCp style number by dividing that Z score by
  • 02:21 3, just like you do in the Cpk measure.
  • 02:23 But be careful, this is not a Cpk measure.
  • 02:27 You don't know if the process is centered.
  • 02:29 In fact, you don't even know if the data is normal.
  • 02:32 All you know about the real world is that a percentage of the units are defective.
  • 02:37 Let's wrap up this discussion by comparing capability, capacity,
  • 02:41 and stability for a process.
  • 02:43 Let's deal with stability first.
  • 02:45 This refers to the variability in the process output.
  • 02:49 We know that all processes have variation,
  • 02:51 but our question is the level of variation.
  • 02:54 We normally discuss this using the process descriptive statistics.
  • 02:58 As we take different samples from the process,
  • 03:01 we're likely to get slightly different descriptive statistics.
  • 03:05 This leads to the concept of a 1.5 Sigma shift that is often used to
  • 03:10 convert sample variation to the overall population variation.
  • 03:14 I'll take a lot more time to talk about this in the next lesson.
  • 03:19 In addition to the process stability,
  • 03:21 an important process metric is the process capacity.
  • 03:25 This is the maximum amount that the process can normally be expected to
  • 03:29 deliver over a period of time, like a week or a month.
  • 03:32 There's a loose connection between capacity and stability.
  • 03:35 When a process is very stable,
  • 03:37 the throughput can often be increased if there are adequate process resources.
  • 03:41 But when stability is a problem,
  • 03:43 the capacity is often curtailed because of the high scrap and rework rates.
  • 03:49 Our third measure is process capability.
  • 03:51 This refers to the ability of the process to consistently meet the specification or
  • 03:56 desired process performance parameters.
  • 03:58 A process can be rock solid stable but
  • 04:01 have poor capability because the normal variation is beyond the spec limits.
  • 04:06 However, if there are stability problems,
  • 04:09 these will manifest themselves in the process capability numbers.
  • 04:13 Recall that Cp and Pp are the best case, but if the process starts to drift, which
  • 04:18 would indicate a stability issue, then Cpk and Ppk will be less than those values.
  • 04:24 And it is the actual values of Cpk and Ppk, or process yield,
  • 04:29 that the customers feel.
  • 04:31 Whether the process output is being measured with variable data,
  • 04:37 or with go-no gauges, or other attribute measurements,
  • 04:42 we can still get a good sense of our process Sigma level.

Lesson notes are only available for subscribers.

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

Gift this course