Measurement Systems

Quick reference

• 00:05 Hi, I'm Ray Sheen, and it should be no surprise to you that during the measure
• 00:09 phase of a Lean Six Sigma project, the team will be measuring, and
• 00:13 how that implies that there is a measurement system.
• 00:16 I want to take a few minutes and talk about that.
• 00:19 A technique that you'll probably need to do in the measure phase is a measurement
• 00:23 systems analysis.
• 00:24 We have an entire short course on how to set up and conduct that analysis, but
• 00:28 at this time, I want to explain why it is needed.
• 00:31 I hope it's obvious to everyone by now that Lean Six Sigma relies on data, and
• 00:36 to do a full and adequate analysis, that data needs to be accurate.
• 00:40 Which begs the question of whether the team can collect accurate data.
• 00:44 A measurement system analysis answers that question by determining the amount of
• 00:49 variation in the measurement that is due to the measurement system,
• 00:52 rather than the variation in the item that is being measured.
• 00:55 If we fully understand the measurement system variation,
• 00:59 it will give us insight into the actual process or product variation.
• 01:03 Without a measurement system analysis, or an MSA,
• 01:06 you could reach a wrong conclusion about the process performance.
• 01:10 Typically, it would be to say that the performance was bad and
• 01:14 out of control when it really was not.
• 01:16 Then you start trying to chase down a problem in the process, and
• 01:19 the problem really with the measurement system being used to measure the process.
• 01:24 Because of measurement system error, you may think the customer is experiencing
• 01:28 a problem of the result being much too high when it is really too low.
• 01:32 We don't understand what the customer is truly experiencing.
• 01:36 So let's talk about measurement system error and its impact.
• 01:40 I'll start with the basics.
• 01:41 All processes have variations, and measuring is a process.
• 01:45 Therefore, there is some level of variation associated with the measurement
• 01:50 system.
• 01:50 The measurement variation will cause errors.
• 01:53 If the variation is large, the error is large,
• 01:56 and then the process data that you collect with that measurement system is invalid.
• 02:00 I was working with a company last year who was having a tremendously difficult time
• 02:05 with some optical alignment steps in their process.
• 02:08 I had them do a measurement systems analysis and
• 02:11 they found that the variation in the measurement system, the error in
• 02:15 the system, was more than twice what was allowed for the entire part variation.
• 02:19 The team's ability to ever get anything to work was just plain luck, that event and
• 02:24 the general experience with the product was the reason for
• 02:28 the Lean Six Sigma project.
• 02:29 To understand the measurement error, the team needs to understand the design of
• 02:33 the measurement system and know the status of the maintenance of the system.
• 02:37 And by maintenance, the primary item there is calibration.
• 02:41 Measurements and measurement error look like this.
• 02:44 We have a right triangle.
• 02:45 The variation in the part that is being measured, the true variation,
• 02:49 is represented by the horizontal side of the triangle, and the variation in
• 02:53 the measurement system is represented by the vertical side of the triangle.
• 02:57 The actual measured variation then is the third side of the triangle.
• 03:02 If measurement error is small, the vertical side of the triangle is short,
• 03:06 and the measured variation almost exactly matched the true part variation.
• 03:11 However, if the measurement error is large,
• 03:13 which means that the vertical side of the triangle is tall,
• 03:17 the measurement variation will be much greater than the true part variation.
• 03:21 Don't get too hung up on the math right now,
• 03:23 we will cover that in more detail in the Measurement Systems Analysis course.
• 03:28 So let's talk about the impact of measurement system error.
• 03:31 Let me use a few illustrations to explain what is happening and
• 03:34 then we can look at the implications.
• 03:37 In the first illustration, the measurement variation is in the middle of the actual
• 03:41 variation, and it's much less than the actual variation.
• 03:44 The measurement variation will have little impact on the accuracy of the data.
• 03:49 The second illustration has a large measurement error,
• 03:52 nearly equal to the part variation.
• 03:54 And in particular, it will produce a significant bias because it is not
• 03:59 centered in the same area as the part.
• 04:01 This third variation will also have a large impact.
• 04:04 While it is centered in the area of the part variation, the measurement that is
• 04:09 taken will likely have very little relation to the actual value of the part.
• 04:13 So what are the implications?
• 04:14 Well, one is that we may reject parts that are actually good but
• 04:18 whose measurement values are recorded as bad.
• 04:21 That can create a great deal of wasted effort in the process because you would be
• 04:26 trying to correct or rework a specific part,
• 04:29 and often tampering with the item until it is truly out of spec.
• 04:33 The third implication is that we can have a wrong understanding of process
• 04:36 capability.
• 04:37 This will lead us to spending a lot of time and effort looking for
• 04:41 root causes and process problems that aren't there.
• 04:44 So there can be many problems getting accurate data if you don't have a good
• 04:48 measurement system.
• 04:49 On many projects, I've had to clean up the measurement system before we
• 04:54 could even begin to understand the process.
• 04:56 Accurate data is foundational in any Lean Six Sigma project.
• 05:01 If you can't trust the data, you can't trust the analysis,
• 05:07 and you don't know what to change.

Lesson notes are only available for subscribers.