4squareviews

In the last post, the authors Mikel Harry, Ph.D., and Richard Schroeder of the book Six Sigma:  The Breakthrough Management Strategy Revolutionizing the World’s Top Corporations, discussed the fact that first-time yield is a crude measure of quality.

To recap their argument, first-time yield takes the number of units that go out of an inspection point defect-free, and divides it by the number of units that go into that inspection point.   So if 100 units go in, and 50 are defect-free, the first-time yield is 50%.   The problem about this metric is that, if you have two product assembly lines, and they both have a first-time yield of 50%, can you both say they have equal levels of quality?   Well, that would depend on the complexity of the parts involved.

If the units on assembly line A are very simple parts that only have 2 ways that each part could be defective, and the units on assembly line B are very complex parts that have 200 ways that each part could be defective, then a 50% first-time yield of product B is actually a lot more impressive than a 50% first-time yield of product A.    By concentrating on the parts that are defect free, it doesn’t tell you how many defects are in each unit that aren’t defect free:   are there 2 defects, 8 defects, or 28?    The first-time yield metric doesn’t give you this information.

The more accurate measure is throughput yield.  This is the number of defects per defect opportunity.    Let’s take our product A and product B from the previous example.    They both have first-time yield of 50%.   What are their respective throughput yields?  Let’s assume each unit has only defect found.    Product A has 2 ways that each part could be defective, so it has a throughput yield of 50%.    Product A has 200 ways that each part could be defective, so it has a throughput yield of 0.5%.   Product B is of a higher quality than product A, and this is readily borne out by the throughput yield metric.

However, as you probably can guess, there are very few manufacturing processes that have only one step.    Most manufacturing processes have several steps.   How do you calculate the yield for multiple-step manufacturing processes?   That is the subject of the next post.

Filed under: Uncategorized |