# PPM Calculation & Reinspection

#### Eximius2286

##### Starting to get Involved
What is the standard practice for PPM measurements involving rejected parts that are reinspected?

I'm measuring parts rejected/parts produced x1,000,000 as to capture our internal PPM measurement.

For example, 10 pieces are inspected and 3 are rejected. The 3 pieces are then reworked and reinspected. Would the reworked assemblies be counted towards the total parts produced? or would these parts be excluded considering it wouldn't impact the assemblies entered into finished goods?

#### Bev D

##### Heretical Statistician
Leader
Super Moderator
Rejected parts are only included ONCE in their original state. UNLESS you are stating shipped ppm levels but since you stated “internal ppm” then you do not add the reworked parts twice to the denominator. However, what really concerns me is that you seem to be using the rejected parts from a small sample and dividing by the total number of parts produced. This falsely reduces the ppm level. The formula to use is the rejected parts divided by the number of parts inspected….

And before anyone gets all huffy about ppm - this formula is not a statistical prediction of ppm rates it is an actual calculation of the defect rate as expressed by the ppm instead of the ‘percentage’ modifier. But in the example presented the ppm would be quite large 3/10 = 30%

#### dubrizo

##### Involved In Discussions
As you're talking about assemblies produced I would ask why you're not simply using FPY/RTY vs. PPM

#### Bev D

##### Heretical Statistician
Leader
Super Moderator
Well ppm (or more accurately defect rate) is a simpler formula and easier for some managers - and engineers - to understand.

#### Eximius2286

##### Starting to get Involved
Rejected parts are only included ONCE in their original state. UNLESS you are stating shipped ppm levels but since you stated “internal ppm” then you do not add the reworked parts twice to the denominator. However, what really concerns me is that you seem to be using the rejected parts from a small sample and dividing by the total number of parts produced. This falsely reduces the ppm level. The formula to use is the rejected parts divided by the number of parts inspected….

And before anyone gets all huffy about ppm - this formula is not a statistical prediction of ppm rates it is an actual calculation of the defect rate as expressed by the ppm instead of the ‘percentage’ modifier. But in the example presented the ppm would be quite large 3/10 = 30%
Thanks Ben, this is the information I was looking for. That is how I was collecting it. We deployed a quality management software here years ago and it was never fully rolled out. It calculates PPM by including reinspected parts into the calculation which left me with a lot of questions.

#### Mike S.

##### Happy to be Alive
Trusted Information Resource
"This is why, when people start talking about parts per million nonconforming, you can be sure that reality has left the building." - Dr. Donald Wheeler

#### Bev D

##### Heretical Statistician
Leader
Super Moderator
Here is the article: “The Parts Per Million Problem

Of special interest is the part regarding predicting ppm from the extreme tails of the Normal distribution. I recommend that everyone read this…it draws a clear distinction between theoretical models and reality.

#### optomist1

##### A Sea of Statistics
Super Moderator
Having a touch of déjà vu here, in the automotive Arena this is typically called FTC, first time capability, the yield for a workstation or a series of workstations absent any rework. Hope this is some cilia too…
Cheers - optomist1

Replies
7
Views
595