# Powder Weight Proportion Measurement Gage R&R

#### sinned

##### Registered Visitor
Help!
We have a measurement on powder particle size proportion using a set of sieves and an electronic scale. During measurement, the powder is sieve to obtain particles of certain sizes and use a scale to weight how much powder is left in the different sieves.

Question: to perform a GRR study of the measurement system, should i include the sieving into the system? Or just the electronic scale measuring the weight of powder ?
If it's the former case, I need to get the 10 samples through sieving and weighing. Should i?

#### Bev D

##### Heretical Statistician
Super Moderator
We have a similar process in my organization. The sieves are a critical element of measuring the particle size distribution so you must have replicate measures of the whole system not just the scale. This is a functional measurement system - so how you set up the test is critical. Technically it is a nested study as the powder cannot be measured twice (lots of powder in the sieves during the first measurement will seriously effect the study results. And different particle sizes will be effected differently due to their mass and the effect of static and vanderwall's forces...

The approach that I found to work best is to take a large sample of powder that will most likely be homogenously mixed and break it into enough sub samples for each replicate you will be testing (If doing the common 10 X 3 X3 study you will need 6 sub samples of 10 different lots).

Make sure you clean the sieves fully in between each rep...

#### sinned

##### Registered Visitor
If this is true, what about the measurement on ash content of powder ?
This involves measuring the weight of material before and after burning in oven with a scale. Do i also need to include the burning process into the system?

#### Bev D

##### Heretical Statistician
Super Moderator
yes - it is part of the measurement system.

#### sinned

##### Registered Visitor
I will try to arrange for that and see how well we can get about the GRR. I am not optimistic on the result as it will be difficult to get samples of full tol. range. To your experience, how well will the GRR be ?