Gage R&R - Destructive Test - Pharmaceutical industry - Tablet hardness


Dan Armstrong

Is it possible to do a gage R&R on an instrument that performs a destructive test? I work in the pharmaceutical industry and I'm looking to determine the reliability of an instrument that measures tablet hardness by cracking them.

KenK - 2009

In general you have two choices when dealing with a destructive test:

1 - find homogeneous parts (same batch or similar) and treat them as pseudo repeats. This can be pretty tough. You'll need to use a nested design instead of a crossed design since these are not true repeats - MINITAB provides a tool for this anlaysis.

2 - find some way to take repeated measurements without destoying something. For example, with a pull test hang weights from the device rather than actually pulling until a break occurs.

Ken K.


Evaluating Destructing Testing

Find Dr. Donald Wheeler's "Evaluating Measurement Systems". Published by the SPC Press. Excellent paperback covering the necessary means for evaluating destructive testing.


Re: Gage R&R

Of course which if it is possible to be made, single that you need to use the method R&R that analyzes the variation part to part.
Do you know it method ?

Any question, contac me via email.



I guess this a good topic to expose my lack of knowledge on the subject as any.

1- Is this machine controlled by the operator or is it automatic, where all the operator does is to place the pill in the fixture and push the button. If it is similar to a Rockwell hardness tester then the operator has no control other than viewing the results on a dial indicator or digital readout.
Doesn't seem like a gage R&R would accomplish anything for the gage relibility.

2- I would think relibility would be stated by the manufacturer, and verified periodically by calibration.

3- I like the idea stated earlier about using within part variation. Sounds like it might give a little more information on the rebility of the process.

Dan Armstrong

On the surface, it would appear that the operator just places a tablet into the machine and pushes a button. However, the test is very sensitive to the orientation of the tablet when it is placed into the tester, therefore an operator that is diligent in placing the tablet correctly may get different results than someone that is less diligent.


It appears then that the fixture is the initial "special cause" for variation and merits further investigation. Have you performed a gage R&R using three different operators? If so, what were the results? With this data you can determine the need for mistake proofing the fixture.

Dan Armstrong

That brings us back to my original post - how can I perform a gage R&R using three operators if the tester destroys the samples? I already know that there is variation from tablet to tablet, so I can't assume that a sample is uniform.
Restated, my problem is this - how can I determine the percentages of total variation that are attributable to gage and operator variation if I can't retest the same samples with different operators?
Top Bottom