I've just completed my first Gage R&R test. I found that if I isolate each operator, so that it's just a repeatability test, some of them are quite poor. The Gage is a rheometer, quite a sensitive machine.

Should I be rating the operator by using this test? or by some other method?

I've just completed my first Gage R&R test. I found that if I isolate each operator, so that it's just a repeatability test, some of them are quite poor. The Gage is a rheometer, quite a sensitive machine.

Should I be rating the operator by using this test? or by some other method?

Welcome to the cove !!

For more (and better) reaction here it would be a good thing if you attached the data of your study.
This way a lot of people here will be able to give you the right comment or answer(s).

Best regards,

Antoine

I've just completed my first Gage R&R test. I found that if I isolate each operator, so that it's just a repeatability test, some of them are quite poor. The Gage is a rheometer, quite a sensitive machine.

Should I be rating the operator by using this test? or by some other method?

What material are you measuring? Over what ranges?

The data would help a great deal.

I've just completed my first Gage R&R test. I found that if I isolate each operator, so that it's just a repeatability test, some of them are quite poor. The Gage is a rheometer, quite a sensitive machine.

Should I be rating the operator by using this test? or by some other method?

In a typical MSA situation, I would say that you could identify a problem in this way. That problem could be the need to adhere to the test instruction, the need for instructions in the first place, etc. I would NOT use it to rate an operator.

Having said that, a rheometer is typically a destructive test, so it is not your typical MSA situation. Performing an MSA on a destructive test is a more difficult situation than in a non-destructive measurement.

An MSA of a destructive test depends greatly on the collection of the test samples and the homogeneity of the product from which the samples are collected. Using an example of 10 parts x 3 operators x 3 repeats/operator, the 10 parts should be 10 separate batches of the material to be tested. The batches should not be consecutive batches, but should reflect the total process batch-to-batch variation. Nine samples should be taken from each batch. These samples should be taken from very close proximity to each other to minimize the effect of within-batch variation. These nine samples should be randomized before giving 3 to each operator. This will further reduce the impact of within-batch variation on an individual operator.

Make sure that the rheometer is at operating temperature and steady state by testing batches prior to the start of the MSA testing. Plotting the data points in the order of testing will help identify any potential warm-up issues.

Good luck! As stated earlier, posting your data will help us help you.

Thanks for the support. I have included the data from my test. I am operator 1, operator 2 is a QC technician, one that I don't have a high opinion of. The tests take some time, so I've done a short form test that I found in a book somewhere. 5 samples, 2 operators, twice each.

I hadn't though of the test as being destructive. The material being tested is a lithographic ink varnish. It is a pseudoplastic polymer solution in a mix of high boiling point mineral and vegetable oils. The polymer is highly soluble, meaning that there will be complete mixing in a batch environment. The material is stable in a metal can. Here we have always regarded the samples taken as being homogenous, that is, a sample taken from the tank represents the material inside the tank, as long as a few litres are tapped off first, to remove residual material in the valve.

The rheometer is used to test the viscosity at shear stress 0 - 2000 Pascal over 2 minutes. The machine takes 120 readings, which are then converted into a value for viscosity and rate index using a mathematical best fit calculation called the Power Law model.

Sorry, but obviously I was hoping to avoid the technicalities.

Thanks for the support. I have included the data from my test. I am operator 1, operator 2 is a QC technician, one that I don't have a high opinion of. The tests take some time, so I've done a short form test that I found in a book somewhere. 5 samples, 2 operators, twice each.

I hadn't though of the test as being destructive. The material being tested is a lithographic ink varnish. It is a pseudoplastic polymer solution in a mix of high boiling point mineral and vegetable oils. The polymer is highly soluble, meaning that there will be complete mixing in a batch environment. The material is stable in a metal can. Here we have always regarded the samples taken as being homogenous, that is, a sample taken from the tank represents the material inside the tank, as long as a few litres are tapped off first, to remove residual material in the valve.

The rheometer is used to test the viscosity at shear stress 0 - 2000 Pascal over 2 minutes. The machine takes 120 readings, which are then converted into a value for viscosity and rate index using a mathematical best fit calculation called the Power Law model.

Sorry, but obviously I was hoping to avoid the technicalities.

After looking at the data, I noticed two things:
Operator 2 averages more than 6% higher than operator 1 on 120Viscosity
. Secondly, the 120 rate index does not show significant differences between operator.

Not knowing this instrument, what could the sources of error be?
Amount of material tested?
Cross-contamination?
Temperature of material?
Data entry?
software operation?
Spindle size, cleanliness?

BTW, why the low opinion of operator 2?

Is there a standardization/calibration process for this instrument?

My experiences with rheometers always involved destructive testing such as curing of rubber. Your example sounds similar to what we called a viscometer. However, we never retested samples, but collected more from the same batch.

Are you retesting the exact same sample or another sample from the same can? If you are using another sample from the same can, you cannot use the standard MSA, which is a Crossed design. You must use a Nested design, which is typically used for a destructive test.

Operator 2's last measurement on Rate Index looks like a bad measurement and should be retested. Otherwise there is not a significant difference between the operators.

The average of Operator 2 is higher on Viscosity. The Range may be higher, but you would need to increase the number of samples to be certain. The difference in averages is probably due to a difference in methods. Do you have a standardized instruction?

We are taking samples from the same can, not using the same sample on the testing equipment. Ok, so I need to go with a Nested R&R rather than a Crossed R&R.

Is it ok to dump what seems like a bad measurement?

We have standardised instructions, but operator 2 is not highly regarded. Not highly motivated, slow and messy. Which is probably where variability in the measurements would come in. Using a rheometer requires that the sample is trapped between the cone and plate, not spread all over the place so that extra material is forming part of the result.

QC is a problem here. The wage rates are not high, and the work can be repetitive and dull for trained people. As a result, the turnover rate is high. QC staff tend to be graduates, they stay for a year or two, then leave for greener pastures.

By the way, where do I post a thread about rating operators for the various tests that they are required to do?

In this case, you should replace the measurement with another from the same batch if possible. Otherwise, your software will have to deal with a missing value. Plus, it further reduces an already small sample size.

There is a Training forum and a Human Factors and Ergonomics in Engineering forum. The Training forum is probably your best bet.