Our product is tested with an instrument called an inkometer. The product is a resin solution used to manufacture printing ink. At the moment we test on our own machine, then send a sample to the independent lab. We are seeing multiple rejections of our product, requiring re-adjustment prior to sending the product to our customer. We have known about our problems for some time, finally I have time to try to address them. I've done a Gage R&R test last year and the total R&R was 23%. Not too bad for a process industry instrument. Nevertheless, we are still encountering rejections.
The specification at the independent lab is 23.5 to 25. Roughly 40% of the samples we send fall outside that range according the the independent lab. See the enclosed files for the graphical information.
At this stage I'm more interested in the instrument itself. It is used by placing 1.32 mls. of solution on the rollers. The machine runs at 800rpm for one minute (90oF) and then takes a reading of the force exerted on one of the rollers. I've been checking the instrument each day recently using the calibration weights and I'm seeing significant variations without even using a sample. Cp without a sample is 1.3.
I need to run an experiment to test for variations in result based on (1) temperature and (2) sample volume. The problem is that I can't easily control how much volume is placed on the instrument, so a standard full factorial DOE is not possible. What is the best way to approach this?

The specification at the independent lab is 23.5 to 25. Roughly 40% of the samples we send fall outside that range according the the independent lab. See the enclosed files for the graphical information.
At this stage I'm more interested in the instrument itself. It is used by placing 1.32 mls. of solution on the rollers. The machine runs at 800rpm for one minute (90oF) and then takes a reading of the force exerted on one of the rollers. I've been checking the instrument each day recently using the calibration weights and I'm seeing significant variations without even using a sample. Cp without a sample is 1.3.
I need to run an experiment to test for variations in result based on (1) temperature and (2) sample volume. The problem is that I can't easily control how much volume is placed on the instrument, so a standard full factorial DOE is not possible. What is the best way to approach this?

