This is new to me and never thought about this. I just love your explanation. If you write any book I will be first person to buy and hopefully you have plans to write some books.
Great explanation. If my Ppk is above 1.67, can I use that as a justification for not doing an R&R since Ppk takes observed std deviation into consideration which includes GRR Std deviation as well
Great question - yes, pretty much. Use some common sense too: if the measurement method can't even detect within the spec range it clearly isn't suitable (for instance, scales that only weigh per 10 kgs are never suitable for a product with spec range 10 +/-1 kg even though they'll give you a great Ppk on paper - a measurement device should be at least 3, preferably 10, times more precise than the spec range). Gage R&R is especially useful in the 0,67 - 1,33 range of Cpk/Ppk, since you want to check if it's the measurement system that is generating much of your observed problems or really just the process itself.
1. Can you please show an example of both the ways how to get R&R from the cpk ?? 2. Also can you give example for the different cases for the both cases ?? ( like one is greater then other)
1. You don’t get R&R from Cpk or vice versa, you’ll need to calculate both indicators separately. The acceptable R&R value relies somewhat on the Cpk. 2a. Example good R&R vs observed variance, poor vs spec range: a fat-water separator should let off water from a fatty waste stream to have 32-35% fat in the resulting slurry. The changing nature of waste (different nature of the fats and varying presence of detergents) makes it difficult to analyse, the measurement variation is about 1% - so R&R vs specs is 33, which seems too high. It’s also a process with a lot of variation: you observe fat contents between 25 an 40% - so R&R vs observed variation is 7%, which is brilliant. You decide to focus on the process, not the measurement system, because for the current reality it’s good enough. (The Cpk of this process will be around 0,8) 2b. Good vs specs, poor vs observed: a filling line filling cartons of 1 kg, with a tolerance range of 950-1050 grams, has an observed spread between 995 and 1010. The checkweigher scales measuring each product in line show a variation of 3 grams: 3% R&R vs specs - brilliant, 20% vs observed variation - OK’ish, not super. You decide to accept the measurement system, as this process is so far from it’s specification limits (has a Cpk of 6,6) that it’s not that critical to have a world class R&R vs observed.
Hi Kevin, sorry for the late reaction. In my R&R file, you won’t find the sigma directly. Rather, you’ll see 99% R&R value, which is 5,15 times sigma. So if you divide that by 5,15 you’ll get the sigma value.
This is new to me and never thought about this. I just love your explanation. If you write any book I will be first person to buy and hopefully you have plans to write some books.
Great explanation. If my Ppk is above 1.67, can I use that as a justification for not doing an R&R since Ppk takes observed std deviation into consideration which includes GRR Std deviation as well
Great question - yes, pretty much.
Use some common sense too: if the measurement method can't even detect within the spec range it clearly isn't suitable (for instance, scales that only weigh per 10 kgs are never suitable for a product with spec range 10 +/-1 kg even though they'll give you a great Ppk on paper - a measurement device should be at least 3, preferably 10, times more precise than the spec range).
Gage R&R is especially useful in the 0,67 - 1,33 range of Cpk/Ppk, since you want to check if it's the measurement system that is generating much of your observed problems or really just the process itself.
1. Can you please show an example of both the ways how to get R&R from the cpk ??
2. Also can you give example for the different cases for the both cases ?? ( like one is greater then other)
1. You don’t get R&R from Cpk or vice versa, you’ll need to calculate both indicators separately. The acceptable R&R value relies somewhat on the Cpk.
2a. Example good R&R vs observed variance, poor vs spec range: a fat-water separator should let off water from a fatty waste stream to have 32-35% fat in the resulting slurry. The changing nature of waste (different nature of the fats and varying presence of detergents) makes it difficult to analyse, the measurement variation is about 1% - so R&R vs specs is 33, which seems too high. It’s also a process with a lot of variation: you observe fat contents between 25 an 40% - so R&R vs observed variation is 7%, which is brilliant. You decide to focus on the process, not the measurement system, because for the current reality it’s good enough. (The Cpk of this process will be around 0,8)
2b. Good vs specs, poor vs observed: a filling line filling cartons of 1 kg, with a tolerance range of 950-1050 grams, has an observed spread between 995 and 1010. The checkweigher scales measuring each product in line show a variation of 3 grams: 3% R&R vs specs - brilliant, 20% vs observed variation - OK’ish, not super. You decide to accept the measurement system, as this process is so far from it’s specification limits (has a Cpk of 6,6) that it’s not that critical to have a world class R&R vs observed.
@@TomMentink thank you very much for detail explanation.
Were do I find the R&R sigma in the calculations in a R&R study
Hi Kevin, sorry for the late reaction.
In my R&R file, you won’t find the sigma directly. Rather, you’ll see 99% R&R value, which is 5,15 times sigma. So if you divide that by 5,15 you’ll get the sigma value.