This is a tricky concept to explain, but when we calculator the operator range, we evaluated only 3 different values (Average of A, Average of B and Average of C). So in that calculation we don't consider the 10 samples that were measured. Which is odd I know, and that m-value reflects the number of operators.
Hey Frank, that sounds hypothetical, and I've never seen that before. Technically, if the EV is large the operator range will reflect/capture/include that variation (hence the subtraction), and operators tend to create variation in a measurement system, so I've never seen AV turn into a negative number. A negative number is impossible, you can't have negative variation.
Thanks for this. What do you do if you can’t collect a sample of parts that span your tolerance. ? Can you just isolate the inspection tool by grabbing the min and the max of the sample parts?
I still cannot understand the AV calculation; why d2=1.912? If the subgroup size=3, the number of combinations of parts and operators (g) = 10 parts and 3 Operators = 10*3=30. Based on the d2 table, we can know that the d2=1.693 instead of 1.192. Can you explain why d2=1.192 in detail in your case study?
Thank you for this excellent tutorial! I do have one question at 14min 10s where you calculate the reproducibility value. How did you come up values for n * r?
When looking up up the d2 factor for reproducibility you stay in the n=1 row, and then your m-value is equal to the number of appraisers. So in my case, I had 3 operators, so my m-value = 3, and my d2 factor = 1.912
@@greenbeltacademy Sorry, I should be little more elaborate on my question. For the Appraiser Variation (AV) equation, the 2nd half of the equation is (EV^2)/(n*r). How did you come up with the value for n and r? I saw you had 10 for n and 3 for r. Can you explain what do they stand for?
Hey Syed! You're using the right approach when looking at the data! That would be a good hypothesis for that situation, and you could prove that theory by observing the technique of that operator, relative to his/her peers, and evaluating their different levels of training/experience
Great question - and that's a tricky part of the whole process, so in that reproducibility calculation we use n = 10 because that sample size is used as an adjustment factor to essentially eliminate EV from the AV calculation. However when using the table to look up the d2 factor for AV you use a different perspective.
I have taken samples of 100x on part and another data set of 10x 10 parts. For calculating repeatability(R-double bar/d2) ,I'm using d2=3.088 on the 10x 10parts. what d2 should I use for the 100x on one part? Thanks
@@greenbeltacademy I want to ask for the minute 17:32, "Thus when the GRR includes 5 samples then d2 = 2.4812 and for 10 samples then d2 = 3.1791". Could you please give the link for the table for more samples other than 4 samples? Thank you...
Sir, firstly thank you for the videos very informative. Question when doing the AV @ 14:15, you said use n=1, but in the video it shows n=10...just checking and want to make sure if this is correct.
Great question! okay, so in that equation, when looking up the d2 factor, the "n" value that we use is always equal to 1, which is why in the equation I use 1.912 as the d2 factor. However, in that same equation (and I get how confusing this is), when calculating AV, that n value is still 10 because there are 10 samples in the gauge R&R
@@greenbeltacademy But *why* are we forcing "n" to equal 1 when selecting a d2 value? The table includes options for when multiple samples are used, which on first glance would indicate we use an "n" of 10. Why is it, then, that we only use the "n" value when calculating the AV and *not* for selecting the correct d2 value?
@@loganfarrow9811 Hey Logan, great question. To be honest, I've never seen an amazing explanation on this approach. I've always understood the sub-group sample size (m) correlating with the number of operators in the experiment because that's the number of measurements that you're including in the range calculation. However, I would assume that the sample size is based on the number of range values included in that calculation, which is simply just 1 since we only calculate the operator range one time.
Thank you for such a nice video! However, there is one question that lingers in my mind after watching this. So basically in Gauge R&R, you want to measure standard deviation (SD), which can be directly computed from your measured data. Thus my question then is, why bother calculating it "indirectly" from the values of range (R) and d2. Thanks!
Another question is the meaning of the R double bar. This is used to calculate the inherent variation of WITHIN the instrument. However, R double bar is calculated by taking the average of the range values from ALL OPERATOR and also from DIFFERENT ITEMS. Does this means that the R double bar also carries with it the variance of "operator to operator" and "item to item"? It kind of does not make sense here.
Hey There! Okay, great question about range versus standard deviation. The answer to this question (I believe) is somewhat historical. Years ago, when the GR&R method was first developed, calculations were done by hand, and range values are easier to calculate than standard deviation. Hence the method today. Now that we have minitab, ANOVA has become a much more popular method for evaluating GR&R, and the ANOVA method does calculate standard deviation directly instead of the range.
@@subasa83No, R-double bar doesn't not include the variation between operators because that variation blocks out any contribution from the operator or the parts. Remember, those range values that we use to calculate R-double bar, is ONLY the variation within a single operator, within a single sample. So any variation that is expressed into those range values comes from the measurement equipment itself.
@@subasa83 hmm that's weird, I thought I had already replied to that question. So honestly the answer has to do with history. Historically, Quality Engineerings made these calculations by hand, and the range value is so much easier to calculate by hand than the standard deviation. Now that we have software like minitab, we use ANOVA which calculates the SD instead of the range value. Hopefully that makes sense!
It was refreshing to watch your video. I wanted to brush up my fundamentals about the Gauge R&R and I watched multiple videos on TH-cam, but your video and explanation was detail, to the point and easy to understand. I am eagerly waiting for part 3 of the series. When are you planning to release? P.S. : - I used to work for Alcon in Houston Facility and had an opportunity to talk to you couple of times. It's good to see your here again online. I wish you the best. :D
Watching from South Africa ,Thank you Andy...I'm doing my Certified quality technician. CQT Training course...I find you videos helping me a lot.. The fact is u make clear example.... hopefully I will pass my CQT exams
Thanks @@egwinharianto2851 ! Okay, so that n is the number of samples used in the Gauge R&R (10 in this example), and r is the number of replicate measurements (3 in this example).
At 14:09 why you took value of N equals to 1?
because n represent sample size and in our case it is equals to 10.
This is a tricky concept to explain, but when we calculator the operator range, we evaluated only 3 different values (Average of A, Average of B and Average of C). So in that calculation we don't consider the 10 samples that were measured. Which is odd I know, and that m-value reflects the number of operators.
@@CQEAcademy so in all case we took the value of n equals to 1 while measuring reproducability?
@@pawandeepsingh8380 Yes!
What if R0 is super small where AV turns into a negative number before the sqrt?
Hey Frank, that sounds hypothetical, and I've never seen that before. Technically, if the EV is large the operator range will reflect/capture/include that variation (hence the subtraction), and operators tend to create variation in a measurement system, so I've never seen AV turn into a negative number.
A negative number is impossible, you can't have negative variation.
Thanks for this. What do you do if you can’t collect a sample of parts that span your tolerance. ? Can you just isolate the inspection tool by grabbing the min and the max of the sample parts?
Hey John, that'll get you fairly close because as long as you've got the Min/Max parts, then your part range will reflect that process variation.
I just started watching your videos which are very informative. Where can I find your Excel template(s) if they are still available? Thanks
You can get them here:
www.Greenbeltacademy.com/GaugeR&R
Thanks Andy, much appreciated. @@greenbeltacademy
I still cannot understand the AV calculation; why d2=1.912? If the subgroup size=3, the number of combinations of parts and operators (g) = 10 parts and 3 Operators = 10*3=30. Based on the d2 table, we can know that the d2=1.693 instead of 1.192. Can you explain why d2=1.192 in detail in your case study?
amazing job, you made the life of GR&R very easy, but as you promised i am still waiting for the 3rd video
Thank you for the feedback, that 3rd video should be coming out in a few weeks!
In Appraiser Variation formula, what is r ? you have calculated n*r as 10*3. Is r same as m ?
r is the number of replicate measurements, which in this example I took 3 replicate measurements of each sample.
@@greenbeltacademy thanks for the reply. Nice video again man. Its really helpful.
@@ahamedthanseeks1440 You're welcome!
If the measurement itself is a combination of operator decision making and the tool is the analysis still valid?
Thank you for this excellent tutorial!
I do have one question at 14min 10s where you calculate the reproducibility value. How did you come up values for n * r?
When looking up up the d2 factor for reproducibility you stay in the n=1 row, and then your m-value is equal to the number of appraisers.
So in my case, I had 3 operators, so my m-value = 3, and my d2 factor = 1.912
@@greenbeltacademy
Sorry, I should be little more elaborate on my question. For the Appraiser Variation (AV) equation, the 2nd half of the equation is (EV^2)/(n*r). How did you come up with the value for n and r? I saw you had 10 for n and 3 for r. Can you explain what do they stand for?
@@ahyungrocks5509 Oh gotcha, okay so in that equation n = the number of samples (10), and r is the number of replicate measurements (3).
@@greenbeltacademy thank you
Absolutely helpful and such great easy to understand language!
Thanks for the awesome feedback!!!
Glad you think so!
Brilliant video. Thanks so much
You're absolutely welcome!
Very informative video series. Thanks for putting in so much effort
You're welcome!!!
How come operator C almost always had a lower range value. Could it be that person has more experience/training with the gage.
Hey Syed! You're using the right approach when looking at the data! That would be a good hypothesis for that situation, and you could prove that theory by observing the technique of that operator, relative to his/her peers, and evaluating their different levels of training/experience
Awesome content sir, Any plan to continue uploading more? Cheers from Arizona
Yes, most definitely, and I apologize to everyone who's been waiting for the 3rd video! It's in process!
@@greenbeltacademy awesome!! I will keep my eyes on it
Look forward to 3rd part.
Why n=10 but d2=1.912? If n=10 in your calculation, does that mean d2=1.716?
Great question - and that's a tricky part of the whole process, so in that reproducibility calculation we use n = 10 because that sample size is used as an adjustment factor to essentially eliminate EV from the AV calculation.
However when using the table to look up the d2 factor for AV you use a different perspective.
I have been waiting for this video for weeks. Thank you.🌹
You're welcome!!!!!
i'm waiting for your 3rd GRR video, appreciate, amazing
You're welcome, that 3rd video should be out in the next month or so.
Thanks so much.
You're welcome!
You're welcome!
Watching from the Dominican Republic! The information has been beneficial for my CQE Preparation. I appreciate the time and effort!
You're absolutely welcome Carla!!!!
thank you so much , You are amazing, You are doing an amazing efforts
You're absolutely welcome and I"m glad to help!!
Andy I passed my Green Belt yesterday. Thanks again for the videos!
Oh wow that's awesome!!! Congratulations!!!
I have taken samples of 100x on part and another data set of 10x 10 parts.
For calculating repeatability(R-double bar/d2) ,I'm using d2=3.088 on the 10x 10parts.
what d2 should I use for the 100x on one part?
Thanks
Great video, this material very very help me to make the calculation for the GRR. Thanks..
You're absolutely welcome!
@@greenbeltacademy I want to ask for the minute 17:32, "Thus when the GRR includes 5 samples then d2 = 2.4812 and for 10 samples then d2 = 3.1791". Could you please give the link for the table for more samples other than 4 samples? Thank you...
@@CobaReviewI don't have a table to share, but if you google it you should be able to find something very quickly
Please can you tell me the link for the excell sheet and when will the third video be available ?
The third video will come out in the next 6-8 weeks!
Here's the link for the excel spreadsheet: www.Greenbeltacademy.com/GaugeR&R
helo any update on the third video@@greenbeltacademy
Hey @@achouchazeddine7336 !!! I'm working on a few other commitments right now, but should have it done in the next month or so.
Sir, firstly thank you for the videos very informative.
Question when doing the AV @ 14:15, you said use n=1, but in the video it shows n=10...just checking and want to make sure if this is correct.
Great question! okay, so in that equation, when looking up the d2 factor, the "n" value that we use is always equal to 1, which is why in the equation I use 1.912 as the d2 factor.
However, in that same equation (and I get how confusing this is), when calculating AV, that n value is still 10 because there are 10 samples in the gauge R&R
@@greenbeltacademy But *why* are we forcing "n" to equal 1 when selecting a d2 value? The table includes options for when multiple samples are used, which on first glance would indicate we use an "n" of 10. Why is it, then, that we only use the "n" value when calculating the AV and *not* for selecting the correct d2 value?
@@loganfarrow9811 Hey Logan, great question. To be honest, I've never seen an amazing explanation on this approach. I've always understood the sub-group sample size (m) correlating with the number of operators in the experiment because that's the number of measurements that you're including in the range calculation. However, I would assume that the sample size is based on the number of range values included in that calculation, which is simply just 1 since we only calculate the operator range one time.
Hello! Great videos. Is Part 3 coming soon? Thank you!
Great, Andy!
Thanks Victor!
Thank you for such a nice video! However, there is one question that lingers in my mind after watching this. So basically in Gauge R&R, you want to measure standard deviation (SD), which can be directly computed from your measured data. Thus my question then is, why bother calculating it "indirectly" from the values of range (R) and d2. Thanks!
Another question is the meaning of the R double bar. This is used to calculate the inherent variation of WITHIN the instrument. However, R double bar is calculated by taking the average of the range values from ALL OPERATOR and also from DIFFERENT ITEMS. Does this means that the R double bar also carries with it the variance of "operator to operator" and "item to item"? It kind of does not make sense here.
Hey There! Okay, great question about range versus standard deviation. The answer to this question (I believe) is somewhat historical. Years ago, when the GR&R method was first developed, calculations were done by hand, and range values are easier to calculate than standard deviation. Hence the method today.
Now that we have minitab, ANOVA has become a much more popular method for evaluating GR&R, and the ANOVA method does calculate standard deviation directly instead of the range.
@@subasa83No, R-double bar doesn't not include the variation between operators because that variation blocks out any contribution from the operator or the parts.
Remember, those range values that we use to calculate R-double bar, is ONLY the variation within a single operator, within a single sample. So any variation that is expressed into those range values comes from the measurement equipment itself.
@@CQEAcademy Tks a lot for your spot-on answer. How about my first question? Why don't we calculate SD directly from the data?
@@subasa83 hmm that's weird, I thought I had already replied to that question.
So honestly the answer has to do with history.
Historically, Quality Engineerings made these calculations by hand, and the range value is so much easier to calculate by hand than the standard deviation.
Now that we have software like minitab, we use ANOVA which calculates the SD instead of the range value.
Hopefully that makes sense!
Great video
Thanks!!!!
It was refreshing to watch your video. I wanted to brush up my fundamentals about the Gauge R&R and I watched multiple videos on TH-cam, but your video and explanation was detail, to the point and easy to understand. I am eagerly waiting for part 3 of the series. When are you planning to release?
P.S. : - I used to work for Alcon in Houston Facility and had an opportunity to talk to you couple of times. It's good to see your here again online. I wish you the best. :D
Also, I'm subscriber now.
What about video 3 guys 😢?
I'm working on it now, should be done in the next 4 weeks
Watching from South Africa ,Thank you Andy...I'm doing my Certified quality technician. CQT Training course...I find you videos helping me a lot..
The fact is u make clear example.... hopefully I will pass my CQT exams
You're absolutely welcome and I really appreciate the positive comment!!!!!!!!
you miss explaining n and r
Hey There! Which N and R value are you referring too?
@@greenbeltacademy thank you for the reply , at 13:54
Thanks @@egwinharianto2851 !
Okay, so that n is the number of samples used in the Gauge R&R (10 in this example), and r is the number of replicate measurements (3 in this example).
Absolutely great explanation. Looking forward the third part.
Thank you!! I"m working on part 3 right now!
Yes me too!
thank you for the videooosss 🥹🥹🥹 help me a lot
Happy to help!