# Destructive Pull Test Gage R&R advice needed

E

#### ErikS

We need to validate our test method for a glue joint. We have the spec that we know we are going to be well above and capable of hitting but we still need to qualify the pull test inspection.

What I have is 100 parts made with one operator/one set up/one lot of components. To do a destructive R&R my understand is that you should have multiple batches (lot) that represent multiple parts? Perform a Nested R&R and go from there. The issue I have is with the assumption that these batches would be equal (and the fact that I don't have the resources/time to make more than the one batch I have). Can I instead just do an ANOVA with 3 operators measuring 10 parts and show that there isn't a significant difference between the operators? I have no way to prove the parts are the "same" but all I really care about at this phase is that the operators are the "same", right?

Erik

#### Bev D

##### Heretical Statistician
Staff member
Super Moderator

I need to perfrom these type of studies on a regular basis. What I do is to take several different manufacturing lots/batches to get as much product variation as I can. Usually 3 lots is sufficient. (If you know your largest component of variation that's great, if you don't it's essential to select from different lots and batches to maximize your chance for product variation in the study)

If I know that product manufactured close together in time and space are more likely to be similar than different - relative to within lot/batch adn between lot/batch variation - I will select 'pairs' or 'sets' of parts that are close together as the repeats for each test operator. If even close parts have a great deal of variation from each other, I just randomly sample from the lot/batch.

I analyze the results first graphically. The results for each operator within a lot are representative of the measurement error confounded with the part to part variation. If the variation of each operator within each of the 3 lots is 'small' compared to the variation bewteen lots, then I am certain that the 'repeatability' of each operator is small enough to detect differnces bewteen lots.

If the operators are all relatively the same within each lot then I am certain that there aren't operator differences.

It is relatively straightforward to calculate the contribution of each effect IF you use a nested study aand pull apart the various components of variation.
Using ANOVA might give you similar answers under specific conditions but it's best to stay with the nested approach as it is the techniclaly correct one and wont' give a you a weird answer that contradicts your eyes...

The keys are:
• More than one lot - 3 is the minimum
• each operator tests each lot - this is crossing the lots but not the parts like a classical R&R

E

#### ErikS

Thanks, Bev, the trick is that I don't have the luxury of multiple lots. I just have the one lot of 100 parts (a financial issue). My concern is that the assumption of a destructive test is that the parts in one lot are similar but you can't prove that part 1 of Lot A is the same as part 2 of Lot A.

Our spec is 10 lbs pull force but the numbers we get is in the 50 lbs range. With any MSA all you are doing is to mitigate the risk that you will call a good part bad or a bad part good, right? So since our actual parts are well beyond the spec limit and (should be) plenty capable (assuming a reasonable st. dev.) then all we care about is that the operator doesn't introduce some sort of preload or something to affect the measurement method.

Thus, won't ANOVA tell me whether or not operator 1 is different than op 2 or op 3?

Thanks,
Erik

#### Bev D

##### Heretical Statistician
Staff member
Super Moderator

With the knowldge you have of your process capabilty - you are correct that one lot should be sufficient. (the caveat being that if your resutls shift low part of the investigation would be to assess the contribution of measuerment error)

You still really don't need ANOVA - a simple graph of the data will tell you if you have operator to operator differences that matter. You might consider putting confidence intervals around the means but still plot all individual values. This should be sufficient.

If you woudl like you can plot your data once complete and we can run through various scenarios so you can understand the choices....

E

#### ErikS

Sounds like a plan, thanks, Bev!!

E

#### ErikS

Attached is the data. I'm open to ideas. For the fun of it I did do an ANOVA an Operator 1 was different (95% conf).

Erik

#### Attachments

• 15.4 KB Views: 335

#### Bev D

##### Heretical Statistician
Staff member
Super Moderator
I have attached my analysis. I plotted the data in a different way and you can clearly see (without aid of math) that Operator 1 is in fact different than the other two.
Of course without replication (a different lot or a different test event of the same lot) you won't know if this happened simply by chance. (replication is the foundation of all science)

The question before you now is do you need to 'fix' operator 1?

#### Attachments

• 28 KB Views: 325
E

#### ErikS

Indeed. We are evaluating the method now, but they "should" be the same. The set up isn't that involved and there really is only one way to test.

Can I make some sort of argument that even with a shift in the operator average we are still way above spec limits and set up higher defined "control limits" (not the + 3 sigma variety) in process to represent that shift? (e.g. if there is a mean shift of 5 lbs from one operator to the next then in process we should have our inspection criteria be at least 15 lbs to pass as opposed to 10 lbs which is the spec limit).

#### Bev D

##### Heretical Statistician
Staff member
Super Moderator
that does work. it is called guardbanding and is a routine approach.

#### Gurumurthy K G

##### Registered
May I get the excel format to conduct the tensile MSA study.
It may be nested Gage RR or pooled std.deviation analysis

S Looking for A Non-Destructive Crimp Test method to replace Pull Test Inspection, Prints (Drawings), Testing, Sampling and Related Topics 2
T Gage R&R studies for Destructive Test Devices (pull and push tests) Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 5
C Sample Size in Destructive Pull Test and Interpretation of Test Result IATF 16949 - Automotive Quality Systems Standard 5
H Destructive Tensile Pull Test Gage R&R Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 7
E MSA requirement for destructive Pull out test Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 1
G Destructive Testing MSA - Weld Bond Pull - Both destructive and non destructive Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 10
GR&R Destructive Data Analysis Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 11
Who are the go to companies for non-destructive hardness testing? General Measurement Device and Calibration Topics 3
Comparing data from destructive testing Inspection, Prints (Drawings), Testing, Sampling and Related Topics 7
Interlaboratory comparison or proficiency testing in destructive testing of welded joints ISO 17025 related Discussions 3
Is capability applicable for a destructive test? Capability, Accuracy and Stability - Processes, Machines, etc. 9
M Identifying Technologies for Non Destructive Examination of steel brazed joints Inspection, Prints (Drawings), Testing, Sampling and Related Topics 1
Do I need part variation while doing Destructive Variable Gage R&R MSA study Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 19
Non-destructive testing methods are special processes? Inspection, Prints (Drawings), Testing, Sampling and Related Topics 18
S Script of R software to conduct Nested GRR for Destructive Testing Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 0
S About NDT (Non-Destructive Testing) and ISO 17020 Other ISO and International Standards and European Regulations 4
Sampling Plan for Destructive Testing Inspection, Prints (Drawings), Testing, Sampling and Related Topics 2
S Magnetic Particle Inspection Non-Destructive Testing Procedure Inspection, Prints (Drawings), Testing, Sampling and Related Topics 9
C Sampling plan for destructive tests on small lots ISO 13485:2016 - Medical Device Quality Management Systems 20
S Destructive Gage RR - Using Crossed - want your thoughts Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 18
T Destructive Tensile Test Sampling Gage R&R Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 9
J Gauge R&R for Destructive Test Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 7
R Destructive Non Reproducibility MSA (Measurement System Analysis) Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 8
Manufacturing Process Validation Destructive Testing Sampling Plan Advice Inspection, Prints (Drawings), Testing, Sampling and Related Topics 1
J Critical Defect Destructive Testing Sampling Inspection, Prints (Drawings), Testing, Sampling and Related Topics 1
K How to perform Verification for a Gage that is used in Destructive Testing Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 2
S Destructive Attribute MSA for Simple Water Leak Check in Trailers Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 5
A Destructive Testing Failure Mode Sampling Plan Inspection, Prints (Drawings), Testing, Sampling and Related Topics 5
S Samples for a Destructive Test included in a Lot Size? Inspection, Prints (Drawings), Testing, Sampling and Related Topics 5
A Destructive Test Sample Size Reduction Statistical Analysis Tools, Techniques and SPC 10
A Destructive Attribute (Crossed) Gage R&R - Need Help! Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 3
O Destructive Attribute Gage R&R without LSL/USL Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 5
M Destructive VTMV Subgroups - A Destructive Variable Test Method to Validate Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 11
E Sequence of Operations - Non-Destructive Testing vs. Trim & Drill Manufacturing and Related Processes 2
C Prices for NDT (Non-Destructive Testing) Inspection for a Boeing 737-400 Inspection, Prints (Drawings), Testing, Sampling and Related Topics 1
B GR&R (Gage R&R) for Destructive Test - Question on how to? Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 2
MSA Gage R&R - Destructive Test with One Sided Specification and One Operator Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 4
C Gage R&R Destructive UTS Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 1
A Destructive Test Sample Size Determination Statistical Analysis Tools, Techniques and SPC 4
A Performance Qualification For Destructive Testing (Impact Testing) Qualification and Validation (including 21 CFR Part 11) 4
I Destructive Testing Measurement System Analysis (MSA) Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 12
C "Destructive" Linearity Study Methodology Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 6
M What Sample Size for Destructive Test and is 'T' Test applicable? Inspection, Prints (Drawings), Testing, Sampling and Related Topics 2
R MSA gage R&R (GR&R) for Destructive Test (Non-Replicable Data) Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 2
N Destructive/Non-Repeatable Measurement Gage R&R Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 5
C Batch Acceptance: Cpk requirements and Sample Size - Destructive Testing Inspection, Prints (Drawings), Testing, Sampling and Related Topics 2
P Nested vs. Crossed Gage R&R Models for Destructive Tests Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 4
D Part Variation, Unilateral Tolerance, Destructive Testing - Gage R&R Experts pls help Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 12
Destructive Attribute MSA (Measurement System Analysis) Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 9
R Destructive Attribute Gage Studies - inserted/pressed in parts Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 6