My first Gage R&R - Paper/Poly/Foil Material Bond Test, need help

zac2944

Involved In Discussions
#1
My company produces extrusion laminated flexible packaging material that is mostly used to package food and pharmaceutical products. This the paper/poly/foil material used to make pouches for many of the products you might see in a grocery store. The illustration below is a basic representation of the process. Typically an outer barrier layer and an interior heat-sealable layer are bonded together with a layer of molten extruded plastic. Bond between these layers is critical to product performance.





Our complaints and returns data shows a growing number of issues related to delamination and poor bond between layers of our material, and so I’m leading an improvement effort to improve this trend. After analysis of the issue data, one of the many projects we decided to launch is to verify with a Gage R&R that we can actually trust the data coming from our lab. We currently don’t use SPC, and simply calibrate our instruments and trust the data. This is convenient, but probably not helping us. We could be calling good parts bad and bad parts good, and making process adjustments based on inaccurate date.

One of the most important tests we do, and want to ensure we can trust, is a Bond Test based on ASTM F 904. This test involves cutting a 1” wide strip of test material, mechanically or chemically separating specific layers on one end, mounting in a tensile tester, and then measuring the average force required to pull the layers apart for 2” after the first 1”. Our specification tolerance is one sided, a minimum (ex: 400gm/in).



Our laminated material is produced in large rolls, typically 60”wide, 40”dia, and may have 20,000 linear feet of material on a roll. These rolls come off a large extrusion laminator at a rate of approximately 1 every 20-30 minutes. We may produce 10 rolls of one specification in a typical job order, and may not run that job again for months; if ever.



For every roll we produce, we roll up about 6’ of material from the top of the finished roll (roll-up sample) and send to the lab for testing as soon as it is produced. For Bond Testing we will typically cut three strips as shown. Properties of the finished material tend to very across the web, and are more homogenous along the machine direction.



I’ve got Minitab set up, I’ve read all I could here on the Cove, elsewhere on the web, and even read a copy of the AIAG MSA Manual, but I’ve still got questions. We have 3 operators, and one testing machine. I should be able to cut as many test pieces as I need from each roll-up sample, and can get 5-10 roll up samples for a Job Order, but I can’t get many different Job Orders (Batches) because we run them so infrequently.

Sorry if all the background info is too much. Below are my specific questions. Thanks in advance for any advice!


  1. What is a Batch, what is a Part, and how many do I need? If I make 10 rolls in one Job Order, is that entire Job Order (all 10 rolls) considered a batch or is each roll a batch? Is each roll considered a part or are the test pieces cut from each roll the parts? Can I use test pieces (Parts) from only one Job Order (Batch)? If not, how many batches are needed? If indeed each Job Order is a Batch, and many batches are required, then I may need years to collect enough samples since we have a long time between running the same job.
  2. What is my Number of Replicates? Assuming that each test piece is a part, and it is a destructive test, how can I test more than once? Minitab says I must have at least 2 replicates.
  3. How can I assume that all parts within a single batch are identical - Because I’m doing destructive testing, I’ve read that I must be able to assume that all parts within a single batch are identical enough to claim that they are the same part. How do I know I can make this assumption? What basis do I have? How identical is identical enough? If I’m unable to make this assumption then part-to-part variation within a batch will mask the measurement system variation. I’ve also read that I need to ensure that I select parts that represent the actual or expected range of process variation. If all parts are identical, then how can they represent a range of process variation?
  4. Which method: Crossed, or Nested? According to Minitab choosing between a crossed or nested Gage R&R Study depends on how my measurement process is set up. If all operators measure parts from each batch, then use Gage R&R Study (Crossed). If each batch is only measured by a single operator, then you use Gage R&R Study (Nested). If I only test parts from one batch, and I have 3 operators, then all operators measure parts from that batch and I use Crossed. Is this correct? I’ve also read that you have to use Nested for Destructive Testing because each part is only measured by 1 operator. So which to use; Nested or Crossed?
I remember taking statistics in school and think it was easy with all the cookie cutter examples involving dice and widgets, but when you actually need to apply these techniques in the real world you quickly realize how necessary it is to have a deeper understanding. :frust:
 
Elsmar Forum Sponsor

Miner

Forum Moderator
Leader
Admin
#2
What is a Batch, what is a Part, and how many do I need?
You just need enough to provide samples that represent your typical process variation. You mentioned that across the web is one source of variation, so collect samples across the web. How much variation to you see roll to roll and lot to lot? Just make sure to incude your major sources of process variation.

What is my Number of Replicates? See following comment.

How can I assume that all parts within a single batch are identical - You mentioned that you do have variation across the web, but not so much in the machine direction. Collect strips three times longer than normal and cut them into three normal size pieces. This will form 3 replicates. This uses the method of subdivided samples.

Which method: Crossed, or Nested? Destructive R&Rs are always Nested.
 

zac2944

Involved In Discussions
#3
Thanks for the advice Miner! I'll post the results of my study here for further comment/review. Maybe it will be benificial to others.
 
S

statdoug

#4
To Miner's excellent advice above, I would only add:
Make sure to assign strip segment (1,2,3) randomly to trial (replicate), and if you plan to have multiple appraisers (N), increase the initial length of the strips by N x 3 so that you have enough for each appraiser to check 3 replicates (again, randomized within a strip).
 

zac2944

Involved In Discussions
#5
This what my proposed sampeling plan looks like:



A, B, and C, are the operators. 3 roll-up samples, get cut into 9 strips, and sub-divided into 54 individual test pieces.

I would take a "roll-up" sample from three master rolls during a job just like we do for regular QA testing. From these 3 rolls I would cut 9 long strips (parts) across the web. The 9 strips would get cut down into 6 test pieces per strip (N=2 per operator). This would simulate each operator measuring the same part twice.

The reason I went with N=2 is because my sample cutter makes a nice 1"x12" strip, which can easily be subdivided into 2 pieces. It wouldn't be too much work to go to N=3, but statisticaly speaking what would be the advantage? Would it make a significant impact on the reliability of my results? I've got to justify this to my QA manager who already isn't excited about doing 54 extra tests.

I plan to randomize the order that the pieces are tested for each operator, and have each operator go through all thier pieces in one session since they each work a different shift. Should I also randomize where in the roll-up sample the operator's pieces come from? Right now it is in order (A, B, C) along the web direction for simplicity sake.
 
P

pawky

#6
"For every roll we produce, we roll up about 6? of material from the top of the finished roll (roll-up sample) and send to the lab for testing as soon as it is produced."

Is there any difference in the process near the end of the roll (slower feed etc?)
it may be advisable to take a sample from different points in a roll if this is the case as the last 6' may have different properties.


Also does perhaps time or temprature play a role in the material failure? if so testing it immeadiately may not be best.

BTW this relates to the regular QA tests and not to the actual gauge R&R, of which i am just learning myself.
 

Miner

Forum Moderator
Leader
Admin
#7
Overall, this looks good. I would randomize the operator order as you asked. N=2 should be okay, and I would code the strips so that an operator does not know when they are testing two parts of a sub-divided specimen.
 

zac2944

Involved In Discussions
#8
Is there any difference in the process near the end of the roll (slower feed etc?) it may be advisable to take a sample from different points in a roll if this is the case as the last 6' may have different properties.
Good question. This process is continuous. If we produce 10 rolls on a job, the machine never stops running. Actually, we might run the machine for weeks on end without ever stopping it as stopping an extrusion process is very costly. All roll-stock raw materials and finished materials are spliced "on-the-fly".

Here's a good example of what it looks like.

Since the process never stops, once you're at steady state for the job there shouldn't be much variation within a roll. There might be normal variation from rolls at beginning, middle and end of a job, and even within rolls.

Also, you cant take a sample from the middle of a roll. You would have to unwind thousands of feet to get to it, and we must maintain certain roll sizes per specification.


Also does perhaps time or temprature play a role in the material failure? if so testing it immeadiately may not be best.
Another great question. Conditionioning of the samples is very important. I will age them all under controlled conditions for 5 days, that way the time difference between tests (each on a different shift) will be small compared to the overall time since manufacture.


Overall, this looks good. I would randomize the operator order as you asked. N=2 should be okay, and I would code the strips so that an operator does not know when they are testing two parts of a sub-divided specimen.
Thanks for all the great advice, will do.
 
Last edited:

zac2944

Involved In Discussions
#9
I finished collecting my data today, and plugged everything into MiniTab for an ANOVA (Nested) Gage R&R. I added Lower Bound spec limit of 250.

I'm new at this, but it doesn't look too pretty. Any insight on how to interpret this would be helpful.

My Data
Code:
Op Part Value
1 1 452
1 1 435
2 1 440
2 1 440
3 1 428
3 1 441
1 2 460
1 2 464
2 2 449
2 2 455
3 2 454
3 2 450
1 3 482
1 3 437
2 3 459
2 3 436
3 3 455
3 3 443
1 4 435
1 4 444
2 4 421
2 4 401
3 4 426
3 4 410
1 5 460
1 5 457
2 5 461
2 5 453
3 5 460
3 5 448
1 6 457
1 6 443
2 6 441
2 6 457
3 6 447
3 6 448
1 7 430
1 7 425
2 7 425
2 7 444
3 7 442
3 7 437
1 8 462
1 8 463
2 8 452
2 8 450
3 8 445
3 8 445
1 9 480
1 9 441
2 9 463
2 9 460
3 9 457
3 9 437
Results
Code:
?????   3/26/2013 4:38:20 PM   ???????????????????? 
Welcome to Minitab, press F1 for help.
 
Gage R&R Study - Nested ANOVA 
Gage R&R (Nested) for Value
Gage name:       Chatillon TCM 201 TSN-00032
Date of study:   3/26/13
Reported by:     Z. Culliton
Tolerance:
Misc:
 
Gage R&R (Nested) for Value 
Source         DF       SS       MS        F      P
Op              2    727.3  363.630  1.08716  0.353
Part (Op)      24   8027.4  334.477  2.58876  0.009
Repeatability  27   3488.5  129.204
Total          53  12243.2
 
Gage R&R 
                            95% Lower  %Contribution  95% Lower
Source             VarComp      Bound   (of VarComp)      Bound
Total Gage R&R     130.823     81.820          56.04      30.23
  Repeatability    129.204     86.966          55.34      20.39
  Reproducibility    1.620      0.000           0.69       0.00
Part-To-Part       102.637     31.744          43.96      11.27
Total Variation    233.460    176.929         100.00
 
Lower process tolerance limit = 250
 
                                95% Lower  Study Var  95% Lower  %Study Var
Source             StdDev (SD)      Bound   (6 * SD)      Bound       (%SV)
Total Gage R&R         11.4378      9.045    68.6268     54.273       74.86
  Repeatability        11.3668      9.326    68.2007     55.953       74.39
  Reproducibility       1.2726      0.000     7.6358      0.000        8.33
Part-To-Part           10.1310      5.634    60.7858     33.805       66.30
Total Variation        15.2794     13.301    91.6764     79.809      100.00
                   95% Lower  %Tolerance  95% Lower
Source                 Bound  (SV/Toler)      Bound
Total Gage R&R         54.98       17.47      13.82
  Repeatability        45.15       17.36      14.24
  Reproducibility       0.00        1.94       0.00
Part-To-Part           33.57       15.47       8.61
Total Variation                    23.34      20.32
 
Number of Distinct Categories = 1
95% Lower Bound = 0.503971
 
Probabilities of Misclassification 
Joint Probability
Part is bad and is accepted   0.000
Part is good and is rejected  0.000
 
Conditional Probability
False Accept      *
False Reject  0.000
* WARNING * Probability of part within spec limits = 1.
 
Gage R&R (Nested) for Value
Graphs
 

Miner

Forum Moderator
Leader
Admin
#10
Beauty is in the eye of the beholder.

What is the end use to which you will put your gauge?

The %Tolerance of 17.47 means that the gauge is adequate for use as an inspection device. Your earlier concern about rejecting good product and accepting bad product is laid to rest.

On the other hand, the %Study Variation of 74.86 and an ndc of 1 means that this gauge is not suitable for SPC and will inflate the process variation observed and make your process capability look worse than it is in reality.

One other thing to keep in mind is that since this is a nested study, we are assuming homogeneity of product. If this is not true, the gauge will appear worse than if it were true. Unfortunately, I do not see a way around this issue.
 
Thread starter Similar threads Forum Replies Date
B Gage R&R Destructive Tests - Tensile tester to check for paper strength Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 3
F How often should Gage R&R's be updated? Is there a recommended time-frame? Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 4
W What is dimensional layout of the gage? Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 5
Q Does a gage used for a visual verification need to be calibrated? Manufacturing and Related Processes 7
T Gage Numbering Using GAGEpack Software 6
S Gage R&R for inline measuring machine Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 9
B Gage calibration Gonogo (calcul of cgk) General Measurement Device and Calibration Topics 0
PastorBee13 Attribute Gage R&R with no bad samples Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 3
H Air Pressure Gage Calibration General Measurement Device and Calibration Topics 6
M Spot Weld Nugget - MSA Gage R&R Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 2
D Gage type and data base maintainence Using GAGEpack Software 2
F Should I perform several Gage R&R for every caliper to cover all specs range of use? Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 9
T Gage R&R study - Ordinal data Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 6
Q Gage Pin Sets? Calibrate/Verify them all or just a sample size? General Measurement Device and Calibration Topics 22
S If 4 operate to measure, what is the value of k2?(MSA gage R&R) Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 1
S Faro Arm, Gage R&R Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 5
8 MSA--Should I Use Parts or Gage Pins? Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 2
C Gage Block Grade 1 - Calibration Measurement Uncertainty (MU) 15
R Gage Measuring 15 Parts at a Time Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 9
Q Gage calibration in disarray ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 12
M Gage R&R automated test equipment Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 5
PQ Systems Recipe for Success! 10 Ingredients for a Healthy Gage Management System Using GAGEpack Software 0
D Gage Repeatability for Faro Arm Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 3
S Comparing measurement results for a part from two ring gage General Measurement Device and Calibration Topics 3
M Quick Gage R&R Question Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 9
PQ Systems Best Practices In Gage Management Using GAGEpack Software 0
PQ Systems Gage R&R - 3 Part Series Using GAGEpack Software 0
I Excel based Gage R&R VS Minitab calculation Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 5
R Repeatability Type 1 Gage and Gage R&R Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 4
W Proper gage to use for receiving inspection General Measurement Device and Calibration Topics 9
A GAGE R&R Binomial with master list (for method validation) Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 0
Q Gage calibration / tracking software General Measurement Device and Calibration Topics 5
A R chart from ANOVA gage R&R test Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 12
M Type 1 Gage r study Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 3
T Plug Gage Calibration Calibration and Metrology Software and Hardware 1
L Gage R&R TMV Acceptance Criteria Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 4
S MSA for attribute relation gage Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 3
E Zero part to part variation - Gage R&R project Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 15
M Gage R&R and right way to measure Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 16
L Gage RandR on automated equipment. IATF 16949 - Automotive Quality Systems Standard 7
B Gage calibration frequency, ISO and IATF - What are the requirements Calibration Frequency (Interval) 3
D Difference between Test Method Validation and Gage R&R Qualification and Validation (including 21 CFR Part 11) 20
R Determining Uncertainty from Gage R&R Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 1
C Correct Calibration Method for Dial Depth Gage General Measurement Device and Calibration Topics 6
C Gage Block Wringing General Measurement Device and Calibration Topics 3
B Gage R&R with NDC=1 Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 19
S Type 1 Gage R&R or something else? Reliability Analysis - Predictions, Testing and Standards 6
G Should I perform Gage R&R only at the beginning of a new project? Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 6
V Thread Plug Gage Pitch GO Diameter out of spec AQL - Acceptable Quality Level 8
G Gage R&R - Where am I going wrong? Part of a FAIR submission (Aerospace) Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 2

Similar threads

Top Bottom