Flame Propagation Precision Analysis Per ASTM E691 Round Robin

1
Flame PropagationPrecision Analysis Per ASTM
E691 Round Robin 7Polyimide Film Metallized
Tedlartm Tape Varying Heat Flux
Monroe Shumate 3/05
2
ASTM E691 Precision Analysis Practice for
conducting an inter-laboratory test program to
determine precision of a test method
Practice has been in use for over 25 years
3
Purpose of an Inter-laboratory Study

• Primary Provide information to prepare
  precision statement on test method
• Secondary- Provide suggestions where to look for
  improvements to test method

4
What is precision ?
Goal
Good AccuracyGood Precision
Poor AccuracyGood Precision
Good AccuracyPoor Precision
Poor AccuracyPoor Precision
5
So whats measured and reported?

• Xbar - Cell Average
• SX - Standard Deviation (Stdev) cell averages
• Sr Repeatability Stdev with-in lab variation
• SR- Reproducibility Stdev between lab variation
• r - Repeatability limit (95 confidence
  2.8Sr)
• R Reproducibility limit (95 confidence
  2.8SR)

6
What does this data show?

• X bar shows average value for this sample/test
• Sx shows how much the values will vary
• Sr shows within lab variation as compared to
  other labs
• SR shows between lab variation
• r - Within lab max. variation or noise
• should be 1/4 of the spec limit
• R Between labs max. variation or noise
• should be 1/4 of the spec limit

7
Test Data Round Robin 7Sample Description
Test Setup

• Polyimide Facing
• Metallized Tedlar Tape
• Fiber Glass AA class 0.34 pcf
• Test at 3 heat flux settings

8
Lab Participation

• 10 Labs
• 5 samples/lab

9
Equipment Setup

• Electric and Gas Radiant panels
• Set up according to FAR Standard with exception
  to heat flux
• Set up performed by each lab

10
Reporting Results

• Record heat flux, room temperature, relative
  humidity, controller temperature settings
• Report
• Flame Propagation
• After Flame

11
ASTM Analysis

• Input raw data from round robin
• Determine average, deviation, and standard
  deviation for the following
• For each lab
• For all labs combined
• Compare individual labs to all labs

12
Initial Test ResultsOne Set of Raw Data
Using this data lets perform precision analysis
a
b
c
g
h
i
j
d
13
Initial test resultsOne set of data
Spec Limit 3 sec max.
So How does this look?
14
Initial test resultsOne set of data
Is X bar (Average) ok?
Spec Limit 3 sec max.
15
Initial test resultsOne set of data
Spec Limit 3 sec max.
How is X bar when you consider Sx or variation?
16
Initial test resultsOne set of data
Spec Limit 3 sec max.
Now for the ultimate question how doesthe spec
limit compare to r or R?
Is R ¼ of the spec limit?
17
Precision analysis All data
Remember R should be ¼ of spec
18
Conclusion

• Test method is not precise for the type of sample
  tested (Need to analyze other types of
  samples)
• Both within and between lab variation needs to be
  addressed this points to both the equipment and
  operator

19
Recommendation

• Continue FAA lab inspection process to refine
  equipment setups for each lab then repeat round
  robin
• Conduct precision analysis for other types of
  samples
• Conduct FMEA failure mode effect analysis to
  list prioritize possible areas for evaluation
  and improvement
• Use data from FMEA to conduct ruggedness test
  (ASTM E1169) or DOE Optimize
  equipment/operator specs

Monroe Shumate 3/05