Title: Flame Propagation Precision Analysis Per ASTM E691 Round Robin
1Flame 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
3Purpose 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
4What is precision ?
Goal
Good AccuracyGood Precision
Poor AccuracyGood Precision
Good AccuracyPoor Precision
Poor AccuracyPoor Precision
5So 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
7Test 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
8Lab Participation
9Equipment Setup
- Electric and Gas Radiant panels
- Set up according to FAR Standard with exception
to heat flux - Set up performed by each lab
10Reporting Results
- Record heat flux, room temperature, relative
humidity, controller temperature settings - Report
- Flame Propagation
- After Flame
11ASTM 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
12Initial Test ResultsOne Set of Raw Data
Using this data lets perform precision analysis
a
b
c
g
h
i
j
d
13Initial test resultsOne set of data
Spec Limit 3 sec max.
So How does this look?
14Initial test resultsOne set of data
Is X bar (Average) ok?
Spec Limit 3 sec max.
15Initial test resultsOne set of data
Spec Limit 3 sec max.
How is X bar when you consider Sx or variation?
16Initial 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?
17Precision analysis All data
Remember R should be ¼ of spec
18Conclusion
- 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
19Recommendation
- 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