2. AGENDA
2 Company confidential
❶ Introduction to Alpha Technologies and presenters
❷ Process improvement using Six Sigma
❸ Case studies: Measurement system
❹ Case studies: Materials variation
❺ Case studies: Machine / Methods variation
❻ Summary and Q&A
4. THESIS
4 Company confidential
Six Sigma quality control and
process improvement requires a
sensitive measurement system
Legacy measurement tools are
insufficiently sensitive to capture
variation in important process inputs
Process improvement and
quality control require new
measurement tools
6. ADOPTING A DATA-
DRIVEN PROCESS
IMPROVEMENT
STRATEGY
6 Company confidential
• Define the system to be improved
and key measures
• Verify the measurement system is
sufficiently sensitive to measure
your variation
• Gather data
• Apply statistical methods to
measurements
• Conduct root cause analysis or DOE
• Implement and institutionalize
improvements
• Update control system (and charts)
Define
Measure
Analyze
Improve
Control The
DMAIC
Process
7. CONTRIBUTIONS TO PRODUCT VARIATION
Company confidential 7
SYSTEM
OUTPUT
MACHINE
Banbury operator
experience
Mill operator
experience
Type of rotors
Type of weighing
scales
Sampling protocol
Instrument sampling rate
Instrument calibration & PM
The “fishbone” diagram
introduced by Ishikawa
again emphasize
measurement system as
among the potential
sources process
variation
HUMAN MEASUREMENT
ENVIRONMENT METHODS MATERIALS
Ambient
temperature
Ambient
humidity
Elastomer source
Moisture
content of fillers
Elastomer MW
/ MWD
Mixer passes
Time to oil addition
Time on mill
9. DEFINITION
WHAT IS A RUBBER
PROCESS ANALYZER
(RPA)?
9 Company confidential
Torque and
pressure
transducer
Oscillating
grooved die
Optional die
gap sensor
Optional
sub ambient
cooling
• Introduced by Alpha Technologies
(Monsanto) in 1992 to advance
beyond simple cure testing with the
Moving Die Rheometer (MDR)
• Measures dynamic rheological
properties of rubber, compounds
and polymer melts as well as stress
relaxation tests.
• Several pre-defined tests are used to
gain deep insights to material
behavior
• Measurements can be taken at a
range of strains, strain rates and
temperatures
• Captures properties before, during
and after cure with one sample
Click to add text
10. DEFINITION
SIGNAL-TO-NOISE
RATIO
10 Company confidential
2
7
8
8
13
13
14
14
15
16
17
18
18
26
31
34
35
ML 1+4
MV SR slope
S" at 50 deg arc
S" at 1000 capm
tand at 50 deg arc
S' at 5 deg arc
tand at 5 cpm
MV SR parameter
S' at 20 deg Arc
S' at 1000 cpm
S" at 5 cpm
tand at 5 deg arc
S' at 200 cpm
S" at 200 cpm
S" at 5 deg arc
S" at 50 deg arc
S" at 20 deg arc
S" at 20 deg arc
S" at 50 deg arc
S" at 5 deg arc
S" at 200 cpm
S' at 200 cpm
tand at 5 deg arc
S" at 5 cpm
S' at 1000 cpm
S' at 20 deg Arc
MV SR parameter
tand at 5 cpm
S' at 5 deg arc
tand at 50 deg arc
S" at 1000 cpm
S" at 50 deg arc
MV SR slope
ML 1+4
• Signal-to-noise ratio (S/N) indicates
the power of a method to measure a
change amidst various causes of
background noise
• Environmental noise
• Electrical noise
• Low sampling frequency
• Mechanical changes
• A higher S/N is required to detect
small molecular or formulation
changes
• The Signal-to-Noise ratio of RPA
Measurements exceed those
available from MV Measurements by
as much as 18X
𝑆
𝑁
=
ത
𝑦2 − ത
𝑦1
𝑆𝑝
1
𝑛1
+
1
𝑛2
where:
𝑆𝑝 =
𝑛1 − 1 𝑠1
2
+ (𝑛2 − 1)𝑠2
2
𝑛1 + 𝑛2 − 2
difference in mean value of two samples
variance in measurements about the means
Measured from
Moony Viscometer
ASTM D6600, Signal to noise ratio
11. DEFINITION
SAMPLING RATE
11 Company confidential
• Low sampling rates only capture a
part of the torque signal
• Higher sampling rates improve
measurement accuracy at low strains
and high frequency
• ASTM D8059 provides guidance on
low strain performance for the
Payne effect
• S* at least 0.0200 dNm
• Strain at least 0.07 %
105
125
145
165
185
0.001 0.010 0.100 1.000 10.000
G'
kPa
Strain (Deg.)
High
Med
Low
12. CASE 1
DETECTING POORLY
PROCESSING SEALING
COMPOUNDS
12 Company confidential
0 10 20
22
25
28
31
34
37
Number of batches
MDR
MH
(dNm)
0 10 20
1.35
1.20
1.05
0.90
0.75
Number of batches
MDR
ML
(dNm)
0 10 20
1,900
1,750
1,600
1,450
1,300
Number of batches
Ultimate
Tensile
Strength
(psi)
0 10 20
700
650
600
550
500
Number of batches
300%
Modulus
(psi)
0 5 10 15 20
30.7
32.5
34.3
36.1
37.9
39.7
41.5
Number of batches
RPA
S'
at
100%
strain
(dNM)
TENSILE
MDR
RPA
PROBLEM
• RMA unable to discern good
from bad batches of rubber
used for a sealing application
(non-fill)
• MDR, Mooney and physical
property tests showed no
differences
SOLUTION
• Develop high strain (100%) RPA
method to capture dynamic
viscoelastic measures
• Good and bad batches became
immediately discernable
Cannot
distinguish
campaigns
Campaign A
Cannot
distinguish
campaigns
Cannot
distinguish
campaigns
Cannot
distinguish
campaigns
Campaign B
14. CASE 2
NITRILE RUBBERS THAT
PROCESS DIFFERENTLY
WITH THE SAME MV
14 | Company confidential
PROBLEM
• Same Mooney viscosity
parameters but customers were
reporting different processing
behavior
• Some customers reported gels
SOLUTION
• RPA frequency sweeps at low
strain
• Customer able to confirm ACN
content, MW and MWD for
each batch
• Batches with gels were now
identifiable
• Advanced methods can be used
to further characterize polymer
15. CASE 3
STUDY IDENTIFIES
POORLY PROCESSING
NITRILE RUBBER
15 | Company confidential
PROBLEM
• Same Mooney viscosity
parameters but customers were
reporting different processing
behavior
SOLUTION
• RPA frequency sweeps at low
strain
• Different lots showed different
low frequency behavior
indicating different molecular
weight distributions
• The peak in tand indicates that a
manufacturer likely blended two
different Mooney resins to
achieve spec
Variation in readings between
labs should be minimal
16. 16 Company confidential
• Dynamic rheology is most often
conducted under very small
strains (“small amplitude”)
• Most processing methods impart
large strains on materials
• Non-linear, large strain responses
can disclose many material
characteristics
• At very large strains (>100%),
torque curves become distorted
• LAOS evaluates the distorted
curve as a sum of different
harmonics
Σ
DEFINITION
WHAT IS LARGE
AMPLITUDE
OSCILLATORY SHEAR
(LAOS) RHEOLOGY?
3rd harmonic 5th harmonic 7th harmonic
17. 17 Company confidential
CASE 4
EPDM PRODUCER
CANNOT IDENTIFY
DIFFICULT-TO-PROCESS
BATCHES
PROBLEM
• Customer reporting problems
with processing batches of
EPDM raw material
• Producer unable to discern
good from bad batches
SOLUTION
• Leveraged Large Angle
Oscillatory Shear (LAOS) higher
harmonic to identify defects
• Producer can now conduct root
cause analysis and DOEs to
minimize defects
19. 19 Company confidential
CASE 5
OPTIMIZATION OF
MIXING EQUPIMENT
AND PROCESS
CONDITIONS
PROBLEM
• Study conducted to identify optimal
configuration and setpoints for
internal mixer
• 3 x 3 study conducted
• 3 rotor designs
• 3 RPM setpoints
• 5 replicates for each condition
SOLUTION
• Used RPA to identify dispersion and
work input
• Identified process configuration that
gave lowest product variability
2-Wing Rotor 4-Wing Rotor
0.53
0.54
0.55
0.56
0.57
0.58
0.59
0.6
0.61
580 590 600 610 620 630 640 650 660
tanD@20.000
Hz
G*@20.000 Hz
2 Wings (40 rpm)
2 Wings (50 rpm)
2 Wings (60 rpm)
4 Wings (40 rpm)
4 Wings (50 rpm)
4 Wings (60 rpm)
N (40 rpm)
N (50 rpm)
N (60 rpm)
Trend Line
N Rotor
20. 0
500
1000
1500
2000
2500
3000
3500
4000
0.01 0.1 1 10 100
G’
(Kpa)
Strain (%)
Payne Effect D8059
001 1-13
007 14-26
024 66-77
011 27-39
017 53-65
015 40-52
20 Company confidential
CASE 5
OPTIMIZING FILLER
DISPERSION IN
COMPOUNDING LINE
PROBLEM
• A captive compounder was having
difficulty discerning the dispersion of
carbon black
SOLUTION
• Payne Effect used to determine the
degree of dispersion of fillers in
compound
• Defective batch could be detected –
suggests that the CB weighing
method needs improved
• RPA even able to distinguish
compounds produced on different
lines
Line 1
Line 2
Defective
batch – too
much CB
21. CASE 6
LAOS TO IDENTIFY
POORLY MIXED
COMPOUNDS
21 Company confidential
PROBLEM
• Customer wanted to optimize
the mixing process of a silica
tread compound.
SOLUTION
• Used LAOS up to 1,000% strain
• Studied different mixing times
• Plotted 3rd harmonic vs. strain %
to look for peaks
• Correlated peak position (left or
right) and peak intensity to
capture quality of mix.
22. TAKEAWAYS
22 Company confidential
❶ ALWAYS verify the measurement system prior to
embarking on process improvement initiatives
❷ An RPA with sufficient sampling rate and modern
design can identify compound and polymer
differences other methods cannot
❸ Advanced methods like LAOS are powerful tools
for probing elastomers and compound to provide
details about molecular architecture