Improvement of Contact Lens Field Complaint Rate Using The DMAIC Approach
Improvement of Contact Lens Field Complaint Rate Using The DMAIC Approach
Improvement of Contact Lens Field Complaint Rate Using The DMAIC Approach
Emil A. Pietri
Manufacturing Engineering
José A. Pesante, Ph.D.
Industrial Engineering Department
Polytechnic University of Puerto Rico
MOLD ASSEMBLY
MALE MOLD
FEMALE MOLD
Figure 2
Figure 1 Contact Lens Tear
Contact Lens Assembly
Key “Proclear” Material Findings Element Description Team Charter
Process: The process in which the All Proclear Omafilcon contact Lens manufacturing at Norfolk
opportunity exists. Plants.
• There is no evidence that product aging is a
Problem Describe the problem that High Noise-Complaints are being received from sales force
possible cause for the torn and split defects Description: needs to be solved, or the in that contact lenses are tearing- before use and in use.
opportunity to be
seen on “Proclear” products in the market. addressed.
Objective: What improvement is • Standardize with automation De-molding operation.
• The study confirmed that “Omafilcon targeted? • Material improvement with in monomer tint. Reduce
handling.
Proclear” material is significantly weaker than
4. Metrics: What are the Name of Baseline Goal Entitlement* Units of
the material used as control for the study Metric Measure
measurements that quantify
(Ocufilcon). Previous studies going back as far program progress and CPM’s 100 25 3.4 to 50 CPM’s
success?
as 2001 had also shown “Proclear” Material to *What is the best the
process is expected to
be weaker when compared with other CVI produce?
5. Business Results: What is the improvement in Cost Cost WIP/ Cash Labor Inc.
materials as well as competitor lenses. business performance? Reduction Avoidance Inventory Flow Savings Sales
Please list any other Reduction
• Tensile strength analysis indicates a difference improvements on a X
between the material formulation utilized for separate sheet as needed.
Program Scope: Which parts of our business Included Excluded
most “Proclear” products and the slightly processes will be Proclear Omafilcon lens
considered? Which Product.
different material formulation utilized for customer segments, Norfolk Facility
organizations, geographies,
“Proclear” Daily Disposable lenses, with the and timeframe?
Team Members: Names and roles of team Emil Pietri Project Leader
Daily Disposable material being stronger. members Brian Charlton
Dan Earnhardt - Engineering
• Tear test show that lens strength is significantly Joe Calcagno - Manufacturing
affected by induced splits to the material. The Klaus Hummel - Manufacturing
Tom Barrett - Engineering
force required to break a good strip of a lens
Benefit to External Who are the final External customer the user will experience a improvement in
versus a strip with a 2mm split is on average 7 Customers: customers, what are their lens tearing by the beginning of March 09.
most critical
times higher (average for materials tested). requirements/measurement
s, and what benefits do we
expect to deliver to them?
Key Project Dates
PHASE I - DEFINE Schedule: Give the key milestones
and dates. Project Start May 08
Define Complete June 08
The Define phase determines the initiation of a Measure July 08
Complete
project, customer requirements and key process Analyze September 08
Complete
outputs variables. To define this process a “team Improve April 09
charter” is created (See Figure 3 above). Complete
Control Complete May 09
A
C D E F
Raw materials B Mixing of
Incoming
specification inspection Storage of monomers Storage of
Transport materials
and suppliers materials monomers
Mix and filter
HEMA
Storage conditions PC HEMA
Polypropylene resin Delivery and receipt In coming inspection monomers EGDMA
Monomers polypropylene AIBN
G H I J K L
Set up Mold Male &
molding Female Molds Convey Molds Mold Fill Mold with Load into
machine Cast Mold with cooling Monomer & Oven
Qualified Tools Assemble Male &
Female Molds
Polypropylene supply
Tool qualification Temperature & Pressure Molds cool and shrink Storage of monomer Load Robot
Barrel temperatures Timing Hamilton Pump Dispenser Chain Index
Preventive maintenance Cooling Assembly Rate Timing
Cooling water to mold plates Wetting & Capping
Mold fit
M N O P Q R
U X
S T V W Dose Blister Tubs,
De-Stack & 100%
Transfer to Hydration Hydrate Transfer to Inspection Insert Inspected
Transfer to Visual
lenses Lenses & Place Foil
Inspection Trays Inspection
Y Z BB CC
Heat-Seal AA DD
Check Sterization of
Blister Tubs Transfer to Sterilizer QC Audit
Seal Decide
lenses in of AQL Pass
and foil Width Disposition
blister tubs Quantity of Lot
Time, Temperature
Pressure, Positioning
Fail - scrap
or re-work
EE FF GG HH
Label, II
QC Audit Decide
Store Pass Transfer to
Box & of labeling Disposition
Package records Distribution centers
of Lot
Figure 4
Process Map
The team selected what key process input analyzing data from the RPM’s higher than 90.
variables were more important in respect to the The result of the FMEA analysis is summarized in
customer requirements in this case the Key (Figure 6 below) the “Pareto Chart”.
processes outputs variables KPOV’s and input them
into the FMEA (See Figure 5). This is the First
Part of the FMEA-selection to start working and
S O D R
Potential Failure Potential Failure
Process Step Key Process Input E Potential Causes C Current Controls E P
Mode Effects
V C T N
What is the
d e te c t c a u s e o r
What is the Key Process Input? In what ways does What is the impact on What causes the Key What are the existing
H o w S e v e re is
th e e ffe c t to th e
H o w w e ll c a n y o u
H o w o fte n d o e s
c u s o tm e r?
cause or FM
process step the Key Input go the Key Output Input to go wrong? controls and procedures
wrong? Variables (Customer (inspection and test) that
Requirements) or prevent eith the cause or
internal requirements? the Failure Mode? Should
include an SOP number.
P a r e to C h a r t o f K e y P r o c e s s In p u t F r o m T h e F M E A
12000
100
10000
80
8000
Percent
RPN
60
6000
4000 40
2000 20
0 0
K e y P ro c e s s I n p u t
fi t
s rs
ab
s n ss ds or
s g te e) e) s) of
)
ss
) y
o ld or
s
re
)
pe si g ce ol at lin ra m us ep ne tr a at tu
ol
d
ri m sw de ro m ar nd ng (ti ho rc ck
ck n m
ar ra
M -c r - f il e a lp of ep - ha c ki o ns a re (f o t
( la
t hi c t io a le ep pe
e u g s i i n n e m s
ol
d
ns
f
pr
o
an t in -
ol
d p it
s/
w t io ta e sp fe - em
ea o ld nd ec ac dg in in 0 (t
-m t ra nc
h m -m co it i
on sp rf (e to 10 s
de er vs H -m de e su s ck T on
st pu ic de ur nd in
t er i n g s tu G d it i
li s t c o in e n
b es a (c g m ic k co
in pr om ge li n ra st re
in
g g ut ra nd pa cu
n dl ld in i- a s to ha n s
o m le
ha -m se ol
d
de vs m
0
10
T
G
RP N 28161584 1224 720 720 720 688 640 448 400 400 224 160 144 120 112 96 64
P e rc e n t 25 14 11 6 6 6 6 6 4 4 4 2 1 1 1 1 1 1
C um % 25 39 50 56 63 69 75 81 85 88 92 94 95 97 98 99 99 100
Figure 6
Pareto Chart from the FMEA
PHASE III - ANALYSIS There was no evidence of differences in
detection capabilities at quality audits visual
The group started the process of data mining
inspection and sampling. The following graphs
and analysis where conclusion were drawn; some of
suggest that there is difference between shifts in
techniques utilized were Box plots, analysis of
picking and placing lenses very manual operation
variance, and comparison by shift and by de-
damage to lens could happen here.
molding equipment (See Figures 7, 8).
Key observations were:
Several Key findings:
• Handling and picking lenses for inspection is
• Thinner Lens Edge suffers greater vulnerability
causing more “tearing” in different shifts (See
to “lens tear”.
Figure 9).
• A significant interaction exists between
• Detection inspection audit is not different (See
Norfolk Cast Mold Line 2 a-sphere Lens (70-
Figure 10).
micron Lens edge thickness). (See Figure 7).
• Tear Rates and De-Mold GT100 #2. This
0.005
interaction is not exhibited to this degree on
other Lines or Lens Types (See Figure 8). P-Value = 0.000
0.004
The difference IS
0.050
0.045
0.040
0.035 0.000
# Lots ==> 737 674 721 653
0.030
0.025
A B C D
0.020 FLIP Shift
0.015 P-Value = 0.000
0.010 Figure 9
0.005
Edge Thkns==> 80 80 80 70 100 80 Plot difference in manual pick and place inspection
0.000
A4BL BLST EMFB PCAB PCT1 PMFB
PCCM Lens Type Each Auditor is assumed to have audited the same number of Lots; 245 each
0.005
Figure 7
Box Plot Difference Lens Type vs. Edge Thickness
0.004
P-Value = 0.415
A vg Tear Rate found, all Lots
0.05
0.001
0.04
0.03
0.000
0.02
P-Value = 0.000 HR JM RBC
0.01
A-Shift Auditor Initials
0.00
A B C D C B C D A B C D A B C D A B C D A B C D A B C D A B C D
A A A A B C C C G1 G1 G1 G1 G2 G2 G2 G 2 G3 G 3 G3 G 3 G4 G4 G4 G4 G5 G5 G5 G5 G 6 G6 G 6 G6 Figure 10
DeMold Equipment/Shift (last letter = shift) Plot no Difference in Auditors
Figure 8
Box Plot Difference in De-molding Equipment
Further studies revealed that cast mold line 2 in Asphere GT100 #4 Tear Trend vs DeMold Shift: 10/1/07-6/23/08
a combination with de mold equipment GT 100 4 Data are exponentially smoothed by factor of 0.99
line had a significant different in tear rate (See 0.11
0.105
Figure 11). There was also observed the GT 100 0.1
F L I P T ea r R a te
0.085 GT4C GT4D
Key observations: 0.08
is imperative. 0.055
Figure 13
0.26 Box Plot Difference De-molding Equipment vs. Shift
0.24
0.22
0.20
P-Value = 0.000 Thinner Lens Edge suffers greater vulnerability
0.18 to “tear” mechanisms. A significant interaction
FLIP Tear Rate
0.16
0.14
exists between Norfolk Cast Mold Line 2 a-sphere
0.12 Lens (70-micron Lens edge thickness) Tear Rates
0.10
0.08
and De-Mold GT100 #4.
0.06
0.04
0.02
PHASE VI - IMPROVEMENT
0.00
2 4 9 12
In Phase VI several improvements were
Cast Mold Line examined and implemented to reduce the incidence
of In-process lens tears, chip edges. After
Figure 11
determining that the critical to quality causes to be
Box Plot Difference Line 2 & De-molding Equipment
lens material strength, handling post cast molding
& poor inspection detection the team projects were
0.15
0.14
configured concurrently.
0.13 • The system called the “Low Volume
0.12
0.11 Automation” to reduce lens handling by
0.10 P-Value = 0.000
operators and aided with automated visual
FLIP Tear Rate
0.09
0.08 inspection system - cameras.
0.07
0.06
• The introduction of a new monomer initiator
0.05 “Perkadox” strengthen the lens to avoid
0.04
0.03 potential tearing when handling by patient.
0.02
0.01
• The introduction of an air gauge system aided
0.00 in the identification of “mold fit” the mold fit is
G4A G4B G4C G4D
GT4_Shift
an important part of the cast molding process
as it determines the gap between the female
Figure 12
mold and the male mold when together, when
Box Plot Difference De-molding Equipment vs. Shift
fit is not monitored properly and corrected at
set up the mold too loose may cause excessive It is important to note that the mold fit is
monomer material thus permitting that at the de important to the lens cast molding manufacturing
mold process lenses can be ruptured as it pulls. stage, with this measurement process the technician
can measure exactly what is the circumference of
The Air Gauge System: The nearly-vertical
the mold and make minute adjustments to the
portion of the critical diameter region in the Male
injection molding process this will avoid potential
and Female Molds is that region where seal-off
problems.(See Figure 14).
occurs, to set the trapped Monomer volume and to
form the Lens edge. This region is typically about
80 microns beyond a “transition edge.”
(See Figure 13 & 14).
LENS EDGE
FIT INTERFACE
Also called:
- Vertical land area
- Land length
LENS
FLAT
GAP
SLEEVE
LAND AREA DETAIL B
DETAIL A NITROGEN
FIT AREA
DETAIL B
18
Figure 14
Mold Fit Regions-lens Edge Figure 15
Cross section of Air Gauge a pre requisite for non extract process, meaning that
The use of the Air gauge provides the the lens will not have to undergo the process of
technician an accurate measurement, this helps the external tint addition and lens hydration, the
early detection of fit too loose or too tight which in formulation tint will be part of the monomer
turn would potentially cause variation in the preparation and will flow with cast molding
process, this variation of fit will yield product process, hydration will occur in blister sealed with
susceptible to tearing or chip edges. As evidenced saline water.
the higher the mold loosens the more propensity of
the product to be tearing at de-molding stations, the
Lens ultimate strength at break
tighter to mold fit the thinner the edge, more
susceptible to tearing as it weakens the area of edge Product using VB6 formulation
Analysis of Means Plot
contact. (See Figure 16). With 95% Decision Limits
(X 10000)
12
UDL=75089.31
10
CL=47464.92
Lines 1, 2, 4, 10, 11, 12 8
LDL=19840.54
Mean
Bartlett's Test
1 -75to-45 6
P-Value 0.000
2
2 -45to-35 57 Lots Levene's Test
A v g M o ld - F it R a n g e ( m ic r o n s )
0
Test Statistic 1.52
PCMT Scottsville
PCM Scottsville
PCTXR Scottsville
PCSPH Hamble
PCM Norfolk
PCMEP Norfolk
PC1D WC
PC1D PR
PCT Norfolk
PCXC Norfolk
P-Value 0.154
3 -35to-25 215 Lots
4 -25to-15 Product type – related to formulation
584 Lots
5 -15to-05 1173 Lots Figure 17
Brake test results new “perkadox” formulation
6 -05to+05 1061 Lots The Low Volume Automation System: The
7 +05to+15 LVA system provides a long range of capabilities
551 Lots
that enhances several issues with the lens
8 +15to+33 100 Lots performance pertaining to tears and split edges. The
system is design to automatically handle lenses
0.050 0.075 0.100 0.125 0.150
after the molding where we believe most defects
Figure 16 occur as discovered in the FMEA de-molding stage.
Tear & Edge Defect Rate Variance by Mold Fit The system provides two automatic camera
Measurement
inspections that will replace the use of human eye;
Key Highlight: The air gauge has the this will be done to 100% of lenses as if done in
capability to identify precisely de gap between regular wet lens inspection by an individual.
female and male molds fit, as observed in the data In addition, with the use of the new monomer
graph (Figure 16). “perkadox” lenses will now be hydrated in the
The new developed monomer formulation: blister it does not have to pass thru the tedious
The implementation of a new “Perkadox” Initiator process of bath hydration as exhibited in the
appears to have more strength (See Figure 17), all process map, this will also eliminate lens handling
“Proclear” product required FDA submissions to and picking the lens manually. (This is called the
the 510K as supplements, approval granted in les extract process). The system was fully implemented
than 180 days, sites were FDA inspected after the on September 2008 with exceptional results
submissions. The formulation already used in other improving internal yields pertaining “lens tearing”
products has more stability when stressed, it is also
and improving-reducing confirmed customer
complaints in the field. Size (mm)
Condition
Key component of the system: 17.5 Description
Actual 10X
X
• (AIS) Automatic inspection system and lens
Any foreign matter that cannot be
handling system. (See Figures 19 & 20). Embedded
0.34 6 3.4 removed which is embedded in
Particles
the lens matrix.
Any particle attached to the lens
Attached to
surface that cannot be removed
surface
by cleaning. Reject if exceeds
Particles 0.11 2 1.14
size or more than three.
Excess lens material protruding
Flash
from the edge.
Enclosed void in the matrix of the
lens. Exception: None in OZ.
Bubbles 0.11 2 1.14
Reject if exceeds size or more
than three.
Results of the project: The following • Part II of the FMEA-RPN’s was reduced in
indicators provided congruent data that the the critical key input variables from the
project was successfully implemented: process (See Figure 23).
R
Actions S O D R
P Resp. Actions Taken E C E P
N Recommended V C T N
800 Standarize Tom B Develop an gauge that can control "air gauge"
design low
detection, 8 10 1 80
damaged lenses
are detected
720 Study of mold fit Dan E Develop an gauge that can control "air gauge"
ve lens tear reate 8 9 1 72
720 Study of mold fit Dan E Develop an gauge that can control "air gauge"
ve lens tear reate 8 9 1 72
Figure 23
Part II FMEA (extract)
Conclusion: The optimization project to The goal was to achieve less than 3.4
reduce the amount of complaints prove to be a complaints per million for tears –split edges by
success, with improvements the market is February 2009 as stated in the charter.
expanding to Asia and growing because of the
recent excellent quality of the “Proclear” product REFERENCES
noticed in the field. [1] Breyfogle F,W, “Implementing Six Sigma” second
The “6 Sigma-DMAIC” process methods edition, 2003, published by John Wiley & Sons, Inc.
gave the team the know how to determine the New Jersey.
Key process input variables that made an effect
in the process for “Tears, Chipped Edge or
Split”[1].
.