Optimizing Atomic Oxygen Resistance on Coated Substrates Using Techoptimizer®

(International Space Station)


Michael S. Slocum
Chief Scientist
The Inventioneering Company
e-mail: slocum1946@aol.com
phone: 714.641.0677

May 1998

ABSTRACT

The role of Champion will be a difficult and trying role. A paradigm shift must be accomplished to insure the successful implementation of the new methodology you know to be monumental. How will you generate support? How will you demonstrate the usefulness of this new tool? These questions must be answered or your impetus will be eroded by the mundane current business practice that may be barely successful. To prevent this from happening the new user must select a current problem and make a case study of it. The problem must be reduced to a generic, non-technical, formulation that will predicate the success of his mission to perform meaningful function analysis and innovative problem solving. This generic formulation will than allow the user to benefit from the embodiment of the TRIZ methodology in the software by using the solution principle(s) that will be presented based on the inherent patent database. This generic solution then, by analogy, must be converted to a case specific solution by the subject matter experts. The successful completion of these tasks will result in reduced cost, reduced failure modes, reduced complexity, solutions to problems, and the potential creation of intellectual property. This will give the champion the support he needs to generate corporation wide acceptance. A case study will be presented that demonstrates the successful completion of these tasks that resulted in all of the aforementioned benefits including corporate acceptance.

The Long Duration Exposure Facility (LDEF), which was retrieved by STS-32, Columbia, after a six year mission to record the effects of atomic oxygen on typical space usage materials, exposed the need for atomic oxygen buffers. This conclusion was made evident after analysis performed on the experimental plates that were exposed during the LDEF’s mission. Silicone paint is employed in spaceborne applications to provide resistance to atomic oxygen, which from 100 to 350 miles from the Earth’s surface comprises nearly 90% of the atmosphere. ITTC, aware of this research, employed silicone paint as a protective coating over the positive mate markers used on the intravehicular and extravehicular connectors that will be employed by the International Space Station.

Although the use of silicone paints decreases the effects of atomic oxygen on the substrates shielded by it, there are numerous manufacturing variables introduced into the connector system. The lack of robustness concerning the application and curing processes associated with the use of room temperature vulcanized paints causes variation in the coating that lead to premature degradation and wear. This degradation, present as particulates, becomes deadly and destructive to any impact site at elevated speeds. Therefore, the application and curing processes must be as robust as possible with very high repeatability to prevent any catastrophic anomalies.

This study focused on the optimization of the silicone paint application process employed, for use on the International Space Station. Advanced functional analysis and problem solving techniques where used that were developed by Altshuller, Tsourikov, et alia. Advanced statistical processes

were employed that were developed by Dr. Genichi Taguchi, R.A. Fischer, L.H.C. Tippett, et alia. A L27 orthogonal array and a bi-level noise factor were selected. The primary quality characteristic was percent surface area degradation, with the ideal goal the minimization of this degradation. Contribution to process variation was calculated for the major factors.

The function model and trimming module suggested (based on function rank / problem rank plus cost) that the silicone coating be removed from the system. The positive mate marker was than chosen as the site for resistance transferal. The problem solving module, Principles, suggested that a composite material be used and this stimulated this scientist to postulate a positive mate marker impregnated with atomic oxygen resistant silicone particulates. This innovate dual functionality would eliminate two steps in the manufacturing process that previously were the source of many manufacturing defects and performance flaws. The number of elements in the system were reduced 40% (approaching the ideal final result), the number of links was reduced 42%, the number of harmful effects was reduced 42%, and the cost per unit was reduced 9.33%. Manufacturing processes were also evaluated with the intention of increasing the robustness of the trimmed system. Results indicate that the pre-cleaning process was the primary contributing factor to degradation resistance (silicone primer application and curing process being second in level of process variation contribution ) . Reduction in degradation was optimized using the “nominal the best I” signal-to-noise ratio. This improved the signal-to noise ration tremendously (20.1dB) and reduced the percentage of degradation affected parts as well as the percent evident degradation in any parts exhibiting degradation (6.1 dB reduction). The percentage of parts affected by degradation was reduced from 35% to 2%. Further enhancements in the chemistry of the materials involved will decrease the dependence on the pre-cleaning process, and the primer application and curing process and will yield a much improved robustness. The first pass yield was increased from 62% to 98%. The amortized, per part, scrap cost associated with non-conforming parts was reduced from $102 to $1.60 . The annual scrap cost associated to this design was reduced from $204,000 to $2,000. More importantly, the lives of the space station crew and the mission will not be jeopardized due to damage caused by silicon debris liberated from our connector systems.

  1. Introduction

It will be assumed that at this point you have evaluated and accepted the unique abilities of TechOptimizer to innovate and solve problems. The major task currently at hand would be to present the software in the proper reference frame to demonstrate its’ effectiveness and prove its’ worth.

A particular product should be chosen to use as a case study. A cross functional team must then be formed in order to obtain the optimum output the software is capable of producing. Product selection must be made keeping in mind the fact that mature systems are not necessarily conducive to innovation. Rather, immature systems, systems designed using arcane design techniques, or designs or processes created using flawed thought processes have a much higher probability of success concerning rate and level innovation using the TechOptimizer software. Therefore, the proper project selection is critical concerning the performance of the software during this evaluation and demonstration period. The performance criteria of the selected device must be well understood so that innovation can be affected keeping the customers articulated voice in mind. A matrix should be created that lists these performance criteria in relationship to one another. This will guide the innovative process and the concept generation. At this point a function model of the existing system should be created. The generation of this function model will assist in the evaluation and justification process of the software by providing at least the following:

  1. An understanding of the components and their interactions
  2. A diagrammatic representation of harmful effects
  3. Component cost control necessary for design- to-cost-per-unit performance
  4. The ability to perform design failure mode effects analysis
  5. Optimization of system
  6. Trimming system components, approaching the ideal final result
  7. Problem management
  8. Problem solving tools (principles, effects, prediction)

The creation of the function model of the existing system is necessary to then perform the optimization techniques available to the TechOptimizer software and then compare the optimized function model with the initial basic function model. Several improvements should be noticed between the two models. The new function model should represent a system that is approaching the ideal final result, which embodies one of the primary principles underlying the methodology responsible for TechOptimizer (TRIZ). Harmful interactions should be reduced or eliminated thereby improving meantime between failure and reducing potential failure modes. This strict parallel comparison should give the team the ability to demonstrate the effectiveness and also vocalize realized cost.

The significant specification parameters are represented in the following table. It is important to note that the function analysis and trimming will create problems (listed in the Problem Manager) that must be mitigated and that this mitigation must entertain the satisfaction of the criteria listed in this table. The true merit of the software will be evident if failure mode and cost reduction are realized without sacrificing any of the critical criteria.

Rank Objective Current Proposed Current SNR Proposed SNR Current Sm Proposed Sm
5 durability cycles 100 500 5 10 20 10
1 longevity 15 30 4 8 12 6
4 voltage dissipation 5k 10k 4 8 12 6
6 complexity 16 5 3 6 12 6
8 cost 4k 2k NA NA NA NA
3 O2 resistance 6 9.5 5 9 9 2
7 pressure 40 lbf 200 lbf 5 10 6 12
2 MTBF 1.5 4 NA 12 NA 20

This particular design was existing based on previous contracted services and had only just entered low rate initial production. There were many process flaws that prevented the proper performance. Optimization was required due to the critical nature of the application. Marginal performance concerning this system was not acceptable. I would consider this to be an almost ideal performance evaluation.

The function model of the existing system was created to better understand the complex interactions and links. The model represented has been simplified and less important links have been omitted (to protect the proprietary nature of this system).

II. Function Model of Current Existing Atomic Oxygen Resistant System:

III. Function Model of Trimmed System (using TechOptimizer):

IV. Trimming Results

The following are generalized statistics representing the benefits realized if the trimming actions are realized.


Components trimmed: silicone coating, PMM primer
Components simplified: None
Functions transferred: 1
Harmful functions eliminated: 3
Cost decrease: $ 2090 per unit


before trimming

after trimming

Improvement



Number of elements: 40%

Number of established links: 42%

Harmful functions: 42%

Useful functions: 42%

Cost: 9.33%

Useful function(s) per component



V. Return on Investment Worksheet

This return on investment worksheet was used to justify the capital expenditure required to purchase the software and minimal training. The analysis was performed using an evaluation license of TechOptimizer and created sufficient financial justification to purchase the software for use on a wide array of current and future projects.


Unit Statistics

Traditional

Using TechOptimizer

Annual warranty costs due to failures

$ 25.57

$ –

Cost per unit to manufacture

$ 4,700.00

$ 2,610.00

Component cost per unit

$ 1,800.00

$ 1,500.00

Total Cost / Unit

$ 6,525.57

$ 4,110.00

Cost Savings / Unit

58.8%

Sale Price / Unit

$ 9,500.00

$ 9,500.00

Units / Annum

350

350

Total Cost / Annum

$2,283,950

$1,438,500

Total Sales / Annum

$3,325,000

$3,325,000

Total Sales – Total Cost

$1,041,051

$1,886,500

Margin

65%

65%

Profit

$676,683

$1,226,225

Delta Percent

44.8%

Delta Dollars

$549,542

Investment
Software

$10,500.00

Training / Facilitation

$3,000.00

Total Investment

$13,500.00

Return on investment in first 12 months

4070.68%

Payback period

0.024565903

months

VI. Background

Space Station prime contractors reported defects associated with the durability of the positive mate marking during installation in flight hardware systems. The importance of the reported defects warranted immediate engineering analysis concerning the mode(s) of failure and the possible contributing factors (performing . System complexity necessitated the employment of advanced methodologies in the resolution phase of this situation. Understanding the adhesion between the substrate and the silicone is of primary importance. Therefor it will be discussed in depth:

6.1 Factors Relating to Adhesive Strength:

6.1.1 Molecular Attraction: The most significant factor in adhesion is the molecular attraction present between the adherend and the adhesive. The molecular forces of attraction are the cause of adhesion and the intrinsic strength of the adhesive bond cannot be stronger than the total molecular forces in operation. There are four types of chemical bonds recognized as being factors in adhesion and cohesion: electrostatic, covalent, metallic, and Van Der Waal’s (orientation, induction, and dispersion).

6.1.2 Inherent Strength of Bond: The nature of the surface of the adherend is a vital factor in bond strength. Solid surfaces are very irregular in their constituents and ability to catalyze therefore each surface will exhibit different degrees of affinity for each adhesive. Adhesives which have a greater adhesive attraction for the adherend surfaces than cohesive attraction within themselves will give better adhesion. The forces governing surface wettability are the aforementioned Van der Waal’s forces. The wetting force of some adhesives for certain adherends are so great that surface contaminants are displaced by the adhesive. It is well recognized that clean surfaces are necessary to obtain adhesive bonds of maximum strength. The physical state of the adhesive must be such that the molecules can come into close contact with the surface of the adherend so that the molecular forces can be operative. This molecular nearness is achieved by: adhesive molecular synthesis in contact with the adherend (polymerization or condensation), Solid material applied with a volatile solvent (the design of the connectors in question as an example), solid adhesive liquefied by heating, and solid material liquefied by the application of pressure.

6.1.3 Residual Strength of Bond: The inherent strength of a bond is reduced during the process of making a bond by a group of factors which cause internal stress development on the bond resulting in lowered bond strength. Factors which cause internal bond stress are: delta from liquid to solid, delta temperature, delta composition, and aging. To bring molecular forces involved in adhesion to bear the adhesive is liquid or semi-liquid at some stage of the bonding process. To convert the adhesive into fixed or hardened state, chemical or physical change is necessary (condensation, polymerization, oxidation, vulcanization (this design), gelation, hydration, evaporation of solvents, reduction in pressure, and cooling).

6.1.4 Measured Strength of Bond: Due to the geometries involved in adhesive test assemblies, the inhomogeneity of the body, characteristics of the testing apparatus, and other contributing factors, the measured strength of the bond may be greater than, but more likely less than the residual strength of the bond. Thus, true residual bond strength is difficult to measure. Therefore, the ability to resist degradation will not be monitored as the primary quality characteristic, but percent surface area degradation will be statistically evaluated and analyzed.

VII. Objectives

The objectives of this experiment were to preclude the re-occurrence of the reported defects concerning the durability of the positive mate marking. Process optimization and innovation were pursued using the nominal the best signal-to-noise methodology as well as analyzing the sensitivity of the quality characteristic with a primary goal of reducing the percent surface area degradation as well as increasing robustness (SNR). Sensitivity must be reduced while SNR increased. The formulae for the nominal the best SNR and Sm are as shown:

equation 1.0

equation 2.0 Sm = {Syi}2 / n; i=1-n

VIII. Experimental Layout

As a result of numerous technical discussions the following process control factors, noise factor and their associated levels were selected for evaluation, see Figures 1.0 and 2.0:

Label

Factor Name

Level 1

Level 2

Level 3

A

Pre-clean

Acetone

IPA

MEK

B

Surface Prep

Scotch-brite

grit 80sandpaper

grit40 sandpaper

C

Primer application

brush

q-tip

spray

D

Primer cure temperature

ambient

100 F

160 F

E

Paint application

brush

q-tip

spray

F

Paint cure temperature

ambient

100 F

180 F

G

Paint cure RH%

30 %

50 %

70 %

H

Primer cure RH%

30 %

50 %

70 %

J

Primer location

overlap

concise

conserved

K

Paint location

overlap

concise

conserved

L

Anodic coloration

natural

gray

black

M

Particulate count

normal

extreme

conserved

N

Stroke pressure

5 ounces

10 ounces

15 ounces

Figure 1.0

Noise Factor and Levels:

Label

Factor Name

Level 1

Level 2

P

Anodic variation

Normal

Abnormal

Figure 2.0

8.1 L27 Orthogonal array:

Figure 3.0

IX. Experimental Results and Analysis

Analysis was performed with the objective of increasing signal-to-noise ratio while reducing the percent surface area degradation. To identify those factors affecting SNR and Sm, analysis of variance was performed for SNR and Sm. The decomposed contribution to SNR and Sm is indicated in the rho column of the ANOVA tables, Figures 4.0 and 5.0:

Signal-to-Noise (SNR):

Source

Factor Name

DF

S

V

F

S’

r

A

Pre-clean

2

107.59

53.79

1,052.46

107.49

44.40

B

Surface Prep

2

13.80

6.90

135.08

13.70

5.66

C

Primer application

2

15.35

7.67

150.16

15.24

6.30

D

Primer cure temp

2

15.11

7.55

147.86

15.01

6.20

E

Paint application

2

18.27

9.13

178.71

18.16

7.50

F

Paint cure temp

2

1.55

0.77

15.24

1.45

0.60

G

Paint cure RH%

2

15.35

7.67

150.22

15.25

6.30

H

Primer cure RH%

2

23.58

11.79

230.69

23.48

9.70

J

Primer location

2

15.45

7.72

151.16

15.35

6.34

K

Paint location

2

1.85

0.92

18.14

1.75

0.72

L

Anodic coloration

2

1.21

0.60

11.86

1.11

0.46

M

Particulate count

2

0.10

0.05

N

Stroke pressure

2

12.83

6.41

125.53

12.73

5.26

(e)

2

0.10

0.05

1.32

0.55

Total

26

242.09

9.31

Figure 4.0

Sensitivity (Sm):

Source

Factor Name

DF

S

V

F

S’

r

A

Pre-clean

2

32.83

16.41

94.31

32.48

49.61

B

Surface Prep

2

4.79

2.39

13.78

4.45

6.80

C

Primer application

2

4.16

2.08

11.96

3.81

5.83

D

Primer cure temperature

2

2.54

1.27

7.30

2.19

3.35

E

Paint application

2

3.44

1.72

9.90

3.10

4.74

F

Paint cure temperature

2

3.42

1.71

9.82

3.07

4.69

G

Paint cure RH%

2

3.51

1.75

10.10

3.16

4.84

H

Primer cure RH%

2

0.88

0.44

2.55

0.54

0.83

J

Primer location

2

1.96

0.98

5.64

1.61

2.47

K

Paint location

2

0.34

0.17

L

Anodic coloration

2

1.28

0.64

3.68

0.93

1.43

M

Particulate count

2

4.64

2.32

13.34

4.29

6.57

N

Stroke pressure

2

1.61

0.80

4.64

1.27

1.94

(e)

2

0.34

0.17

4.52

6.91

Total

26

65.48

2.51

Figure 5.0

9.1 Signal-to- noise ratios:

SNR and sensitivity calculations for the 27 treatments evaluated in this Taguchi methodology are indicated in the following chart. The signal-to-noise ratio and the sensitivity of the current baseline process is reflected by treatment number 8:

Treatment

Signal-to-Noise Ratio

Sensitivity

1

13.6078

33.5777

2

15.8149

32.3519

3

14.0491

31.6902

4

19.6600

32.0468

5

14.4102

31.9590

6

14.0657

31.5782

7

18.8600

31.4855

8 BASELINE

10.5289

31.5987

9

15.8739

31.0493

10

15.6049

31.8256

11

14.5509

31.7807

12

13.9941

31.5165

13

15.7641

31.5269

14

14.5164

31.1693

15

10.9996

29.5182

16

10.6804

30.2768

17

10.6804

30.2768

18

8.9560

29.4406

19

9.3152

29.3748

20

9.3152

29.3748

21

9.3457

29.7241

22

9.7743

29.5448

23

13.8507

25.0515

24

9.7799

30.0660

25

12.0310

30.7739

26

8.7086

29.5188

27

10.7535

29.7178

Figure 6.0

9.2 Response graphs:

Figure 7.0

Figure 8.0

X. Confirmation

Optimal process factors and levels based on analysis of variance ( Figures 4 and 5) and the response graphs for SNR and Sm ( Figures 7 and 8) were selected and are represented in the following table:

Variable

Factor Name

Level

Reason for Selection

A

Pre-clean

3

reduce Sm

B

Surface Prep

2

reduce Sm, increase SNR

C

Primer application

1

reduce Sm

D

Primer cure temperature

3

reduce Sm, increase SNR

E

Paint application

2

reduce Sm

F

Paint cure temperature

1

reduce Sm, increase SNR

G

Paint cure RH%

3

reduce Sm

H

Primer cure RH%

3

reduce Sm, increase SNR

J

Primer location

2

reduce Sm, increase SNR

K

Paint location

1

reduce Sm

L

Anodic coloration

1

reduce Sm, increase SNR

M

Particulate count

3

reduce Sm

N

Stroke pressure

2

reduce Sm, increase SNR

Figure 9.0

Predictions of SNR and Sm were made for the selected optimum variable levels and are shown in the table below as Optimum Theoretical. The actual confirmation data was computed to yield SNR and Sm and these values are represented in the table as Optimum Experimental. The percent error for each prediction versus actuality is also listed:

Process Condition

Signal-to-Noise Ratio

Sensitivity

Optimum Experimental

27.86 dB

25.00 dB

Optimum Theoretical

30.60 dB

25.50 dB

Percent Error

8.9

2

Figure 10.0

A total SNR and Sm comparison is reflected in the following table with dB deltas for each phase of optimization:

Process Condition

Predicted SNR

Predicted Sm

Actual SNR

Actual Sm

Baseline

N/A

N/A

10.53 dB

31.59 dB

Improved Process

30.60 dB

25.50 dB

27.86 dB

25.00 dB

Improvements

N /A

N/A

17.33 dB

6.59 dB

Figure 11.0

XI. Conclusions

Analysis indicated that the largest contributors to process variation was the pre-cleaning process, contributing approximately 44 % of the process variation. Methyl ethyl ketone was adopted as the surface preparation solvent of choice due to its’ ability to reduce variation and the overall percentage of degradation perceived in the silicone paint. This improved the signal-to noise ration tremendously (20.1 dB) and reduced the percentage of degradation affected parts as well as the percent evident degradation in any parts exhibiting degradation (6.1 dB reduction).

The percentage of parts affected by degradation was reduced from 35% to 2%. Further enhancements in the chemistry of the materials involved will decrease the dependence on the pre-cleaning process, and the primer application and curing process and will yield a much improved robustness. The first pass yield was increased from 62% to 98%. The amortized, per part, scrap cost associated with non-conforming parts was reduced from $102 to $1.60 . The annual scrap cost associated to this design was reduced from $204,000 to $2,000. The number of elements in the system was reduced 40% (approaching the ideal final result), the number of links was reduced 42%, the number of harmful effects was reduced 42%, and the cost per unit was reduced 9.33%. The proposed composite PMM and atomic oxygen protective coating is novel as well as patentable. The increase in profit was substantial. More importantly, the lives of the space station crew and the mission will not be jeopardized due to damage caused by silicon debris liberated from our systems.