EVALUATION OF PROPYLENE CARBONATE
IN AIR LOGISTICS CENTER (ALC)
DEPAINTING OPERATIONS
by
Foster Wheeler Enviresponse, Inc.
Edison, NJ 08837
EPA Contract 68-C9-0033
FW Project 750 181071 04
Project Officers
Johnny Springer, Jr. and Kenneth R. Stone
Pollution Prevention Research Branch
Risk Reduction Engineering Laboratory
Cincinnati, Ohio 45268
RISK REDUCTION ENGINEERING LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
NOTICE
The information in this document has been funded wholly or in part by the United States Environmental
Protection Agency under EPA Contract 68-C9-0033 to Foster Wheeler Enviresponse, Inc. It has been
subjected to the Agency's peer and administrative review, and it has been approved for publication as an
EPA document. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
-------
FOREWORD
Today's rapidly developing and changing technologies and industrial products and practices frequently
carry with them the increased generation of materials that, if improperly dealt with, can threaten both public
health and the environment. The U.S. Environmental Protection Agency (EPA) is charged by Congress with
protecting the nation's land, air, and water resources. Under a mandate of national environmental laws, the
Agency strives to formulate and implement actions leading to a compatible balance between human
activities and the ability of natural systems to support and nurture life. These laws direct the EPA to perform
research to define our environmental problems, measure the impacts, and search for solutions.
The Risk Reduction Engineering Laboratory (RREL) is responsible for planning, implementing, and
managing research, development, and demonstration programs to provide an authoritative, defensible
engineering basis in support of the policies, programs, and regulations of the EPA with respect to drinking
water wastewater, pesticides, toxic substances, solid and hazardous wastes, and Superfund-related
activities. For this project, the EPA applied the resources of its Waste Reduction Evaluations at Federal Sites
(WREAFS) Program, with the support of the Strategic Environmental Research and Development Program
(SERDP) to provide assistance to Tinker Air Force Base, Oklahoma.
The Pollution Prevention Research Branch of the Risk Reduction Engineering Laboratory has instituted
the WREAFS Program to identify, evaluate, and demonstrate waste minimization opportunities in industrial
and military operations. This report examines propylene carbonate solvent blends as a substitute for methyl
ethyl ketone in ALC depainting operations.
E. Timothy Oppelt, Director
Risk Reduction Engineering Laboratory
-------
ABSTRACT
This report summarizes a two-phase, laboratory-scale screening study that evaluated solvent blends
containing propylene carbonate (PC) as a potential replacement for methyl ethyl ketone (MEK) in aircraft
radome depainting operations. The study was conducted at Oklahoma City Air Logistics Center (OC-ALC)
at Tinker Air Force Base (TAFB). TAFB currently uses MEK to depaint B-52 and KC-135 aircraft radomes
in a ventilated booth. Because MEK is highly volatile, many gallons vaporize into the atmosphere during
each depainting session. Therefore, the U.S. Environmental Protection Agency (EPA) is supporting studies
to Identify effective, nonvolatile, less toxic substitutes for MEK.
The first phase of this study screened the performance of three solvent blends provided by Texaco
Chemical Company. These blends contained varying concentrations of PC, n-methyl pyrrolidone (NMP),
dibasic ester (DBE), and other organic solvents. The performance of each blend was compared with that
of MEK—both by the paint removal time and by a visual estimate of the amount of paint removed without
any visible substrate damage (removal efficiency). The best performer was PC Blend 2, which contained
25 percent PC, 50 percent NMP, and 25 percent DBE. This solvent blend was then compared with MEK
during the second phase of this study. The Phase 2 tests measured paint removal time and efficiency, paint
adhesion, flexural properties, weight change of the substrate after paint removal and hardness of unpainted
substrate test panels.
Phase 2 test results revealed that PC Blend 2 performed favorably in comparison with MEK in removing
paint from the fiberglass/epoxy (F/E) test panels and in subsequent paint adhesion tests, despite an
indication of possible substrate damage. A preliminary economic analysis performed on PC Blend 2
estimated TAFB would save over $30,000 the first year of operation by replacing MEK with PC Blend 2. PC
Blend 2 should continue to be evaluated as a substitute in the TAFB radome depainting operation.
Additional qualification testing, required by the Air Force, and a full-scale demonstration project are
recommended before implementation.
This report is submitted in fulfillment of EPA Contract 68-C9-0033 under the sponsorship of the U.S.
Environmental Protection Agency Risk Reduction Engineering Laboratory (RREL). This report covers a
period from February to September 1993 and work was completed as of September 30,1993.
IV
-------
CONTENTS
Page
Notice ................... ........ • • ............. • ....................... .!! \;
Foreword .......... ...... ..................................... ..... ..... ;
Abstract
Figures
vj
Tables ........ . ....... .................................................. ..
Abbreviations and Symbols .......... ................ • ................. ....... ."
Acknowledgements ...... . ........ ----- ---- .................... • • ---- ; ...... - Vl"
1. Introduction • • '
2. Conclusions •
3. Recommendations • •
4. Background • • • • • • '
Existing Depainting Procedure • • °
PC Solvent Blends • • • ®
5. Test Program Methodology • J°
Prescreening • • •
Screening Procedures • "
Evaluation Tests • • • • • • ••-.•• -'
6. Results and Discussion • • • • ^
Screening Procedure 1°
Evaluation Tests - - • • • • • • ^1
7. Cost Analysis • • • • „
8. Quality Assurance Analysis • • • ~*
Data Quality • • '• ^
Deviations • •
33
References
Appendices
A. Material Safety Data Sheets
13. Procedure for Weight Change Test
C. Laboratory Test Results
-------
FIGURES
Number
1 TAFB radome depainting operation
Page
- 7
TABLES
Number
1
2
3
4
5
6
7
8
9
10
Solvent Properties Summary
Laboratory Tests and Sample Requirements Summary
Screening Procedure Results
Depainting Simulation Results
Hardness Test Results
Rexural Properties Results
Paint Adhesion Test Results
Weight Change Test Results
Economic Comparison of PC Blend 2 and MEK
Data Quality Requirements Summary
Page
8
12
16
18
19
21
23
25
26
29
vi
-------
ABBREVIATIONS AND SYMBOLS
ALC Air Logistics Center
ASTM American Society for Testing and Materials
DBE .dibasic ester
EPA U.S. Environmental Protection Agency
°F Degrees Fahrenheit
F/E fiberglass/epoxy
FWDC Foster Wheeler Development Corporation
FWEI Foster Wheeler Enviresponse, Inc.
HaO water
HP horsepower
in. inches
L liter
lbf pound force
MEK methyl ethyl ketone
mil unit of thickness; 1 mil = 0.001 inches
mm/mm millimeter/millimeter
MSDS material safety data sheet
NMP n-methyl pyrrolidone
OC-ALC Oklahoma City Air Logistics Center
PC propylene carbonate
psi pounds per square inch
PPRI3 Pollution Prevention Research Branch
QAPP Quality Assurance Project Plan
RREI. Risk Reduction Engineering Laboratory
RSD percent relative standard deviation
SEM scanning electron microscope
SERDP Strategic Environmental Research and Development Program
SS stainless steel
TAFEJ Tinker Air Force Base
TCLP Toxicity Characteristics Leaching Procedure
UCT universal coordinated time
WREAFS Waste Reduction Evaluations at Federal Sites
vii
-------
ACKNOWLEDGEMENTS
This report was prepared under the direction and coordination of Johnny Springer, Jr. and Kenneth
Stone, Project Officers, Pollution Prevention Research Branch in the Risk Reduction Engineering Laboratory,
Cincinnati, Ohio. Contributors and reviewers for this report were S. Garry Howell of the Office of Research
and .Development, Angela Burckhalter, Carlos Nazario, and Albert Arrieta of Tinker Air Force Base. The
authors also wish to acknowledge the assistance of Doug Culpin and Dr. Tom Marquis of Texaco Chemical
Company.
This report was prepared for EPA's Pollution Prevention Research Branch by Seymour Rosenthal and
Ann Hooper of Foster Wheeler Enviresponse, Inc. (FWEI) under Contract 68-09^0033. Assistance was
provided by Jeffrey Blough and Stuart Van Weele of Foster Wheeler Development Corporation, Livingston,
NJ, and James P. Stumbar of FWEI.
viii
-------
SECTION 1
INTRODUCTION
This report summarizes a two-phase, laboratory-scale screening and measurement study to evaluate
the potential of substituting solvent blends containing propylene carbonate (PC) for methyl ethyl ketone
(MEK) in radome depainting operations at Tinker Air Force Base (TAFB)., By eliminating the use of MEK,
TAFB will reduce its use of this hazardous substance and the resulting problems associated with its disposal
and air emissions.
This project is supported by the U.S. Environmental Protection Agency's (EPA) Risk Reduction
Engineering Laboratory (RREL). In keeping with the Agency's responsibility to advise and cooperate with
otherfederal departments on environmental risk reduction, the Pollution Prevention Research Branch (PPRB)
has managed a technical support effort known at the Waste Reduction Evaluation at Federal Sites (WREAFS)
Program WREAFS was established to conduct research, develop, and demonstrate opportunities to reduce
the'generation of waste from federal activities. Through WREAFS, the EPA provides support to federal
facilities in researching, developing, and demonstrating pollution prevention technologies and transferring
lessons learned from the federal community.
Since 1988, WREAFS has conducted research and development efforts under funding from both EPA
and other federal agencies via interagency agreements. In 1990, Congress established the Strategic
Environmental Research and Development Program (SERDP) as a multiagency effort to support
environmental research and development programs. One of the objectives of SERDP that is addressed by
this project is the identification of technologies for national defense purposes that assist government and
private sector organizations in addressing environmental concerns. This report is a deliverable of the
WRIEAFS/SERDP programs.
This project also focuses on EPA's 33/50 Voluntary Reduction Program, whose goals are to reduce
generation of 17 hazardous substances by 50 percent by the end of 1995 based on the 1992 baseline. One
of those 17 chemicals is MEK, a solvent used in aircraft maintenance operations such as cold cleaning
applications for removing resins, coatings, and adhesives.
From previous research and background documents (2), RREL has identified PC as a possible
alternative to MEK. To date, PC has not been performance-tested as a substitute for MEK in radome
depainting operations. The primary objective of this project was a proof-of-concept study to evaluate the
performance of solvent blends containing PC as a substitute for MEK.
The program was conducted in two phases. The first phase of the study screened three solvent
blends containing PC in varying concentrations; all three blends were provided by Texaco Chemical
Company (Texaco). Texaco chose a blend rather than pure PC due to the fact PC's properties would be
enhanced by the addition of other solvents. The performance of each blend was compared to that of MEK
as measured by the paint removal efficiency and the amount of visible substrate damage from radome test
panel specimens provided by Oklahoma City Air Logistics Center (OC-ALC).
-------
Since the PC solvent blends showed adequate paint removal performance from the radome test panel
specimens, the most effective solvent blend was chosen for additional testing and comparison to MEK. The
second phase included radome test specimen depainting using a simulated depainting spray procedure to
determine the paint removal time and efficiency as compared to the current procedure at OC-ALC. The tests
performed on the ^specimens determined the effect of the PC solvent blend and MEK on the composite
substrate and included flexural properties, hardness, weight change, and paint adhesion. Microphotographs
of the surface and interface of the failed flexural test panels provided visual evidence of possible damage
or lack of damage between the radome fiber and epoxy matrix. >
The sections that follow provide summarized conclusions of the study and recommendations for
additional studies to confirm the potential use of a PC solvent blend as a viable substitute for MEK and
provide steps for its introduction into the existing TAFB depainting operation. The report discusses the
background and methodology of the overall test program. Test results, cost and quality assurance analyses,
and references are provided. Material Safety Data Sheets (MSDS) for the solvents, specific test
methodologies and protocols, and definitive laboratory test results are provided in the appendices.
-------
SECTION 2
CONCLUSIONS
The evaluation results indicate that PC Blend 2 is a potentially viable solvent for replacing MEK; further
evaluation will be required for proper qualification of its use in TAFB's radome depainting operations.
During screening, the three PC blends provided by Texaco removed paint from condemned radome
specimens and appeared to perform better than the pure MEK. PC Blend 2 performed the best and was,
therefore, selected for further evaluation tests. These tests included depainting simulation, hardness, flexural
properties, paint adhesion, and weight change.
Depainting simulation results were invalid because six of the eight test panels contained a different
paint system from the one TAFB currently uses. Although the paint system is no longer used, PC Blend 2
removed the paint faster and more completely than the MEK. The two panels painted with the current
radome paint system were tested with PC Blend 2 and were depainted in less than an hour. Because of
the inconsistencies in the test panels, results of this test were disregarded and not used in subsequent tests.
To obtain removal time and efficiency for depainting, panels painted with the current paint system were
depainted in the simulation unit and then repainted for the paint adhesion test. PC Blend 2 removed the
paint in slightly more time than MEK and required a little more scraping for total removal. Both solvents
removed 100 percent of the paint. After depainting, the panels were observed indicating possible removal
of the top layer of resin by PC Blend 2. The impact of this observation requires further study.
Results from the hardness test indicates that neither PC Blend 2 or MEK embrittle the fiberglass/epoxy
(F/E) substrate. The solvents did not affect the flexural properties. Examination of the failed test panels with
a scanning electron microscope (SEM) indicated no significant damage to the fibers or the fiber matrix
interface. The weight change test panels exhibited weight loss for both solvents after immersion for four
hours, although the amount was negligible. The MEK-immersed panels indicated lower weight loss. The
paint adhesion rating for both MEK and PC Blend 2 represented complete paint adhesion after a depainting/
painting cycle.
Due to limitations of the test procedure, test panels used in the hardness, flexural properties, and
weight change testing were cut from prepreg sheets; test panels for screening, simulated depainting, and
paint adhesion testing were cut from condemned F/E radomes. However, since the F/E substrate is the
same for all tests and there is direct comparison between MEK and PC Blend 2 for each test, it is assumed
that conclusions for all tests are valid.
-------
SECTIONS
RECOMMENDATIONS
Based on the tests performed for this evaluation, PC Blend 2 is a potential substitute for MEK in TAFB's
radome depainting operation. Since the evaluation tests indicate that PC Blend 2 may damage the
substrate, further study is required before full-scale implementation of this solvent can occur. Possible
alternative future courses of action are:
• Further evaluation of the potential adverse effects of the solvent on the substrate, and either
• Reformulation of the solvent blend to eliminate or reduce any identified damage, or
• Full-scale demonstration project of PC Blend 2 in TAFB's radome depainting operation if no damage
is Identified.
The first alternative entails evaluation of whether the extent of resin substrate removal would preclude
the use of the PC Blend 2. This evaluation requires thickness measurements of the resin substrate. The
depainting operation should be performed over several depainting/painting cycles; thickness of the resin
coat should be determined and compared with the minimum allowable thickness to determine suitability of
the PC Blend 2. These tests should be repeated with MEK from the sump to permit valid comparison.
Texaco suggested the second alternative, reformulation of the blend to prevent or lessen the substrate
damage. This requires tests to determine which blend component damages the substrate, followed by
testing of a reformulated Wend. The component that causes damage may be isolated by formulating three
solutions: PC/NMP, PC/DBE, and NMP/DBE, each with the same concentration as in PC Blend 2. Radome
panels would be tested with these three solutions to determine paint removal efficiency and substrate
damage. PC Blend 2 could then be reformulated and tested with less of the substrate damaging
component.
In a full-scale demonstration project, several areas should be addressed. One area is the disposition
of the spent solvent. The EPA does not identify the three components of PC Blend 2 (PC, NMP, or DBE)
as hazardous in 40 CFR, Part 261. According to MSDS, the three components are all biodegradable and
can be sent to an industrial treatment plant, if local regulations allow. However, this should be tested on
a small scale before actual discharge of spent solvent to TAFB's treatment plant. E.I. duPont de Nemours
(DuPont) recommends DBE, as a pure component, be recycled in a vacuum distillation unit which indicates
that PC Blend 2 can also possibly be distilled and reused. Also, to ensure that the spent PC Blend 2 is
disposed properly, testing procedures, such as Toxicity Characteristics Leaching Procedure (TCLP), should
be performed on the spent solvent to confirm the nonhazardous designation.
Also to consider is the logistics of retrofitting the existing MEK operation. The three components are
not a hazard concern, as is MEK. However, these solvents have some equipment limitations, especially with
pump seals. Incompatible materials are Buna-N, Viton®, and Hypalon. Recommended materials are Teflon™,
polyethylene, and ethylene propylene rubber, in addition, PC Blend 2 is not volatile and will remain in the
-------
sump for reuse until operators determine it is ineffective. A filtration device may have to be installed in the
sump to prevent clogging the pump and piping by paint chips. Depainting booth operators should continue
to wear personal protective equipment including respirator, chemical resistant clothing, eye protection, and
gloves. - •-'--
In addition to equipment, the operation must be slightly altered. The time required to remove paint
with PC Blend 2 is relatively comparable to MEK. As with the current MEK operation, manual scraping will
probably be necessary when using PC Blend 2. However, in the current operation, the MEK evaporates
quickly from the radome surface, avoiding any need for a drying step. PC Blend 2 has extremely low
volatility and will not quickly evaporate from the surface. An additional step to remove solvent from the
surface will have to be added to the depainting operation. The radomes could be heated to evaporate the
solvent from the radome surface, by either using infrared panels, moving them to a heated booth, or by
spraying them with a heated air curtain. Texaco has also suggested using a volatile solvent, such as
dipropylene glycol monomethyl ether, which will dissolve the PC Blend 2 and evaporate. As of this writing,
dipropylene glycol monomethyl ether is not characterized as an ozone depleter or hazardous air pollutant
under the Clean Air Act. Use of this chemical requires personal protective equipment as in the depainting
operation. It is important that any solvent used for this purpose should, as a minimum, be nonhazardous
as a solid waste and air pollutant. Part of a qualification program for PC Blend 2 may include testing a
solvent for drying.
-------
SECTION 4
BACKGROUND
EXISTING DEPAINTING PROCEDURE
Aircraft radomes are painted for appearance and protection from the environment. These protective
coatings applied to the radome must not adversely affect the operation of the enclosed equipment, such
as radar. Military aircraft radomes are typically fabricated using fiberglass and epoxy or polyester composite
material. Large radomes usually consist of a honeycomb structure with a thin coating. The protective
coating on F/E radomes consists of a primer and polyurethane rain-erosion coating, followed by a
pdyurethane antistatic topcoat. Removal of these protective coatings, without damage to the F/E substrate,
presents many challenges.
Currently, the radome depainting task at TAFB is performed using MEK. Radomes handled by TAFB
Include those from KC-135, EC-135, B-52, B-1, and E-3A aircraft. A radome piece can either be a nose
radome, "top hat," or side radome. The nose radome, located on the tip of the aircraft, is an oblong dome
shape. The "top hat" is the piece located above the nose radome on the aircraft; side radomes are located
on the sides of the nose radome. Paint systems vary for different radomes.
Radomes requiring repair are removed from the aircraft and sent to the Composite Repair Facility
(TAFB Building 2211), which houses the radome depainting operation. Figure 1 depicts TAFB's radome
depainting operation. In this operation, paint is removed in a large ventilated booth by subjecting it to a
MEK shower to loosen the paint. The MEK attacks the primer via scribed breaks in the topcoat. According
to operators, the paint starts to bubble after about 30 minutes of continuous showering. As the primer
dissolves, the topcoat is flushed away from the radome by the MEK shower.
Topcoat residue is filtered from the MEK and flows to a sump for recycling back to the spray header.
The operation typically takes 1 >6 to 3 hours. According to TAFB, a large percentage of the MEK is lost to
the atmosphere through the booth exhaust system. After the MEK depainting process, the remaining paint
residues are removed by hand sanding. Topcoat chips are captured in a sump and disposed as hazardous
waste. In 1991, 719 pounds of topcoat chips were disposed, and an estimated 8,250 gallons of MEK
evaporated to the atmosphere.
Because of its properties, MEK is an ideal solvent for the radome depainting operation. MEK lowers
the viscosity In paints and adhesives and is relatively inexpensive. However, MEK exhibits the following
chemical and physical properties which contribute to its classification as a hazardous material:
« Rammable
• 1.8% lower explosive limit
• 20 °F flash point
-------
-------
• High vapor pressure
• High evaporation rate
• Not miscible with water to any great extent
For these reasons, EPA and TAFB want to eliminate the use of MEK and replace with a nonhazardous
or less hazardous alternative.
PC SOLVENT BLENDS
Texaco provided the solvent blends containing PC. The solution compositions (by weight) are as
follows:
• PC Blend 1
33%% propylene carbonate
33V4% n-methylpyrrolidone
331/*% dibasic ester
• PC Blend 2
25% propylene carbonate
50% n-methyl pyrrolidone
25% dibasic ester
• PC Blend 3
15% propylene carbonate ;
15% n-methylpyrrdlidone
15% methyl Isoamyi ketone
40% dibasic ester
15% dipropylene glycbl monomethyl ether
Texaco provided the physical properties of these blends, summarized in Table 1, along with properties
of MEK. Texaco also provided the MSDS for these blends, included in Appendix A. An MSDS for IMEK is
also included. Characteristics of the three major components of the solvent blends are discussed below.
TABLE 1. SOLVENT PROPERTIES SUMMARY
Solvent/Blend
MEK
PC Blend 1
PC Blend 2
PC Blend 3
Flash
Point
(°F)
20
215
210
144
Vapor
Pressure
(mm Hg @ 68° F)
70.21
0.18
0.20
6.51
Specific
Gravity
(@77°F)
0.8023
1.0985
1.0797
0.9947
Viscosity
(cps@77°F)
0.41
2.175
2.105
1.919
PC Is a clear liquid which has excellent solvent properties of high flash point and low toxicrty; a low
evaporation rate eliminates the concern of air emissions. PC is typically used as a solvent in such
applications as surface coatings, dyes, fibers, and plastics. !
8
-------
NMP is a highly-polar, general-purpose, organic solvent. Like PC, NMP also has a low evaporation
rate and high flash point. Typical applications of NMP include paint removal and engine cleaning.
DBE Is a product manufactured by DuPont and is a mixture of dimethyl succinate (24 percent by wt.),
dimethyl glutarate (60 percent by wt.), and dimethyl adipate (15 percent by wt.) and <1 percent of methanol
and water. DBE is a clear, colorless liquid with a mild odor. It is readily soluble with alcohols, ketones,
ethers, and most hydrocarbons, and only slightly soluble in water and higher paraffinic hydrocarbons. DBE
is stable at normal ambient conditions and, therefore, can be handled and stored without need for
precautions to prevent auto-oxidation or hydrolysis. DBE is principally used as an industrial cleaning solvent
and paint remover.
-------
SECTION 5
TEST PROGRAM METHODOLOGY
An EPA Level IV Quality Assurance Project Plan (QAPP) was approved and used as the basis For the
propylene carbonate solvent evaluation.
The overall objective of this study was to determine whether a propylene carbonate solvent blend
removes paint from a radome without damage to the F/E substrate.
Prior to this program, Texaco laboratory personnel conducted an independent prescreening exercise.
FWEPs evaluation consisted of a screening phase and an evaluation phase. The prescreening and two study
phases are discussed below.
PRESCREENING
For the prescreening, Texaco developed a computer program to predict properties of various solvent
blends and to select potential blends, based on criteria entered into the program. Each blend contained
PC. A Wend was chosen rather than pure PC to enhance paint removal properties of PC. The selection
criteria for the PC solvent blends was developed by FWEI and included: j
• Nonhazardous mixture
• Low volatility
• Safe to handle
• Rash point > 140°F
• Biodegradable
Texaco's database produced over 60 solvent blends. The blends were made up and tested on panels
cut from a condemned radome by applying solvent to the panel's surface with an eyedropper.
In addition, commercial paint removers Zip Strip and Klean Kutter were selected as standards.
According to Texaco, these paint removers softened paint in 1-2 minutes and scraped off fairly easily. The
solvent compositions are: \ ' ' . .,
Zip Strip Klean Kutter
79% methylene chloride 35% methylene chloride
9% mineral spirits 20% toluene
9% ethanol 20% acetone
3% metnanol 25% methanol
PC Blends 1 and 2 removed the paint as well as, or better, than Zip Strip and Wean Kutter. PC Blend
3 and two other formulations attacked the paint less vigorously, requiring slightly more scraping effort to
remove it. The other two formulations were:
10
-------
• 25% propylene carbonate ;
50% n-methylpyrrolidone
25% methyl isoamyl ketone
• 25% propylene carbonate ,
50% n-methylpyrrolidone
25% n-amyl acetate
The three solvent blends selected for testing were made up in one-gallon batches and sent to the
TAFB laboratory.
FWEI conducted Phase 1 screening procedures at TAFB. Phase 2 was the evaluation of substrate
damage and paint adhesion of the solvents. The tests are described in the following sections and
summarized in Table 2.
SCREENING PROCEDURES
The purpose of the screening test was to determine if any of the propyiene carbonate blends remove
paint as effectively as MEK and if so, to select the best of the three blends for more detailed testing.
Screening was performed at TAFB Chemical Laboratory in Building 3001. The screening phase
involved qualitative testing of the PC solvent blends to determine their effectiveness in removing paint from
radoime panels. Pure MEK was evaluated for comparison. The screening test was performed in a laboratory
hood.
The three solvent blends and MEK were each poured into two liter beakers with enough volume to
ensure total immersion of the test panels. TAFB provided approximately two in. square test panels made
of F/E honeycomb from a condemned radome. The panels were unscribed. The radome was a "top hat"
section from a B-52 aircraft The back side and edges of the test panels were masked with aluminum foil
tape to prevent damage to the honeycomb structure. The panels were immersed face down in the bath.
After 30 minutes, the panels were removed and visually examined for signs of bubbling of the paint
coating. Observations were recorded. The specimens were put back into the bath and examined at 1, 2,
4, 8, and 24 hour intervals after initial immersion. During each examination, the amount of paint removal
effectiveness was estimated and recorded. As recommended by the TAFB Materials Group, the panels
remained in the bath for 24 hours, even if the paint was removed prior, to provide some visual observation
on any possible damage to the substrate.
EVALUATION TESTS
Although the three solvent blendsi removed paint effectively, PC Blend 2 performed best during the
screening procedure and was selected to undergo further testing. The testing included a simulation of the
depainting procedure to determine the time and removal efficiency for a spray operation, with a more
accurate representation of the depainting operation. TAFB chose other tests to analyze the solvent's effects
on the substrate. Tests evaluated hardness, flexural properties, paint adhesion, and weight change. Foster
Wheeler Development Corporation (FWDC) personnel performed the tests at FWDC's John Blizard Research
Center in Livingston, New Jersey.
11
-------
CO
i
LU
i
a
UJ
cc
o.
I
Q
111
§
DC
O
m
CQ
03
Q. T3
H- .2
® ?5
teff
1
3
Z
Solvent
I
£
ubstrate
CO
1
**
u
£
T- CM CO
-a -a -a ON
CO CD CO ^
CD CQ CQ vf ^x
/% -_ -» in m
a. a. a. 2 2
0)
1
CO
I
S.
1
a
i honeycomb
•£
5
"-
<
z"
f
i
o
CO
CM *3" CM O
CM CM
TJ TJ
1 I
CQ NX CD xx
* % iij f ^ M i
0.2 Q.2
1
13 I
CO
•Si •§
st 5
£ to* c
If c5
i honeycomb
jb.
5
UJ
u.7
<
z
c
o
ta
•3
CO
O)
1
a
•* ri-
CM
C
5 N£
f_5 l^J
0.2
1
c
preg laminate
S
Q.
UJ
uT
I
CM
Q
CO
i
i
CO
m in
CM
CQ sx
CD UJ
Q. 2
1
i
c
preg laminate
CO
0.
UJ
UL
9
S
0
t'
2
Q.
1
1
LL.
Tf TT
CM
CQ v^
C3 ^^
0.2
CO
1*
c
s.
i
O
i honeycomb
.•e
5
UJ
u.
in
8
O
o
1
i
^. ^J. <4- Tt
CM CM
1 1
CD CO
CQ ^ OQ y
o uj o !^
a. 2 o. 2
"8
S
1 3.'
•1
~ ^
£ CO
a "i
§. ^
2 co
a. co
UJ (0
"»•» T-
U. CO
^_
z
1
(0
6
D)
1
-------
Simulated Departing Procedure ;
The simulated depainting procedure was set up for both PC Blend 2 and MEK, in two separate,
identical units. The apparatus consisted of a Kleer-flo Cleanmf ster parts washer fitted with a Vi in. diameter
spray nozzle. The flow rate of the parts washers was approximately seven liters per minute. Both units had -
to be modified with a 0.07 HP orbital magnetic drive centrifugal pump because the original pump's seal
material was incompatible with both MEK and PC. The nozzle sprayed the two in. square test panels made i
of F/E honeycomb cut from a condemned B-52 radome. The panels were initially masked with aluminum '
foil tape on the edges and back to protect the honeycomb structure, but after it was noticed the tape was ;
holding down the edges of the paint and interfering with the paint removal process, the tape was pulled back :
from the front edges of the panel.
The times required for the solvents to bubble and totally remove the paint were recorded. Along with
the time, a qualitative judgment of paint removal was removed was recorded at various time intervals. After '.
bubbling, the paint was removed by hand or a blunt-edged wooden spatula. :
Hardness i
Hardness tests were performed in accordance with ASTM Test Method D 2583-87. Four test panels ;
were used for each solvent. The panels were two in. squares cut from an unpainted F/E prepreg sheet used
for another project at TAFB. A prepreg panel is made of fiberglass fibers that have been soaked or !
impregnated with a polyester or epoxy resin. The surface of a prepreg panel is identical to an unpainted
F/E radome surface. Fabricated prepreg panels are used in tests where an actual radome cannot supply
the required surface conditions or dimensions. Panels from the prepreg sheet had to be used for this test
because the Barcoi Impressor requires a flat surface for measurement. The condemned radome panels '
available for this experiment had a slight curvature, making them unacceptable for this test. !
The hardness test determines indentation hardness of a material using a Barcoi Impressor, Model No. ;
GYZJ 934-1. Indentations are made on the specimens and the hardness measured. In accordance with
ASTM D 2583, ten measurements were made on each of the four panels, both before and after contact with
the solvent. The measurements are no less than Va in. from the edge or from other measurements. The ;
Barcoi Impressor was checked for calibration prior to each set of measurements with standard test disks
supplied with the impressor. The panels were sprayed with solvent in the depainting simulation unit for two
hours, and gently wiped dry with a paper towel.
Flexoral Properties Test
The flexural properties test was performed in two parts. The first part measured flexural properties of '
the test panels in accordance with ASTM Test Method D 790-92, Test Method I, Procedure A. Five test
panels were used for each of the solvents. The test panels are made from an F/E prepreg sheet used for ;
another project and were provided by TAFB and cut to test specifications. Panels from the prepreg sheet
were used instead of condemned radome pieces because flat panels with 1" x 3" dimensions were required
for this test. The condemned radome panels available for this experiment had a slight curvature, making
them unacceptable for this test.
This ASTM test measured the flexural strength of the panel and allowed inspection to determine if any ;
damage by the solvents occurred to the interface of the F/E and laminate structure. The panel was first ;
subjected to the solvent in the depainting simulation unit for two hours. After drying one hour, the panel
was subjected to load in a tensile machine until breakage of outer fibers occurred or when a maximum strain
of 0.05 mm/mm was reached. :
13
-------
During the second part, the failed test panels were examined with a scanning electron microscope
(SEM). To prepare for SEM examination, the panels were cut into one in. squares with the failure break in
the center of the panel. The panels were then mounted onto an aluminum stud using carbon paint, which
provides a conductive bond between the stud and the panel. The test panels were then sputtered with gold
in a vacuum chamber to make them conductive. After about ten minutes of sputtering at various angles to
ensure the gold is applied underneath the fibers, the panel is then placed in another vacuum chamber and
viewed by the SEM. The surface was examined at 30X, 300X, and 1200X magnification and the cross-
section was examined at 1200X magnification to observe any fiber/matrix interface disbonding.
Microphotographs were made at each magnification for representation of the conditions observed. Test
panels not exposed to any solvent were also viewed by the SEM to obtain a baseline comparison.
Paint Adhesion
Paint adhesion testing was performed in accordance with ASTM Test Method D 3359-92a, Method A.
Four unscribed F/E with honeycomb test panels were used for each solvent. The test panels were painted
with TAFB's current paint system.
After paint removal with the solvent in the depainting simulation unit, the test panels were repainted
by TAFB personnel with the rain-erosion coating system currently applied to B-52 radomes. The same
thickness of coating and painting procedure used for actual radomes was applied to these test panels,
approximately a 10 to 12 mil coat For the. tape adhesion tests, two incisions were made in the panels to
the substrate layer. A pressure sensitive tape was then applied to the intersection of the cuts for a period
of 90 ± 30 seconds. The tape was one in. wide PermaCel 99™ (manufactured by PermaCel, New Brunswick,
NJ). After tape removal, the X-cut area was visually inspected. The adhesion is rated according to a scale
of OA (removal beyond area of X) to 5A (no peeling or removal).
Weight Change Test
The purpose weight change test is to determine if the substrate has been damaged. TAFB developed
the procedure for performing the weight change test (provided in Appendix B).
The test involved weighing a clean, unpainted F/E prepreg test panel before and after immersion in
the solvent. A weight loss occurs if the solvent has eaten away the substrate. A weight gain could also
occur If the solvent is absorbed into the substructure through microcracks in the substrate. Weight change
measurements were recorded in grams. The test panel remained in the solvent for four hours, four times
longer than the time required to strip the panel in the screening test. After immersion, the panels were
gently wiped dry by hand, and dried in a 150°F oven for one hour. Four weight change tests were
performed for each solvent. The balance used in this test was calibrated with Bureau of Standards Class
S weights each day of testing. Before each measurement was made, the balance was tared to 0.000 grams.
Panels from the prepreg sheet had to be used for this test because the condemned radome panels
may have given false readings. TAFB materials engineers felt that the PC Blend 2 may have absorbed into
the honeycomb layers and give a false weight gain. Due to the conical shape of the radome, the
honeycomb structure is not in contact with the solvent during the actual depainting operation. Therefore,
it was agreed by TAFB and FWEI that panels cut from a prepreg sheet would be more representative of
actual conditions. :
A parallel experiment was also conducted on a standard to validate the drying procedure. The
standard panels are 2V6 in. square 316 stainless steel (SS) plates, and were subjected to the same testing
conditions as the composite. An inert material, the 316 SS should not absorb or be attacked by the solvent.
Any gain in weight indicates solvent residue is present on the surface and did not evaporate.
14.
-------
SECTION 6
RESULTS AND DISCUSSION
SCREENING PROCEDURE
FWEI screened the three propylene carbonate blends and MEK at TAFB Chemical Laboratory in
Building 3001 from August 9 through 13, 1993. Four 24-hour runs were performed with each solvent.
In addition, a fifth beaker containing a mixture of 12.5 percent by weight water in MEK was evaluated.
This was to determine the potential reuse of an MEK/water mixture which would be recovered from an
adsorptive resin system, such as the Purus PADRE™ unit, being evaluated simultaneously at TAFB by FWEI
for MEK vapor recovery.
Results of the screening tests are presented in Table 3. As the table indicates, the MEK/water mixture
removed the paint in approximately 30 minutes. Although the three PC blends removed the paint, PC Blend
2 removed ft most rapidly. For each solvent, paint was removed by hand after the paint had completely
bubbled from the surface. The paint bubbled in one piece, which included primer layer, rain-erosion coating,
and topcoat. Both MEK and the PC blends attacked the paint in this manner, i.e., by dissolving the bond
between the coating and the substrate.
A possible explanation provided by Texaco for why PC Blend 2 was effective at paint removal is due
to solubility parameters. This theory is that two substances should be soluble with each other when the
solubility parameter of the solvent is equal or nearly equal to the solubility parameter of the solute (in this
case, paint) although for this study, the paint formulation was not defined. This theory also explains why
the individual components of a blend may not be effective paint removers by themselves, but when
combined with other solvents, increase their effectiveness significantly.
The MEK did not perform as well as the MEK/water mixture. After the first run, significant evaporation
occurred in both the MEK and MEK/water beakers. To minimize the evaporation, a plastic bag was placed
over both beakers after about eight hours into the second run. On the third and fourth runs, a watchglass
was placed over the beakers during the entire 24-hour period.
As Table 6-1 indicates, MEK performance became worse in Runs 3 and 4 when the beaker was
covered. Given the performance of the MEK/water and the MEK in the first two runs, it appears that the
MEK absorbs water from the air, which enhances its paint removal properties. The behavior of the
MEK/water mixture versus pure MEK is similar to that of a solvent blend versus a pure solvent, as discussed
above. Also, since the volatile MEK is evaporating, the composition of the MEK/water mixture changes with
decreasing concentrations of MEK as time increases. Absorption of water by MEK can be checked by doing
moisture analyses on several MEK samples obtained from the drainage sump during TAFB departing
operations. The TAFB operators in the radome departing operation also indicated that MEK sometimes
must be used in conjunction with manual scraping techniques to completely remove the paint. The removal
time also varies from about 1V6 hours to several hours.
15
-------
III
DC
UI
DC
Q
UJ
8
DC
a.
DC
O
CO
CO
Ul
m
i
O
DC
S
o
£
CO
s
z
*
•I
CM
i
1 30 Minutes
o
f=
at o>
1 5
.0 J3
S S
CD CD
t 1
Q Q
CO
2
f-§ CD
m fS
co "c3.cn
B 1 0
S O CM
«?
5
50% Bubbling
No Change
30% Bubbling
No Change
*~ T- CM CO XT
"=§§§§
s. & cc ir cc
CQ
o
0.
O)
3
1
Q.
i
"
0)0)0)
Complete BubbI
Complete BubbI
Complete BubbI
10% Bubbling
25% Bubbling
5% Bubbling
90% Bubbling
No Change
011 T- CM CO f
"2 c n c c
IddcSd
CO
o
Q.
CD
m
S
CD
3.
E
3
CO CO
c c
3 3
•§ g"§
s Is
CD 3 CO
Q. CD Q.
s a? e
3 S<3
CO
3
if if
CQ 3,jC CQ
g | o §
80 2 i-
co
CQ 3 ca co
s^SS
CO (8 2 2
No Change
Slight Bubbling
No Change
No Change
m i- CM CO Tf
"9 c c c c
IdSdDl
CQ
0
Q.
iS
£> X)
xi ja
3 3
CO CQ
a? a?
CO O)
c c
3 3
•§ ~9 o> o>
o)
O CO CO CO
' S S S 5
-------
The inconsistency of the paint removal times for the test panels could possibly be explained by the
condition of the panels prior to depainting. Some panels had a mottled appearance, suggesting certain
areas of the topcoat layer were worn, exposing the white,; rain-erosion coating layer. For these panels, paint
removal would be easier since the paint thickness was thinner than when originally applied. In Run 4, panels
having similar surface conditions were selected. These panels did not have a mottled appearance and,
therefore, provided better comparison of the different solvents. As reported in Table 3, complete bubbling
took longer in Run 4 than in the previous runs. To satisfy requirements in the QAPP for this project, three
additional runs should have been made to confirm results of Run 4. However, because of schedule and
limited supply of condemned radome panels with similar surface conditions, this was not possible. The
varied surface conditions of the radomes are, however, representative of the radomes encountered in the
actual depainting operation.
After the 24-hour period, each of the test panels were evaluated for visible substrate damage. No
apparent substrate damage was visible to the naked eye, although white spots indicated some fibers may
be exposed. The TAFB materials engineer examined the panels visually and under magnification. He
concluded that there could be possible fibers exposed, although all panels, including the MEK, exhibited this
characteristic.
TAFB and FWEI conducted a separate experiment in which panels were partially immersed in the
solvents (MEK/H2O and PC Blend 2 were used since they were removing paint the fastest). When the paint
bubbled and peeled away from the panel's surface, the paint was manually removed. After the panels had
dried, the paint on the unimmersed portion was partially peeled away by fingernail. This exposed portion
of the panel had a darker color of substrate and TAFB concluded it indicated the top layer of resin (referred
to as the "gel coat") had been possibly removed by both PC Blend 2 and the MEK/H2O. To validate this
observation, this experiment should be repeated with pure MEK for comparison and panels should be further
evaluated at magnification.
Based on screening results, PC Blend 2 was selected for further testing to evaluate the effect of the
solvent on the F/E substrate. MEK was also evaluated in these tests.
EVALUATION TESTS
FWDC performed evaluation tests at their John Blizard Research Center in Livingston, New Jersey.
Evaluation tests consisted of simulated depainting, hardness, flexural properties, paint adhesion, and weight
change. ,
Simulated Depainting Procedure
Results from the depainting simulation are given in Table 4, and includes the time of removal for each
test panel and an observation on the degree of paint removal. The PC Blend 2 removed the paint faster and
more completely than MEK. It was discovered during this exercise that the paint system on these panels
was different from the one used in the screening test. TAFB confirmed that these panels were painted with
a neoprene-based coating which consisted of a yellow primer, white primer, and black topcoat. This paint
system bonded very well to the substrate, and was extremely difficult to remove. TAFB discontinued using
it over six years ago. Apparently, the condemned radome from which these panels were cut still had this
paint system on it.
The current paint system consists of a dark red primer, followed by a white, rain-erosion coating and
black topcoat. This paint system was on the radome test panels used in the screening tests. Both paint
systems appear identical on the test panels. Only when the panels are stripped can they be identified.
17
-------
TABLE 4. DEPAINTING SIMULATION RESULTS
Sample No.
DP-01
DP-02
DP-03
DP-04
DM-01
DM-02
DM-03
DM-04
Solvent
PC Blend 2
PC Blend 2
PC Blend 2
PC Blend 2
MEK '
MEK
MEK
MEK
Removal Time
(min)
228
212
44
58
248
308
307
154
Removal Efficiency
100% removal
i
100% removal
100% removal
100% removal
100% removal
100% removal
100% removal
100% removal
Runs 3 and 4 for the depainting simulation with PC Blend 2 had the current paint system and,
therefore, removed paint in less than one hour and in one piece, as in the screening. In Runs 1 and 2, the
PC Blend 2 removed the old paint in about 31/£ hours with vigorous scraping. PC Blend 2 appeared to
remove the neoprene paint system more completely than the MEK. The four panels tested with MEK,
however, had the neoprene system. MEK took an average of 4V6 hours to remove this paint, with vigorous
scraping.
Due to schedule considerations, this test could not be repeated with panels coated with the current
system. However, the panels for the paint adhesion test had the current paint system. An additional step
of recording the paint removal time was added to,that test.
Hardness
Hardness results are summarized in Table 5. Hardness measurements were made on unpainted F/E
prepreg test panels. Ten measurements were made on each panel prior to exposure of solvent. The panels
were then sprayed with solvent in the depainting simulation unit for two hours. After drying, ten additional
hardness measurements were made on each panel. Table 5 lists the measurements and the arithmetic
average.
The hardness test objective was to achieve a Barcol hardness of 55 or greater. As indicated in Table
5, hardness measurements met this objective and also did not change significantly after being exposed to
solvent Measurements ranged from 75 to 85 Barcol hardness units, with an overall average of 80.4. These
results show that both MEK and PC Blend 2 do not chemically embrittle the substrate.
Flexural Properties Test
Flexural properties test results are listed in Table 6. For each test panel, measurements of length,
width, depth, and rate were made. Flexural strength, maximum strain, and modulus of elasticity were
calculated by the following formulas: ;
18
-------
TABLE 5. HARDNESS TEST RESULTS (Per ASTM D2583-87)
Sample No.
HP-01
HP-02
HP-03
HP-04
Solvent
PC Blend 2
PC Blend 2
PC Blend 2
PC Blend 2 ,
1
Measurement
1
2
3
4
5
6
7
8
9
10
Average
1
2
3
4
5
6
7
8
9
10
Average
1
2
3
4
5
6
7
8
9
10
Average
1
2
3
4
5
6
7
8
9
10
Average
Total average
Barcol hardness
Before
81
80
80
82
79
82
79
79
80
78
80
79
82
81
82
78
79
81
77
80
78
79.7
81
79
79
80
82
82
82
79
79
81
80.4
80
82
80
80
79
81
78
82
77
79
79.8
80 ± 0.3
After
82
84
81
80
; so
83
84
84
84
84
82.6
80
79
80
80
79
79
80
81
80
79
79.7
81
81
81
82
81
80
81
82
79
79
80.7
79
82
80
76
80
77
'. 80
80
80
80
79.4
80.6 ±1.4
19
-------
TABLE 5. (Continued)
Sample No.
HM-01
.
HM-02
HM-O3
HM-04 ,
Solvent
MEK
MEK
MEK
MEK
i
Measurement
1
2
3
4
5
6
7
8
9
10
Average
1
2
3
4
5
6
7
8
9
10
Average
.1
2
3
4
5
6
7
8
9
10
Average
1
2
3
4
5
6
7
8
9
10
Average
Total average
Barcol hardness
Before
81
82
82
81.5
81
81
82
82
81.5
80
81.4
79.5
81
79
78
81
80
81
79
78.5
79
79.6
79
80
80
80
80
83
78
82
79
79
80
76
80
79
80
78
79
79
82
76
80
78.9
80 ± 1.1
After
i 79
82
82
79
81
82
81
83
83
82
81.4
85
82
82
81
, 80
83
82
80
81
82
81.8
82
80
79
82
82
82
82
80
77
, 79
' 80.5
: 78
79
81
80
81
80
81
75
82
79
79.6
80.8 ± 1.0
20
-------
O)
i
UJ
cc
(0
UJ
HI
a.
O
cc
a.
u.
id
UJ
CQ
I-
-of
If
*!
K
1?
IE""
s-4
ifc
01
,—k
It
.'L
IT
S*
}l
1
Solvent
1
R
"o "o "o "o "o "o
X X X X X X
S § ° 8 § to
co co co co co co
CM CM CM CM CM J5
O o O O O O
^? ^j ^3 * ^D CNJ
C 2 S M- C S!~
1- CM T- T- CM i-
OOdddd
O O O O O O
U) 1** U5 O CO ^J
p p O p P p
CM 01 CM fl> O 4 CM CM CM
\
CM CM CM CM , CM
•O T3 "O 'D T3
c c c c c
Q> Q> QJ
^
co
CM"
csj csi oj csi csj
^ ^: ^ ^ ^
111 LLJ UJ UJ UJ
5 5 5 5 5
i i i i i
LL. U. U. U. U.
t
X
•
co
1
o
CD
O>
^
CD
•cb S
'5-
«? d
co ^
m o>
CM 5
0 O
o o
+i
2 10
~s
« .+'
CM 8
S *
&«
< CQ
m C C
S.2 °
111
££
co
c
o
T3
O
o
T3
1
I
a.
S
2
ca
a.
co
CD
I
TJ
i
I
I
CM
-------
Rexural strength:
S = 3PL
2bd2
where: S » stress in the outer fibers at midspan (psi)
P - load at a given point on the load-deflection curve (lbf)
L = support span (in.)
b = width of beam tested (in.)
d = depth of beam tested (in.)
Maximum strain:
r« 6Dd
L2
where: r = maximum strain in the outer fibers (in./in.)
D = maximum deflection of the center of the beam (in.)
L = support span (in.)
d - depth of beam tested (in.)
Modulus of elasticity:
EB= L*m_
4bd3
where: EB = modulus of elasticity in bending (psi)
L - support span (in.)
m = tangent slope to the initial straight-line portion of the load deflection curve (lbf/in.)
b = width of beam tested (in.)
d = depth of beam tested (in.)
Values required for the calculations were measured by the test.
Load deflection curves plotting load (Ibs) versus deflection (mils) were produced for each test panel.
These curves are presented in Appendix C as Figures 2 through 17.
The data presented in Table 6 demonstrates the exposure to either PC Blend 2 or MEK did not affect
the flexural strength of the panels. Although most panels failed at approximately 72,000 psi loading, test
panels O-4, FP-1, FP-4, and FM-2 had Idwer flexural strength, failing at approximately 52,000 psi loading.
These four panels failed with a straight break across the test panel rather than with the zigzag pattern
exhibited with the stronger panels. Since this occurred for the unexposed, PC Blend 2 exposed, and MEK
exposed panels, the probable explanation is that those four test panels were cut from a weaker section of
the prepreg sheet A statistical analysis revealed that a bimodal distribution existed, proof that the four
panels were taken from a different sample population than the stronger specimens. Comparison of
individual readings within the respective populations show that flexural strength is unaffected by exposure
to either solvent.
The second part of the test required observation of the failed test panels under an SEM. The panels
failed in the first part of the test, but did not break in half. The specimens were manually broken in half so
22
-------
they could be viewed both at the surface and cross-section of the failure by the SEM. After proper mounting
and preparation, the panels were viewed at various magnifications to determine the conditions of the fibers
and matrix. Microphotographs were taken at 30X, 300X, and 1200X magnification for representative test
panels for each solvent. Selected microphotdgraphs are presented as Figures 18 through 22 in Appendix
C. .'. . .
Figures 18 through 20 represent microphotographs made of the panels not exposed to solvent (i.e.
Samples O-3 and O-4). Sample O-4 was examined because of its low flexural strength and its atypical
failure pattern. As a comparison of Figures 18 and 20 indicates, Sample O-4 at 30X magnification has a
much straighter and cleaner break than that of Sample O-3 at the same magnification. This type of break
also occurred on Samples FP-01, FP-04, and FM-02. However, the microphotographs show no evidence
of fiber damage on Sample O-4.
The remaining microphotographs were made for panels exposed to solvent. Figures 21 represents
Sample FP-05, exposed to PC Blend 2. Figure 22 represents Sample FM-Q5, exposed to MEK. No
difference in appearance was evident when comparing the two solvents' microphotographs at the same
magnifications. At 300X magnification, Sample FM-05 appears to have fibers protruding in every direction,
whereas Sample FP-05 has unidirectional fibers. However, this is a function of the area selected for
photographing. As can be seen in the top portion of Figure 21, Sample FP-05 also has fibers in all
directions.
The SEM microphotographs indicate no damage by either solvent of the fiber matrix interface or of
the fillers themselves. Had damage occurred, the microphotographs would have shown noticeable gaps
where the fibers interface the matrix (most obvious at the 300X magnification). Also, individual fibers appear
to be intact, indicating the solvent did not attack the resin binding the fibers.
Paint Adhesion
Table 7 reports the paint adhesion results and also lists the paint removal time for preparation of each
test panel.
TABLE 7. PAINT ADHESION TEST RESULTS (Per ASTM D3359-92a)
Sample No.
PC-01
PC-02
PC-03
PC-04
PM-01
PM-02
PM-03
PM-04
Solvent
PC Blend 2
PC Blend 2
PC Blend 2
PC Blend 2
MEK
MEK
MEK
MEK
Removal Time
45 min.
21 min.
25 min.
25 min.
30 min.
29 min.
25 min.
25 min.
Paint Adhesion Rating
5A
5A
5A
5A
5A
5A
5A
5A
Removal times for PC Blend 2 averaged about 30 minutes. The MEK samples had paint bubbling
much quicker than the PC panels (after about 10 to 15 minutes). After bubbling, paint on the MEK samples
23
-------
removed easily. PM-2 and PM-4 required scraping to remove paint on edges whjch were not in the direct
spray pattern of the MEK. The PC Blend 2 panels required minimal scraping to remove 100 percent of the
paint. Sample PC-01's longer removal time is attributed to less vigorous scraping than the other panels.
The behavior of the paint on PC Blend 2 panels was to bubble after about 20 minutes and peel off live to
ten minutes later. Scraping was required to remove the loosened paint from the substrate.
Paint removal efficiency of panels for this test is similar to that of Runs 1 and 2 in the screening
exercise. (In Runs 1 and 2, the beakers were left uncovered; in Runs 3 and 4, the beakers were covered
to prevent evaporation and removal efficiency decreased significantly.) One possible explanation is that the
MEK draws moisture from the air to form an MEK/water solvent blend. Water is a polar solvent which
increases the polarity of the solvent blend and, therefore, enhances the paint removal capability of the MEK.
The panels were observed after air drying for one hour. The MEK panels appeared darker than the
PC panels.
The paint adhesion ratings determined by the paint adhesion test were 5A for each test panel.. This
rating indicates no peeling or removal of paint by the pressure-sensitive tape occurred, suggesting complete
paint adhesion following a depainting/painting cyde.
Weight Change Test
Results of the weight change test are shown in Table 8. Weight changes are given in grams and as
a percentage of the total weight. The test panels exposed to both PC Blend 2 and MEK have weiglrt loss
indicating slight substrate damage although considered to be negligible by FWEI and TAFB. Weight
measurements for the 316 SS standards are also given in Table 8. Weight changes for these samples are
within the accuracy limits of the balance. Therefore, the solvents had completely evaporated from the
surface of the test panel.
24
-------
TABLE 8. WEIGHT CHANGE TEST RESULTS
Sample No.
WP-01
WP-02
WP-Q3
WP-04
WM-01
WM-02
WM-03
WM-04
Standards
WP-01
WP-02
WP-03
WP-04
WM-01
WM-02
WM-03
WM-04
Solvent
PC Blend 2
PC Blend 2
PC Blend 2
PC Blend 2
MEK
MEK
MEK
MEK
Weight (grams)
Before
8.124
8.290
8.177
8.065
8.435
8.337
8.300
8.596
After
8.113
8.281
8.157
8.040
Average
8.427
8.331
8.287
8.587
Average
Weight Change
Grams
-0.011
-0.009
-0.020
-0.025
-0.016
-0.008
-0.006
-0.013
-0.009
-0.009
%
0.14
0.11
0.25
0.31
0.20
0.09
0.07
0.16
0.10
0.11
PC Blend 2
PC Blend 2
.PC Blend 2
PC Blend 2
MEK
MEK
MEK
MEK
118.612
120.937
120.118
120.710
119.581
119.464
118.044
120.700
118.610
120.935
120.119
120.710
119.581
119.465
118.044
120.700
-0.002
-0.002 ,
+0.001
0.000
0.000
+0.001
0.000
0.000
25
-------
SECTION 7
COST ANALYSIS
An economic analysis was performed for replacing MEK with PC Blend 2, based on available
information. Results of this analysis are summarized in Table 9. Assumptions used in the analysis are
discussed in the following paragraphs.
TABLE 9. ECONOMIC COMPARISON OF PC BLEND 2 AND MEK
Cost raw material ($/gal)
Usage (gal)
Total raw material co«t ($)
Disposal cost ($/lb)
Quantity disposed (Ibs)
Total disposal cost ($)
Labor rate ($/hr)
Ami of labor required (hr)
Total labor cost
TOTAL
PC Blend 2
$9.01 1
3.0253
$27,255.25
SLsa2
3.1202
$5,990.40
$58.35
312
$18,20520
$51,450.85
MEK
189 drums @ $284.13/drum2
31 drums @ $215.29/drum2
12,100*
$60,374.56*
$1.92*
3.1202
$5,990.40
$58.35
260
$15,171.00
$81,535.96 I
Notes: i
1 Texaco estimate
2 1992 data as supplied by tAFB ,„„„,- -i«-
3 Estimated usage of PC Blend 2 is assumed to be
-------
increased from 150 drums in 1991 to 220 in 1992. However, TAFB agreed to use 1992 figures since it is the
last available year.
The estimated usage of PC Blend 2 is assumed to be V4 the usage of MEK; this is a conservative
estimate since PC Blend 2's vapor pressure is 1/10 that of MEK. The quantity of paint chips disposed is
assumed to be the same for both PC Blend 2 and MEK. Based on discussion in Section 3, spent PC Blend
2 is assumed to be either discharged to an industrial wastewater treatment facility or recycled. If this is not
feasible, an additional disposal cost for the spent solvent would be added to the total disposal cost for PC
Blend 2. The MEK is assumed to evaporate 100 percent, therefore, the spent MEK quantity is zero.
TAFB provided the standard labor rate for the depainting area and is a 1993 value. The amount of
labor was calculated from an average of the results obtained during the testing program and from the actual
depainting operation. In preparing for the paint adhesion test, in which test panels painted with TAFB's
current paint system were depainted in the simulation unit, both PC Blend 2 and MEK took about 30 minutes
to remove paint from a two in. square panel. According to TAFB personnel, paint starts bubbling at 30
minutes and takes between 11/£ to 3 hours to completely depaint the radome. Since both MEK and PC
Blend 2 removed paint at the same rate, the time estimated for paint removal of a complete radome with
both PC Blend 2 and MEK was three hours. An additional two hours were added to PC Blend 2 and MEK
to account for set-up and clean-up time, as well as necessary scraping and disposal. An additional hour
was added to the PC Blend 2 to represent a step to remove solvent residue. This operation is assumed to
occur once a week for an annual total of 312 hours for PC Blend 2 and 260 hours for MEK.
27
-------
SECTION 8
QUALITY ASSURANCE ANALYSIS
The QAPP, "Evaluation of Propylene Carbonate in ALC Depainting Operations," dated August 12,1993
was used as a basis for this study.
The purpose of the QA/QC program was to fulfill two related purposes:
• To provide an organized framework for the laboratory testing program, and
• To control data quality, within pre-established limits, to ensure that it was adequate to achieve the
project objectives.
The overall QA objective is to evaluate the feasibility of substituting a PC solvent formulation for MEK
in depaintlng aircraft radomes. To support this objective, the following goals had to be achieved:
• The PC solvent blend removes paint as effectively as MEK in 24 hours or less,
• Substitution of the PC solvent blend reduces hazardous waste disposal and eliminates discharge
of a hazardous air pollutant, and
• The PC solvent blend does not damage the substrate surface or interface and allows subsequent
application of paint to adhere to the substrate.
DATA QUALITY
The quality control program defined in the QAPP was implemented during the testing program. Some
measurements were identified as critical for meeting the primary project objectives. Other noncritical
measurements were performed to provide supporting information for the primary objectives. Critical
measurements were screening, removal rate and efficiency, paint adhesion, and flexural properties.
Noncritical measurements were weight change and hardness.
The QAPP contains quantitative specifications for analytical data quality. Adequate data quality is
defined by meeting criteria for precision, accuracy, and completeness. A comparison of the data quality
requirements, as defined in the QAPP and what was actually achieved, are provided in Table 10. A summary
of calibration standards used for this test program is presented in Table 1 of Appendix C.
Based upon the test results, the data quality was adequate to meet the program objectives. Accuracy
objectives were met for all critical tests. Although data quality criteria for precision and completeness were
not achieved for some of the tests, supplemental tests were added to address those testing areas which
28
-------
i
V)
I
LU
2
LU
gc
O
LU
DC
£j
i
o
\-
0
6
LU
_J
m
u
t)
ll
1
!=
Q
;3
&
(0
e
o
2
1
i
^
Q.
a.
-a
i
Q.
"3
Q.
a
2
Is
o
2 a
5 i
E
«5
in
O)
88
co co-
in
+i
f
+i +i
+i
o o
«.
1
£
a*
>.§
|1|
°ll
IS.!
eS£
S
§11
^*~
in u> tn
o> o> o
1 1 1
~-z?~
coSS
TT T-: co
SSS
Q.
*f
i, in./in.
ral strength
ilus of elasti
^ J "^
Siri
'€
Q.
1
3
X
0>
C
8
8
sr
o
S
i
••!
\E
5
C
1
T2
1C
1
'5
£
Si .52 «
§ »
O *
en
CM
-------
failed to meet the QA objectives. When the test data are considered in their totality, the project met overall
program objectives. These factors are discussed below.
Precision
The precision criteria are expressed as the percent relative standard deviation (RSD) of the replicate
values from each test. The data met the precision criteria for the removal efficiency, flexural properties, and
paint adhesion data. It did not meet the precision criteria for paint removal rate (time).
Paint Removal Rate-
The paint removal rate was to be originally determined from the depainting simulation tests. Since the
specimens provided had a different paint system, most of the data from the depainting simulation tests were
invalid. Time considerations did not allow rerun of the depainting simulation tests so corrective action
consisted of obtaining paint removal times from the paint adhesion tests. This data gave a 11.6 percent RSD
which slightly failed the precision criteria. This failure should not affect the main conclusion that paint
removal times between the base MEK solvent and the proposed PC Blend 2 were comparable for the
following reasons:
• The criterion for this test was set too tightly.
- With the exception of sample PC-01 that was scraped too lightly, both the MEK and PC Blend 2
data had the same variability.
- The main purpose for paint removal time data was to show that paint removal times were
comparable. Average paint removal times were comparable.
- The paint removal times were obtained from actual specimens that had been exposed to erosive
forces that flying as an external part of an aircraft creates. The specimens showed variations in
wear. The paint should be easy to remove from specimens having greater wear.
- There was no previous data to set a reasonable precision criterion for the paint removal times.
• Since the material tested does not provide true replicates, the precision criterion should have been
based upon comparing the standard deviations achieved in the base MEK tests and the PC Blend
2 tests.
Rexural Strength-
The flexural strength tests met the precision criteria. The samples tested failed around two different
strengths, 71,762 psi and 52,400 psi. Within each population the RSD for extreme values were within the
10 percent precision criteria. Data concentrated around two sample means indicate a bimodaj distribution.
This was confirmed by a statistical analysis that showed that the specimens were taken from two distinct
sample populations within 95 percent confidence limits (See Table 6).
Accuracy
Accuracy for removal time was defined by calibration of the stopwatch against the WWV broadcast
of the Universal Coordinated Time (UCT) over a 63-hour time period. The stopwatch was found to tie two
seconds slower than UCT after the 63 hours. Removal efficiency is a visual determination and, therefore,
usual accuracy definitions do not apply. However, accuracy can be estimated, since the human eye can
discern ±20 percent. The .accuracy for the flexural properties measurements is defined by ASTM Test
Method D 790-92. Accuracy objectives were achieved for these measurements. Although accuracy
30
-------
requirements for the paint adhesion test are given in the QAPP as 1 classification number, further review of
ASTM Test Method D-3359-92a indicates bias cannot be established.
Completeness
Completeness is the percentage of the total measurements judged to be valid. Completeness
objectives were not achieved for removal rate (both depainting simulation and paint adhesion). Removal
rate (for depainting simulation) values were not complete because six of the eight test panels were painted
with an obsolete paint system. For the paint adhesion removal rate, a completeness value of 87.5 percent
represents seven out of the eight panels as valid.
DEVIATIONS
The following deviations from the procedures and test plan defined in the QAPP occurred:
• The amount of panels tested was less than specified in the QAPP, which was a combination of two
substrates and two paint systems for each test. TAFB could provide only one substrate with a
single paint system per test. Test panels provided were either F/E with honeycomb or unpainted
F/E prepreg. As discussed in Section 5, hardness, flexural properties, and weight change tests
were not conducive to panels with a honeycomb structure and, therefore, used the unpainted F/E
prepreg. Since the base material is the same (F/E), FWEl would not expect a difference in results
obtained with either form.
• An additional solvent blend of MEK with 12.5 percent concentration by weight water, was analyzed
during the screening test. The purpose of this additional test was to evaluate the potential to reuse
MEK with water which could be recovered in the Purus PADRE™ unit, which FWEl is studying
simultaneously at TAFB.
• During screening, the test panels were immersed face down into the solvent baths rather than
suspended using wires clamped to a ringstand. It was discovered the panels floated and it was
more practical to immerse the specimens.
• Panels for the screening and evaluation tests were not scribed, as discussed in the QAPP. TAFB
explained radomes are not always scribed, and they wanted the test panels to represent actual
radome conditions.
• An additonal test was performed to better evaluate possible resin damage after screening. Two
panels were partially immersed in beakers: one containing MEK/H2O and the other PC Blend 2.
After the paint bubbled, it was removed manually. After sufficiently drying, the unimmersed area was
then peeled from the immersed area with a fingernail, exposing the substrate under the unimmersed
area. This was then compared to the immersed area for any visual signs of resin removal.
• After the departing simulation runs were conducted, it was discovered some of the panels were
painted with a neoprene-based coating which was discontinued over six years ago due to its
difficulty in paint removal. The surface of these test panels were identical to those with the current
paint system. Out of the eight panels tested, six were painted with this system. Because of this,
the times recorded for the depainting simulation are not representative since they do not reflect the
current radome conditions. Due to limited sample size (specimens containing the same integrity
as the specimens used in the screening were not readily available at the testing site), the test was
not redone; instead, times were recorded for the specimen preparation step for the paint adhesion
test.
31
-------
• For the hardness test, panels were cut from an unpainted F/E prepreg sheet. The panels were
exposed to the solvent in the depainting simulation unit for two hours. This time was chosen
arbitrarily since the depainting simulation did not represent the paint system currently used at TAFB.
However, the removal time for both solvents was later determined to be about 30 minutes; based
on this finding, two hours was a reasonable exposure time.
• The paint adhesion test measured only one "X" cut on the test panel, rather than the three specified
in the QAPP. Each "X" cut, as specified in ASTM D 3359, was 1.5 in. long. Test panels were
provided by TAFB and as specified by FWEI, were two in. square. Therefore, only one "X" cut could
fit on each test panel. Since the results were consistent for both solvents, and four samples were
tested, the extra two "X" cuts would probably give similar results.
• Supporting measurements of temperature and relative humidity were not recorded during tests
conducted early in the test program because a measuring device was not readily available. During
the flexural properties and paint adhesion tests, this information was measured and recorded, as
recommended by the respective ASTM test methods. These measurements are reported with the
test results in Appendix C.
• The weight change test procedure was revised, since issuing the QAPP. The revised procedure is
included in Appendix B as Revision 2. Changes include unpainted F/E prepreg panels used as test
specimens. Since the panels were unpainted and did not contain the honeycomb structures, no
depainting with MEK or masking the test panel was required to prepare for the test. Also, the
panels tested with PC Blend 2 did not undergo a solvent rinse after removal from the simulation unit;
it was decided not to introduce another variable into the testing process. The panels were wiped
gently with a paper towel and dried in an oven for one hour. As the results from the standard
panels indicate, the solvent appeared to sufficiently evaporate and did not affect test results.
• After the flexural properties test, the failed test panels were broken in half at the failure site to allow
examination with the SEM. The panels were then mounted and viewed by the SEM at various
magnifications. Microphotographs were made at 30X, 300X, and 1200X magnification rather than
the 40X and 1000X indicated in the QAPP. The latter magnifications seemed to give a representative
view of the fiber matrix. The SOX magnification provides an overall view of the failure of the panel.
The 300X magnification views the fiber and matrix interface, while the 1200X magnification gives a
close-up of the fiber breakage. '
32
-------
REFERENCES
1. Foster Wheeler Enviresponse, Inc., "Quality Assurance Project Plan - Evaluation of Propylene
Carbonate in ALC Depainting Operations," August 12, 1993.
2. Battelle, "Assessment of Selected Oklahoma City Air Logistics Center Overhaul/Repair Processes for
Reduction of Chemical Wastes," September 30,1992.
3. Pollution Prevention International, Inc.," Use of 1,1,1 -Trichloroethane, Xylene, and Methyl Ethyl Ketone
in the Aerospace Industry," EPA/540/0-00/000, U.S. Environmental Protection Agency, Cincinnati,
Ohib, September 1992.
4. ASTM D3359-92a (Vol. 6.01), Tape Adhesion Test.
5. ASTM D 2583-87 (Vol. 8.02), Barcol Hardness Test.
6. ASTM D 790-92 (Vol. 8.01), Rexural Properties Test.
33
-------
------- |