EPA/600/2-85/107
September 1985
TRANSFER EFFICIENCY OF
IMPROPERLY MAINTAINED OR OPERATED SPRAY
PAINTING EQUIPMENT SENSITIVITY STUDIES
by
K. C. Kennedy
CENT EC Corporation
Reston. Virginia 22090
Contract Number 68-03-1721, Task 1
EPA Project Officer:
Charles H. Darvin
Air and Energy Engineering Research Laboratory
Hazardous Air Technology Branch
Research Triangle Park, NC 27711
AIR AND ENERGY ENGINEERING RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
EPA Headquarters Library
-------
TECHNICAL REPORT DATA
jP'.au nod Imuniciumt on tht rtrrnt btfort compltfint)
NO
EPA/600/2-35/107
I TITLE AND SUBTITLE
Transfer Efficiency of Improperly Maintained or
Operated Spray Painting Equipment Sensitivity
Studies
i. REC
B. REPORT DATE
September 1985
>. PERFORMING ORGANIZATION CODE
T AUTMOR(S)
K.C. Kennedy
I. PERFORMING ORGANIZATION REPORT NO
i PERFORMING ORGANIZATION NAME ANC AOCRESS
Centec Corporation
11260 Roger Bacon Drive
Reston. Virginia 22090-5281
10 PROGRAM ELEMENT NO
11. CONTRACT/GRANT NO.
68-03-1721. Task 1
13 SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OP REPORT AND PERIOD COVERED
Task Final; 6/83 - 6/85
14. SPONSORING AGENCY CODE
EPA/600/13
6 •^""•"TAnviioTM AEERL project officer is Charles H. Darvin. Mail Drop 54. 919/
541-7633.
6. ABSTRACT _. ' ,
The report gives results of an investigation of the impact of common indus-
trial operating and maintenance practices on the efficiency of spray painting sys-
tems. The investigation included independent research, as well as assistance from
both representatives of the spray painting equipment manufacturing industry and
users of spray painting equipment. The results indicate strong directional response
in painting efficiency to certain common painting practices.,'
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
Pollution
Spray Painting
Maintenance
Operations
b IDENTIPIEnS/OPEN ENDED TERMS
Pollution Control
Stationary Sources.
Transfer Efficiency
COS
-------
NOTICE
This document has been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names
or commercial products does not constitute endorse-
ment or recommendation for use.
11
-------
FOREWORD
When energy and material resources are extracted, proc-
essed, converted, and used, the related pollutional impacts on
the environment and even on health often require that new and
increasingly more efficient pollution control methods be used.
The Air and Energy Engineering Research Division (AEERD) at
Research Triangle Park, North Carolina, assists in developing
and demonstrating new and improved methodologies that will
meet these needs both effectively and economically.
The research described herein was undertaken to address
how spray painting transfer efficiency is affected by operating
and maintenance parameters. Air pollution impacts, energy,
and materials resource conservation are affected by loss of paint
and solvenc in poorly operated or maintained spray painting
facilities.
Pour major types of spray painting equipment were tested
to determine their sensitivity to certain preselected operating
or maintenance parameters.
This is the first published research into a very expensive
industrial and environmental problem.
ABSTRACT
This report is submitted in fulfillment of Contract
Number 68-03-1721, Task 1. It describes sensitivity studies
conducted on four types of spray systems to determine the
effects of improper operations or maintenance on trtnsfer
efficiency. A Draft Standard Transfer Efficiency Kethod was
used for the test program. Three different target configura-
tions were painted for each spray system.
Test results show the strong effect proper selection of
spray conditions has on transfer efficiency. The particular
3evel of response for specific factors varies frou spray system
to spray system, and from target configuration to target con-
figuration. Case-specific regressions were developed for each
spray system and target type. These are presented and discussed
in the report.
ill
-------
CONTENTS
Foreword ill
Abstract ill
Figures vii
Tables viii
Abbreviations and Symbols x
Acknowledgment xi
1. Introduction 1
2. Conclusions 3
3. Mass Flow Rate Comparison 4
4. Air Atomized Conventional Spray Equipment 7
Equipment Description 7
Operating and Maintenance Variables 8
Experimental Design 10
AAC Test Performance 13
Test Results ........... 17
Statistical Analysis 17
AAC Test Conclusions 25
5. Air Atomized Electrostatic Spray Equipment 30
Equipment Description 30
Operating and Maintenance Variables 31
AAE Test Performance 38
Test Results 40
Statistical Analysis 40
AAE Conclusions 44
6. Airless Conventional Spray Equipment 50
Equipment Description 50
Operating and Maintenance Variables 50
Experimental Design 52
Test Performance 52
Test Results 56
Statistical Analysis 56
ALC Conclusions 59
7. Airless Electrostatic Spray Equipment 64
Equipment Description 64
Operating zi,d Maintenance Variables 65
Experimental Design .... 65
ALE Test Performance 70
Test Results 73
Statistical Analysis 75
ALE Conclusions 77
-------
CONTENTS (Continued)
3. Comparison of Targets 83
Background 83
Comparison of Variables Identified as Significant 83
Worth Assessment of Three Target Configurations . 85
Appendices
A. Draft Standard Method for Spray Fainting Transfer
Efficiency Operations and Maintenance Testing .... 87
B. Quality Assurance/Quality Control Plan Sensitivity
Studies on the Effects on Transfer Efficiency of
Improperly Maintained or Operated Spray Painting
Equipment 114
C. AAC Test Equipment and Paint Specifications 136
D. AAE Test Equipment and Paint Specifications 139
E. ALC Test Equipment and Paint Specifications 142
F. ALE Test Equipment and Paint Specifications 145
G. Glossary of Statistical Terms 148
vl
-------
FIGURES
Number Page
1 Air atomized conventional air cap
(frontal view) 16
2 Air atomized electrostatic air cap
(frontal view) 33
3 Air atomized electrostatic electrode position
test levels 34
4 Airless paint spraying system 51
5 Airless electrostatic air cap showing
test levels for electrode position 72
A-1 Target configurations; for air atomized
conventional and electrostatic spray guns. . . 90
A-2 Target configuration for high speed bell ... SI
A-3 Set-up for paint supply equipment and platform
scales 97
A-4 Permissible methods for measuring conveyor
speed * 101
A-5 Vertical cylinder wrapping technique 105
A-6 Flat panel foil attachment technique 106
B-1 Project organization as related to corporate
structure 120
B-2 Data validation responsibilities 127
vii
-------
TABLES
Number Page
1 Mass Flow Comparison Data 5
2 Operating and Maintenance Variables for AAC
Spray Equipment 9
3 Experimental Variables Selected for Testing
AAC Spray Systems 10
4 AAC Experimental Design 11
5 Levels of Operating and Maintenance Variables
Tested on AAC Spray Painting Equipment .... 15
6 Air Atomized Conventional Test Results .... 18
7 AAC F-Statisties and Associated Probability . . 21
8 AAC-FP Comparison of Predicted Versus Actual
Transfer Efficiencies 26
9 AAC-VC Comparison of Predicted Versus Actual
Transfer Efficiencies 27
1G AAC-Graco Comparison of Predicted Versus
Actual Transfer Efficiencies 28
11 Operating and Maintenance Variables for AAE
Spray Equipment 32
12 Operating and Maintenance Variables for AAE
Spray Painting Equipment 36
13 AAE Experimental Design 37
14 Levels of Operatinq and Maintenance Variables
Tested on AAE Spray Painting Equipment .... 39
IS AAE Test Results 41
16 AAE F-Statistics and Associated Probibility . . 46
17 AAC-FP Comparison of Predicted Versi s Actual
Transfer Efficiencies 47
i8 AAC-VC Comparison of Predicted Versus Actual
Transfer Efficiencies 48
19 AAC-Graco Comparison of Predicted Versus
Actual Transfer Efficiencies 49
20 Levels of Operating and Maintenance Variables
Tested on ALC Spray Painting Equipment 53
21 ALC experimental Design 54
22 Order of Performance of ALC Test Runs 55
23 ALC Test Results 57
24 ALC F-Statistics and Associated Probabilities . 59
25 ALC-FP Comparison of Predicted Versus
Actual Transfer Efficiencies ... 60
vlll
-------
TABLES (Continued)
Number Page
26 ALC-VC Comparison of Predicted Versus
Actual Transfer Efficiencies 61
27 ALC-Graco Comparison of Predicted Versus
Actual Transfer Efficiencies 62
28 Operating and Maintenance Variables for
ALE Spray Equipment 66
29 Experimental Variables Selected for Testing
ALE Spray Equipment 68
30 ALE Experimental Design 69
31 Levels of Operating and Maintenance Variables
Tested on ALE Spray Painting Equipment .... 71
32 ALE Test Results 74
33 ALE P-Statistics and Associated Probabilities . 78
34 ALE-PP Comparison of Predicted Versus
Actual Transfer Efficiencies 79
35 ALE-VC Comparison of Predicted Versus
Actual Transfer Efficiencies 80
';6 ALE-Graco Comparison of Predicted Versus
Actual Transfer Efficiencies 81
37 Comparison of Significant Factors Identified
by Three Target Configurations 84
38 Worth Assessment Model Comparing Target
Configurations 86
A-1 Nomenclature for Spray Painting Transfer
Efficiency Tests 104
B-1 Spray Painting Transfer Efficiency Precision,
Accuracy and Completeness Objective 122
B-2 Performance Audit Requirements 122
ix
-------
LIST OF ABBREVIATIONS AND UNIT CONVERSIONS
ABBREVIATIONS
ASTM
AAC
AAE
ALC
ALB
EPA
Pan air -
PP
PSIG
O&N
QA/QC
TE
VC
VOC
American Society for Testing and Materials
air atomized conventional paint spray equipment
air atomized electrostatic paint spray equipment
airless conventional spray equipment
airless electrostatic spray equipment
United States Environmental Protection Agency
shaping air or horn air
— flat panel (target configuration)
— pounds per square inch, Ib/in , gauge
- operating and maintenance
- quality assurance/quality control
- transfer efficiency
- vertical cylinder (target configuration)
- volatile organic compounds
UNIT CONVERSIONS
To go from
°C
cm
9
kg
Jcg/L
kPa
L
m
m
m/s
mVs
rps
s
To
°P
in
Ib
Ib
Ib/gal
psig
gal
ft
mils
ft/min (fpm)
ftVmin
rpm
min
Multiply by
,8°C
32
2.54
0.0022
2.204
8.328
0.145 kPa -14.7
0.264
3.281
3.937 x
196.85
211R.8
0.017
60.0
10
-------
ACKNOWLEDGEMENT
The contributions of Graco, Incorporated are gratefully
acknowledged. Graco donated laboratory facilities, test equip-
ment and supplies, and provided technical support for the tests
described herein.
The knowledge and experience of Ray Myers, Professor
of Statistics at Virginia Polytechnic Institute and State
University and author of "Response Surface Methodology" and
"Probability and Statistics for Engineers and Scientists," was
invaluable to this effort. Myers devoloptd the experimen-
tal design and evaluated the test data using the Statistical
Analysis System. Myers1 contributions as a statistical
consultant are grateful)y acknowledged.
xi
-------
SECTION 1
INTRODUCTION
This test program was initiated to develop information
about how spray painting transfer efficiency (TE)* is affected
by operating and maintenance variables. Four basic types of
spray equipment were selected for the test program: air atomized
conventional (AAC), air atomized electrostatic (AAE), airless
conventional (ALC), and airless electrostatic (ALE).
Operating and maintenance (O&M) variables were developed
for each of these equipment types through a literature search,
by industry contacts, and through manufacturers of spray equip-
ment. Over thirty separate variables were identified. Based on
an evaluation of the possible effect of each variable on TE (for
each type of equipment) and on the ability to simulate the
variable in a laboratory, the most significant variables were
selected for testing in this program. Up to 7 variables were
selected for testing on a single equipment type.
An experimental design was developed to address selected
operating and maintenance variables for each type of equipment.
In sach case the design consisted of a fractional factorial
design augmented by a "star" design and a set of replicates.
The process of identification of operating and maintenance
variables, and of developing appropriate experimental designs is
detailed in "Subtask Report: Sensitivity Studies on the Effects
en Transfer Efficiency of Improperly Maintained or Operated
Spray Painting Equipment." Levels for testing each variable
were developed on-site prior to testing.
Once the test program was well defined, CENTEC began
contacting companies with well equipped spray painting labora-
tories to locate a qualified test site. The electrostatics
laboratory at Grace, Incorporated in Minneapolis, Minnesota, was
qualified and willing to participate in the test program. Graco
provided the laboratory, spray equipment, technicians, and some
other materials for testing.
*TE is the amount of paint solids deposited on a target
divided by the amount of paint solids sprayed at the target
multiplied by 100 percent.
-------
The test program was conducted in February 1984. Each
equipment type was the subject of a single experiment consisting
of up to 34 test runs. Each experiment lasted one week, for a
total of 4 weeks of testing. Sections 4, 5, 6, and 7 of this
report describe the performance and results of each experiment.
Section 2 summaries the overall conclusions from the test
program and accompanying test results.
As described in the Draft Standard Test Method (Ap-
pendix A), all tests took place with two target types, flat
panel (FP) and vertical cylinder (VC). Graco has, for their own
purposes, developed a transfer efficiency determination method
utilizing a different target design. In all of the testing
described in the report, the "SPA" targets (Standard Test Method
Targets) were first painted at a given set of conditions,
followed by painting the Graco target set under the same condi-
tions. Thus all transfer efficiency results in this report are
reported according to flat panel, vertical cylinder, and Graco
target results.
-------
SECTION 2
CONCLUSIONS
EFFECTS OF OPERATING AND MAINTENANCE VARIABLES ON TE
AAC transfer efficiency was most strongly affected by
restricted air lines. This effect was pronounced over all
three target types tested, and should be considered the most
prominent OSM variable tested for this type of spray system.
Fan air adjustments had a strong effect for two of three target
types, and should be considered a major effect as well. Re-
stricted paint lines and booth air rates had significant
effects, although not as strong as restricted air lines or
fan air.
AAR tranfer efficiency was effected by the highest
number of variables. The most prominent effect was voltage,
followed by restricted air lines and restricted paint lines.
Booth air, gun cleanliness, fan air, and electrode position
also had significant effects. These effects were not con-
sistent across all target configurations; the VC and Graco
target configurations were much more sensitive to AAE test
variations than FP targets. The FP target restricts the
ability of electrostatic spray to wrap around and increase TE.
ALC transfer efficiency was overwhelmingly affected by tip
erosion. Restricted paint lines were found significant for
Graco targets, but the effect of tip erosion was overriding
in all cases.
ALB transfer efficiency effects are similar to both the
AAE and ALC systems: voltage and electrode position had the
largest effects, but effects of other factors were contingent
on target configuration.
OTHER CONCLUSIONS
The Graco target configuration was found to be the most
sensitive target design for detecting O&M effects. Transfer
efficiency regressions had the tightest fit for Graco results,
and the target configuration was the most comfortable to use
experimentally. Thus, the Graco target represents the most
desirable target design tested to date. It is recommended
as the standard target for all future TE tests.
-------
SECTION 3
MASS FLOW RATE COMPARISON
The original test method called for determining the paint
mass flow rate using platform scales and a stopwatch. (Refer to
Appendix A.) The paint supply pot rested on platform scales,
and readings were taken as t^e paint flow was initiated and
stopped for each test run. & stopwatch was used to time the
interval between scale readings. Mass flow rate was determined
by dividing the total weight difference by the elapsed time.
While this method had proven satisfactory in determining paint
mass flow for TE testing, no airless spray equipment had been
tested. Airless pumps create vibration problems in using
platform scales. Several sources recommended using a mass
flow meter to determine paint flow rate. These sources main-
tained that a mass flow meter would be simpler to use, easier to
read, and more precise than the platform scale/stopwatch
method. \ mass flow comparison test was designed to evaluate
the benefits and any CA/QC implications of these two measurement
techniques.
The QA/QC plan for the TE test program specified require-
ments for determining paint mass flow in terms of weight and
time measurements. To ensure that the mass flow meter met
these requirements, a test was set up to directly compare
methods. The platform scale was set up, calibrated, and zeroed.
The paint supply pot was placed on the scales. The paint flow
was routed through the mass flow meter, which was also zeroed
and calibrated. TE test runs were simulated by spraying atom-
ized paint into an empty spray booth using AAC spray equipment.
A test run time of 16 seconds was selected, similar to the time
for earlier runs at TE testing sites elsewhere.
Seven test runs were performed. In each run mass flow
meter and platform scale/stopwatch readings were taken simul-
taneously. The experimental results are presented in Table 1.
It is readily apparent from Table 1 that the standard deviation
of the mass flow meter data was significantly lower than for the
original mass flow determination method, while the average mass
flow rates were virtually identical.
Table 1 presents the results of seven experiments
performed to compare flow rates as determined by the
the mass flow meter and by the platform scale/stopwatch
-------
method. The use of the platform scale and stopwatch is
described in the Draft Standard Test Method and has been
used in all testing to date. The use of the mass flow meter
would offer certain simplifications in the proposed test
procedure. The experiments listed in Table 1 were undertaken,
then, to determine if the two flow rate measurement methods gave
substantially equivalent results in order to justify the sub-
sequent use of the mass flow meter.
TABLE 1. MASS FLOW COMPARISON DATA
Experiment Platform scale 6 stopwatch Mass flow
number method, g/s meter, g/s
1 10.16 9.99
2 10.05 9.99
3 10.09 9.99
4 9.96 9.99
5 10.04 9.99
6 9.74 9.91
7 9.93 9.93
Mean 10.00 9.97
Standard Deviation 0.14 0.03
It was determined that the new method would be accepted
if it provided readings within the accuracy specifications
set for the flow rate determination, +2 percent. The standard
deviation of the flow rate determination by scale and
stopwatch had been estimated to be 0.1 g/s. At a flow rate
of 10 g/s, then, a maximum acceptable difference of 0.2 g/s
was set, or a ratio of acceptable difference to standard
deviation of 2.0.
-------
The risk of falsely accepting the mass flow meter as
meeting these criteria, the g-risk, was set at 0.05. That
is, no more than a 5 percent risk was desired that the sample
would be judged to have come from an acceptable population
when it really came from an unacceptable r/opulation. The
ct-risk, or the risk that the two methods night be judged dif-
ferent when they actually are equivalent, was set at 0.1
(double-sided test).
The required number of observations to control the o and
0-risks to these levels under the stated conditions is seven.
The significance of the difference between the means was then
determined by performing a t-test at the 0.1 level.
The t-statistic as determined for the data of Table 1 is
0.458 with 12 degrees of freedom. Since the value of t is well
below the critical value at the 0.1 level, it may be stated that
the two methods do not differ by more than 0.2 g/s at the stated
levels of risk.
-------
SECTION 4
AIR ATOMIZED CONVENTIONAL SPRAY EQUIPMENT
EQUIPMENT DESCRIPTION*
A Graco Model 800 manual air spray gun was selected for AAC
testing. The Model 800 gun was considered typical of pressure-
fed external mix air spray equipment. The spray gun was equipped
with a 021-10^806 Graco air cap, and a 1.2 nun fluid tip. Paint
flow was manually initiated by opening a valve on the paint
supply line. Paint flow was measured by a daily-calibrated
Micromotion mass flow meter (see Section 3).
A standard Graco black enamel was selected as the test
paint. The paint averaged about 28 weight percent solids
during AAC tests and was adjusted to 29 seconds {12 Shell
cup) viscosity. A 16 L (4 gal) batch of paint was mixed for
AAC testing. One batch was sufficient to complete AAC testing.
Paint was mixed and stored in a 20 L (5 gal) Graco Model 210-393
pressure tank, which was kept in a temperature-controlled
booth. The paint pressure tank, stirrer, regulators, viscosity
measurement equipment, and some supply lines were kept in the
temperature-controlled booth at 25 C + 1 *C throughout TE
testing. All paint supply lines (spanning 6-8m) were Insulated.
AAC tests were conducted in a Dynaprecipitator water wash
spray booth. Air flow was in the direction of paint spray,
normal to the targets. Air flow was adjusted by opening or
closing a vent on the booth exhaust duct. With the vent closed,
the booth air rate was 61 cm/a (120 fpm) at the plane of the
targets, and varied from 51 cm/s to 71 cm/s (100 fpm to 140 fpm)
across the booth face during testing. With the vent open, the
booth fan pulled air directly from the room and through the
booth (instead cf. taking suction from the booth alone), lowering
the booth aiv flow rate to 36 cm/s (70 fpm) and varied from
25 cm/s to 46 cm/s (50 fpr= to 90 fpm) across the booth face
during testing. Targets were kept from being blown back from
the spray equipment by a polymer pipe frame mounted between the
targets and the water wash.
*Air atomized conventional (AAC) spray equipment is charac-
terized by the use of air as the atomizing agent for the
paint spray.
-------
A variable-speed electric conveyer system (Reliance Electric
Company) was used to carry the targets in front of the spray
equipment. All AAC runs were made with the conveyor set at 10.6
cm/s. Very little fluctuation was observed in conveyor speed
during AAC testing.
Foil weights were determined on Precise laboratory scales
accurate to 0.01 g. Weight-percent-solids samples and dishes
were weighed on 0.0001 g accuracy laboratory scales.
A forced air gas-fired oven was uued for curing weight
percent solids samples and painted foil TE samples. TE samples
were mounted on a large rack for curing, while weight Percent
solids sample dishes were placed on a makeshift shelf for curing.
Both were cured at 148.9*C (300*P) for 20 minutes. After
the first few TE samples showed signs of contamination (dirt
flecks in finish), the oven was cleaned out daily by vibrating
the walls and then vacuuming.
A Micro Motion mass flow meter was used for paint mass flow
determinations. Section 3 discusses the use of a mass flow
meter in comparison to the digital scales/stopwatch mass flow
determination method specified in the test procedure (Appen-
dix A).
Medium temper 4x10~5tn (1.5 mil) thick, 15.24 cm (6 in)
wide aluminum alloy foil was used to cover VC and FP targets.
(Refer to Appendix A, Test Method.) Medium temper 4 x 10 m
(1.5 mil) aluninum alloy foil 38.1 cm (15 in) wide was used to
cover Graco targets during testing.
The test method in Appendix A was strictly adhered to for
AAC testing, as were the QA/QC requirements of Appendix B.
After each EPA test run was completed, a separate run was made
using Graco targets. A summary of all AAC test equipment
specifications is presented in Appendix C.
OPERATING AND MAINTENANCE VARIABLES
During an earlier phase of this project, industry represen-
tatives, consumers, and manufacturers identified 17 operating and
maintenance variables considered important in achieving optimum
TF for AAC equipment. These variables are listed in Table 2.
Five variables were selected for testing on the basis of
the number of tines it was identified by different sources, the
anticipated size of effect on TE, the ability to simulate it
within the prescribed test methodology, and finally, the limi-
tation of laboratory time. The five selected teat variables
were:
o Restricted air lines
o Booth air rate
o Gun cleanliness
o Restricted paint lines
o Fan (or horn) air
-------
TABLE 2. OPERATING AND MAINTENANCE VARIABLES FOR
AAC SPRAY EQUIPMENT*
Atomizing air
Booth air rate
Booth configuration
Cure schedule (time, temperature)
Paint discharge technique
Equipment design
Plash off
Gun cleanliness
Gun condition
Gun-to-target distance
Operator error
Paint mass flow rate
Paint characteristics
Restricted air supply
Restricted paint supply
Shaping air (fan air)
Target configuration
* As mentioned by industry sources contacted
-------
Some of the variables could be quantitatively simulated
(for example by varying paint pressure), while others could only
be simulated qualitatively.
EXPERIMENTAI DESIGN
An experimental design was developed to accommodate the
limitations of testing while addressing the effects of each
variable as completely as possible.
The first restraint on experimental design was the avail-
ability of laboratory time: only about 30 test runs could
reasonably be completed during a week of testing. The second
limitation was the number and type of simulation levels for each
variable. Only two levels of linear air velocity (booth air
rate) were possible in the test laboratory, while three levels
of fan air (sometimes called horn air or shaping air) were
achievable, and five or more levels of the other variables could
be simulated. Table 3 presents the type of variable (quantita-
tive/qualitative) and levels to be accommodated in the experi-
mental design.
TABLE 3. EXPERIMENTAI, VARIABLES SELECTED FOR TESTING
AAC SPRAY SYSTEMS
Factor Quant./ No. of
ID Factor description qual. test levels
A Restricted atomizing
air lines Quant. 5
B Booth air rate (linear
velocity) Quant. 2
C Gun cleanliness Qual. 5
D Restricted paint lines Quant. 5
E Fan air (sometimes
called horn air or
shaping air) Qual. 3
A central composite experimental design was selected as the
most thorough way to examine the effects of these factors with
the fewest number of test runs. The experimental design is
characterized by combining a fractional factorial design portion
with a "star" portion, a uc; men ted by replicates.
Table 4 presents the AAC experimental design. In this
table, the abbreviations "a," "1," "0," "-1," and "-a" denote
10
-------
TABLE 4. AAC EXPERIMENTAL DESIGN
Run Number
1
2
3
4
5
6
7
8
9
10
1t
12
13
14
15
16
17
)8
19
20
21
22
23
24
25
26
27
28
29
30
-a
a
0
0
0
0
0
0
a
a
a
a
a
a
FACTOR
B C
-1
0
0
-a
a
0
0
0
0
a
a
a
a
a
a
0
0
0
0
-a
a
0
0
a
a
a
a
a
a
_E
1
•1
-1
•1
-1
-t
-1
1
-1
1
1
-1
1
1
1
1
0
0
0
0
0
0
Where:
A
B
C
D
E
Restricted air lines—test at 5 levels: a,1,0,-1,-a
Booth air rates—teat at 2 levels: 1,-1
Gun cleanliness—teat at 5 levels: a,,1fO,-t,-a
Restricted paint lines—test at 5 levels: a,1,0,-l,-a
Pan air—test at 3 levels: 1,0,-1
11
-------
the level of each variable to be tested.
base level with a good spray pattern. Level "-a" denotes the
poorest level of a variable to be tested. The intermediate
levels "1," "0," and "-I" were determined along equal spacing
from "a" to r-a" for the particular variable. Level "a"1 will
be different for each experimental design and for different
variables in the design. It remains constant for a given vari-
able in a given design.
Levels for AAC variables were determined in pretest
trials as described in the following subsection.
The first 16 test runs in the experimental design (Table 4)
are the fractional factorial portion of the design. When the
results of several factors are to be studied, a factorial design
it usually the most efficient method to use.* The basic idea
of factorial design is to alter several aspects of a test at a
time, but in such a way that the effects of individual altera-
tions can be determined. Fractional factorial designs sacrifice
some ability to test for interaction between factors but are
able to test for main effects very efficiently.
Runs 17 through 24 in Table 4, are the "star" portion of
the experimental design. This portion of the experiment tests
tbe effects of variables at the extremes of their range (for the
system under test, at "a" and "-a"). The star design broadens
the range of information gathered in the test. The star portion
of the design allows extra degrees of freedom in order to assess
lack of fit.
The experimental design used in the AAC case involves a
central composite design for factors A, C, and D. A classical
central composite design on all five variables was impossible
because of the necessity oi: using only 2 levels of variable B
and 3 levels of variable G.
The last six runs of the test design are replicates.
Replicates ace provided at the base condition of all variables
to provide a measure of i:he test precision.
*Youde~n"r W. J. and Steiner, E. H. , Statistical Manual of the
Association of Official Analytical Chemists, Arlington, VA,
1932; and Davies, 0. L., Design and Analysis of Industrial
Txperiments, Great Britain"^19T§! and Myers, Raymond H. ,
Response Surface Methodology, 1976.
12
-------
AAC TEST PERFORMANCE
AAC testing began on February 8, 1984. Equipment set up,
target assembly and hanging, foil cutting and preweighing, and
other preparatory activities were completed <-arlier in the week.
The morning of February 8 was spent mixing and adjusting test
paint to desired specifications. (See Appendix C for paint
specifications.) About 16 L (4 gal) of paint was mixed in the
20 L (5 gal) capacity paint pressure pot. The paint was stored
inside a temperature-controlled booth at 25*C + 1"C. Once
the paint was adjusted, all equipment and line? were rechecked
for proper installation and freedom from obstruction. The mass
flow meter was calibrated and zeroed. Haas flow calibration was
double checked against unatomized paint capture a-.d found to be
within 0.4 percent of the meter reading, as required.
Preliminary paint weight-percent-solids determinations were
made using the recommended ASTM method, and using Grace's own
technique. (The ASTM method is included in Appendix A.) Basi-
cally, the ASTM method required sampling the paint and adding
0.5 g (+ 0.1 g) of the paint sample to a dry preweighed 58 mm
aluminum sample dish. Solvent (3 mL) was added to the sample
prior to curing to spread the paint sample evenly in the dish.
The Graco method required a 15 mL paint sample be taken and
spread out by gravity into a 30.48 cm (12 inch) preweighed
aluminum dish. The results of the two methods did not closely
agree, and some unacceptable variance in the ASTM method was
also noted. According to the QA/QC plan, TE data cannot be
accepted unless all component measurements meet precision
requirements.
A number of weight-percent-solids determinations was made
to resolve the differences. In running these samples it was
discovered that the ASTM-method aluminum dishes were coated with
an oily compound to keep them from sticking together in storage.
This coating had to be burned off before the dish could be ised
in weight-percent-solids determinations. The weight of the oil
on the preweighed dish varied, causing the net weight percent
solids to vary as well. It was also discovered that uneven
distribution of the paint (in either method) caused differences
in curing and consequently in weight percent solids. The latter
problem manifested itself most frequently in the Graco method,
and almost none at all in the ASTM method when proper care was
taken to assure the dishes were level during curing. As detailed
in the TE test procedure, the ASTM method was used for all EPA
weight-percent-solids determinations in this report. Graco
amended their weight-percent-solids determination to follow ASTM
recommendations, but continued to take their weight-percent-
solids samples from the paint line rather than from the paint
pot as th^ TE test procedure requires. Graco weight-perc«nt-
oolids samples continued to vary somewhat from EPA values,
apparently due to the sampling technique or position. (The term
"EPA values" as used here refers to the determinations nade
following the Draft Standard Test Method of Appendix A.)
-------
While one group of technicians was performing weight-per-
cent-solids determinations, a second group of technicians was
setting up the equipment at base levels of each variable. Base
level ("a") was determined by Graco experience with the test
paint and spray painting system. Selection of base level was
confirmed by checking the spray pattern at base level for a good
pattern. No adjustments were required from Graco-recommended
base levels after checking the spray pattern. Base levels were
thus determined as shown in Table 5.
Deteriorated levels were selected by setting all variables,
except the subject variable (for each variable in turn), at
the base level. The subject variable was altered until a
significantly worse spray pattern could be discerned. The spray
pattern was checked by spraying onto a paper target for 5 or €
seconds then observing the resulting pattern. Deteriorated
factor levels ("-a") thus determined are shown in Table 5.
Intermediate factor levels were calculated to be evenly
spaced from the base level ("a") to the deteriorated level
("-a"). Intermediate levels are also shown in Table 5.
Selection of levels for gun cleanliness were made by trial
and error pattern testing of air caps with different holes
plugged. The resulting pattern of plugged holes for deterio-
rating levels of gun cleanliness is shown in Figure 1. Gun
cleanliness must be considered qualitative because of the nature
of the progressively more plugged air cap. Atomization and TE
may be affected as much or more by the geometry and design of
the plugged holes than by the total plugged area.
Booth air rates were determined by the only two available
levels. Neither level should be assumed to be an ideal level.
Base level was selected at the normal air flow level for the
booth, rather than the artificially lowered level. Rates at
each level were measured using a hot wire anemometer.
Level selection was completed on February 8, 1984. AAC TE
testing began on February 9, 1984. TE runs were made in a
randomized order based on the experimental design in Table 4.
During a single 16-hour experimental day 15 runs were made.
Three of the runs were thrown out because of incomplete data,
underspray, or losing paint from the targets because of dripping
or accidental contact with wet targets. These three runs were
repeated at the end of the day.
TE testing of AAC equipment was completed on February 10,
1984, after pertorming the remaining 15 runs. One run (Run 10)
was identified as an outlier by the QA/QC analysis (Refer to
Appendix B); it was repeated immediately. Weight-percent-solids
samples were taken at the completion of testing as required by
the draft Standard Test Method.
14
-------
TABLE 5. LEVELS OF OPERATING AND MAINTENANCE
VARIABLES—TESTED ON AAC SPRAY PAINTING
EQUIPMENT
Quant/ No. of
Factor qual. levels Test levels
A. Restricted automizing
air lines* Quant. 5 a= 239.0 kPa (20 psig)
1= 218.6 kPa (17 psig)
0= 197.9 kPa (14 psig)
-1= 177.2 kPa (11 psig)
-a= 156.6 kPa (8 psig)
B. Booth air rate +
(linear velocity) Quant. 2 1= 0.61 m/s (120 ft/min)
-1= 0.36 m/s (70 ft/min)
C. Gun cleanliness t Qual. 5 See Figure 1
D. Restricted paint lines Quant. 5 a= 180.7 kPa (11.5 psig) z
1- 170.3 kPa (10.0 psig)
0= 160.0 kPa (8.5 psig)
-1- 149.7 kPa (7.0 psig)
-a= 139.3 kPa (5.5 psig)
E. Fan air I Qual. 3 1s wide open
0= one turn shut
-1= two turns shut
* Measured at the spray gun.
+ Actual booth air rates varied from 100 to 140 ft/min for level "+1"
and 50 to 90 ft/min for level "-1." Average air velocities are used in
this table.
t Deteriorating gun cleaniness was simulated by blocking air cap holes
as shown in Figure 1.
z Measured at control panel approximately 20 feet from spray gun.
f Fan air (sometimes called horn air or shaping air) was adjusted by
setting the control knob on the gun wide open, then turning it the
required number of turns towards the closed position.
15
-------
Level +ai all holes open
Level +1: 3 holes plugged
Level Oi 4 hcles plugged Level -1: 6 holes plugged Level -a: 8 holes plugged
Figure 1. Air atomized conventional air cap (frontal view) showing
selection of test levels for gun cleanliness
-------
TEST RESULTS
TE's were calculated according to the Draft Test Standard
Method. The final AAC results are presented in Table 6. Some
corrections were made to the original TE test data because of
mathematical errors or incorrectly recorded weights. These
corrections are reflected in Table 6.
STATISTICAL ANALYSIS
Regression equations were developed to fit the TE data for
each target design. The regressions developed for AAC equipment
are based on the data in Table 6, which were developed according
to the experimental design in Table 4. Both qualitative and
quantitative variables were coded into the regression analysis
according to the level rather than their numerical value during
the test. The coding procedure is a simple "centering" and
scaling of variable levels. The variable levels for the quanti-
tative variables were evenly spaced. The following represent
the coded or "design units."
Quantitative Coded level
variable level for regression equation
a
+1
0
-1
-a
2
1
0
-1
-2
By using these coded values for levels, the regressions
become useful for either SI (Le Systeme International d'Unites)
or standard U.S. industrial units. Actual test values at different
levels than those tested here may be coded by interpolating
linearly according to the above table. Thus, a value exactly
halfway between "-1" and "-a" would take on the numerical value
of -1.5 in the regression equation.
Transformation to design units is standard procedure when
one builds models based on a planned experiment involving
quantitative variables. It allows for interpretation of re-
gression equation and tests to be in terms of units that are
scale free and determined by the region of experimentation
selected by the scientist.
Qualitative variable levels were coded in much the same
manner, assigning either zero or one to the level for each
variable. The "0" level of each qualitative variable was
arbitrarily set at zero for all cases in this report. For
17
-------
TABLE 6. AIR ATOMIZED CONVENTIONAL TEST RESULTS
Percent Transfer Efficiency
Run number Ft VC Graco
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
6^.5
85.9
83.8
•»7.1
73.8
75.8
77.8
57.4
82.0
58.2
60.0
85.3
61.0
68.7
70.7
65.0
78.1
63.5
66.7
64.8
59.5
73.5
80.5
65.7
58.3
59. 1
58.4
58.2
60.4
57.1
11.9
15.2
16.5
14. 0
13.7
13.0
12.3
11.0
15.7
10.7
9.6
16.4
10.7
13.6
13.5
12.6
15.1
11.7
12.2
11.5
11.2
13.7
15.7
11.7
10.3
11.2
9.5
11.2
11.3
10.5
27.8
39.9
41.7
36.3
33.2
34.6
38.5
25.8
38.2
25.0
26.1
38.4
29.0
31.4
32.3
29.5
38.9
30.1
30.3
29.5
28.7
33.2
35.9
28. i
25.1
26.2
26.2
27.9
28.3
25.7
18
-------
qualitative variables tested at three levels, the value of the
variable associated with that particular level is denoted by
dummy variables xRd1 and *nd2» which take on the values:
Factor Experimental Regression level
X
n
x
n
x_
level
+ 1
0
-1
«nd1
0
0
1
Znd2
1
0
0
Qualitative effects can only be determined on a relative
basis. For all of the regressions in this report, the "0"
experimental level has been designated the base level of
comparision for qualitative O&M factors.
For qualitative variables at five levels, the assignment of
regression levels proceeds according to:
Factor Experimental Regression level
X
n
x
n
x
n
x
n
x
level
a
+1
0
-1
-a
"nd1
0
0
0
0
1
*nd2
0
0
0
1
0
Xnd3 '
0
1
0
0
0
-------
Because x, and x- cannot be measured on a continuous quanti-
tative scale, they are termed qualitative variables. The
various levels of such variables used in the experiment are
represented in the analysis of variance and the regression
analysis by "dummy variables.* Thus, for gun cleanliness, x-,
and for fan air, x_, dummy variables were introduced. For gan
cleanliness, dummy variables take on the following values:
Level
a
4-1
0
-1
-a
Gun cleanliness
*3d1
0
0
0
0
1
0
0
0
1
0
0
1
0
0
0
1
0
0
0
0
Gun condition
See Figure 1
See Figure 1
See Figure 1
See Figure 1
See Figure 1
Thus, for gun cleanliness at level *+1", all dummy variables
except x->d3 take on the value of zero; x,d, takes on the
value of one.
In the case of fan air, dummy variables x,.d1 and x,.d2
take on the values shown in the table below.
Fan air
Level
+1
0
-1
*5d1
0
0
1
*5d2
1
0
Gun setting
wide open
one turn shut
two turns shut
The analysis of variance results '.n a comparison of the
variance associated with each experimental variable to the
inherent error associated with repeat observations. This com-
parison is accomplished by forming the P statistic, the ratio for
each value of F has a probability associated with It given the
number of degrees of freedom in the numerator and in the denomi-
nator. When the probability of achieving a given value of F by
chance is less than 0.05, the effect is said to be significant
at the 0.05 level. The F statistics and associated probability
for all factors found to be significant are presented in Table 7.
(See Appendix G for a glossary of statistical terms.)
20
-------
TABLE 7. AAC F-STATISTICS (F) AND ASSOCIATED
PROBABILITY (P)*
Effect
X1
Plat Panel
F P
140.89 .0001
27.84 .033
Vertical Cylinder
F P
30.45
4.49
.0027
.0376
X3d2
X4
X5d1
X5d2
X1X2
x2
X4
X2X5d1
x.x.d.
—
64.72
163.76
21.27
11.59
32. US
10.21
22.42
- —
.0005 8.66
.0000 16.88
.0058
.0192
.0024
.0241
.0520
—
.0321
.0093
-
-
-
-
-
Grace
F P
36.32 .0000
8.57 .0022
6.36 .0660
5.33 .0120
53.17 .0000
28.14 .0001
6.01 .0800
7.66 .0034
9.43 .0015
*F and P are dimensionless terms. Refer to Appendix G for a
definition of those terns.
21
-------
In the case of all equipment types, a regression model was
postulated that applied for all target types. The model considered
contained linear and sometimes quadratic effects for the con-
tinuous variables and dummy variables for the quatitative variables.
Certain interactions were put into the model on the basis of the
best engineering experience available. All possible interactions
could noc be estimated due to limitation of time and resources.
In all cases, the experimental design was constructed to accommo-
date the model terms'. Terms that were significant on the basis
of an F-test were retained and the final regression model is
reported for each target type.
In the case of air atomized conventional, the following
model terms were considered:
o linear in x.
o quadratic in x^
o linear in x.
o dummy variables in x3
o linear in x.
o quadratic in x.
o dummy variables in xg
o interactions between x, and x2
o interactions between x_ and x5
Regression models were constructed using mainframe SAS*
capabilities for e;-ch of the three target types including all of
those affects found to be significant at the 0.05 level. The
resulting models and their associated R (proportion of the
overall variance explained by the regression) ere presented
below:
Flat Panel Target
The regression model developed for AAC FP is:
TE - 68.95 - 3.12x1 - 0.98x2 +
"•Statistical Analysis System, SAS Institute, P.O. Box 10066,
Raleigh, NC 27605.
22
-------
The negative coefficient on x. indicates a tendency for TB
to decrease with an increase in restricted air levels. However,
there is an important interaction between the two factors as
evidenced by a positive and significant coefficient of x^j.
While this interaction does suggest that TE continues to
decrease with an increase in restricted air levels, the magni-
tude of that increase depends on the booth air level. The same
is true for x-, booth air. The mixed coefficients on x4 and
X? indicate tnat TE increases with an increase in paint
line levels for the low paint line levels, but the amount of
increase tapers off as the paint line levels become larger.
The heavy positive coefficient on x-d. suggests that TE
increases when "-1" level is used oft fan air. The negative
coefficient on xsd, suggests a decrease in TE at the
•+1" level of fan air.
Since both the magnitude and direction of «-he effect of fan
air on TE are dramatically different at different- levels of fan
air, the operator must be very careful to establish the appropri-
ate fan air level for optimum TE.
The proportion of the overall variance explained by the
regression (R ) is 0.97. This R indicates a tight fit
of the regression model to experimental data. The standard
deviation of replicate runs was 1.098, well within the targeted
2.0 standard deviation** (expressed in units of transfer effici-
ency ).
The error in the regression model due to lack of fit was
determined to be insignificant at the 0.05 level. The P-
statistic for lack of fit was 1.08 (probability-0.49).
Vertical Cylinder Target
The regression model developed for &AC VC is:
TE- 11.99 - 0.91X, - 0.28x2
+ C.44x4 + 2.83X36,
The P-statistics are shown in Table 7.
The negative coefficient on x. indicates that TS decreases
with an increase in air pressure. The negative coefficient
on x0 indicates that TE decreases with an increase in booth
air rates. The positive coefficient on x. indicates that TE
increases with increasing paint pressure. The positive coef-
ficient on fan air, only at level -1, indicates there is
**CENTEC Corporation, "Development of Draft Stan-Sard Test Method
for Spray Painting Transfer Efficiency," for USCPA under Con-
tract 68-03-1721, Task 2.
23
-------
something different about fan air at this level than at other
test levels. Since fan air is a qualitative factor, it can only
be speculated that certain levels of fan air affect TE more than
other levels. Reducing fan air from the "0" level improved
transfer efficiency, but increasing fan air did not signifi-
cantly degrade TE.
The proportion of the overall variance explained by the
regression (Rz) is 0.79. This R is lower than for the FP
target, probably due to the lack of overall variation in TE
level. In the FP case, the total standard deviation for all
experimental data was 9.61, while it is only 2.030 in this case.
The FP target TE's were more strongly affected by experimental
variations and were thus easier to model. The effect of experi-
mental variations on VC TE's is so small that it is almost below
the targeted 2.0 standard deviation for replicates. Small
effects are difficult to tightly fit with renression models.
The standard deviation of VC replicates was 0.706. This
low standard deviation can also be attributed to the overall
insensitivity of the VC targets to experimental variables,
The error in the regression model due to lack of fit is
insignificant: F - 1.S3 (probability - 0.33).
TE - 32.22 - 1.61X, -1.24x2
Graco Target
.22 - 1.6
+ 1.23x4 - 0.64xJ
•f 0.67*^2 •*• 1.73x2x,d2
The negative coefficient on x. indicates that TE decreases
with an increase in air pressure (restricted lines}. The
negative coefficient on x, indicates a similar decrease in
TE with increasing booth air rates. Cnce again, the positive
coefficient on the interaction between x. and x, indicates
that these trends are not constant but rather depend on the
level of the interacting variable. For example, the negative
trend of TE with respect to air pressure Is not as pronounced
when booth air rate is high, according to the regression equation.
The positive coefficient on x3 (only at level d.) indicates
that TE is affected for one level of gun cleanliness. The effect
of gun cleanliness on TE at other experimental levels is insig-
nificant.
Again, the positive coefficient on s. and negative sign on
the quadratic term suggests & nonlinear effect of paint pressure.
There is a positive slope on paint pressure until one goes beyond
the x »1.0 level, roughly. At that point, the effect becomes
negative.
24
-------
The strongly positive coefficient on x& at the d, level
contrasts dramatically with the strongly negative coefficient at
the d~ level. This indicates that fan air can substantially
increase or decrease TB. Again, very careful selection of fan
air levels is warranted by these results.
Finally, two interactions are noted: restricted air lines
and booth air rate interact to affect TB, and booth air rate
interacts with fan air (only at the d2 level) to affect TE.
The proportion of the overall variance explained by the
regression (R > is 0.97. The standard deviation of repli-
cates is 1.261, well below the target value. The error due to
lack of fit is insignificant: P - 0.92 (P » 0.773).
Tables 8, 9, and 10 present a comparison of predicted
values, based on the derived regression, with observed values
for the flat panel, vertical cylinder, and Graco targets res-
pectively. The residual is the difference between predicted and
observed values. The 95 percent confidence limits for the wean
give the upper and lower bounds of the range within which the
mean of transfer efficiency (the "true regression") at
each experimental condition lies with 95 percent confidence.
AAC TEST CONCLUSIONS
The regressions previously presented illustrate the
differences that target configuration can make in TB. Even
with these differences, however, there are basic consistencies
between the results. Three factors (restricted atomizing air
lines, restricted paint lines, and fan air) were identified as
significant for all tested target configurations. A fourth
variable, booth air, was significant for FP and Graco targets
and very nearly significant for VC targets as well. The con-
sistency of these results strongly implies that these four
factors have a critical influence on TE regardless of target
configuration.
Thus, selection and maintenance of appropriate atomizing
air pressure and paint pressure should be given regular atten-
tion by the operator. Pan air rates have o strongest influence
on TE across all target types, as demonstrated by highly
significant P-value and the large coefficient in each of the
AAC regressions. Pan air levels are often set by individual
operators according to their own judgment. Por optimum TE,
plant management should determine optimum spray painting
conditions through test runs and then specify those conditions
for the operator.
Some reev.iluation of booth air rates may be justified by
the test reouits, whicli indicate that the lowest level of booth
air rate should be selected to maximize TB. Care should be
taken to adhere to all safety and environmental regulations,
25
-------
TABLE 8. AAC-FF COMPARISON Of PREDICTED VERSUS ACTUAL TRANSFER EFFICIENCIES
to
en
Observation
number
1
2
3
4
5
6
7
6
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Observed
value
62.50
85.90
83.60
77. »0
73.80
75.80
77.30
57.40
82.00
58.20
60.00
85.30
61.00
68.70
70.70
65.00
78.10
63.50
66.70
64.80
59.50
73.50
80.50
65.70
58.30
59.10
58.40
58.20
60.40
57.10
Predicted
value
64.69
87.87
82.13
78.81
74.87
73.57
79.32
56.47
80.61
57.43
63.17
84.55
61.71
67.46
70.44
62. n
77.15
62.62
67.92
67.92
59.49
70.98
80.23
63.13
58.98
58.18
58.98
58.98
58.98
58.98
Residual
-2.19
-1.97
1.66
-1.71
-1.07
2.22
-1.52
0.92
1.38
0.76
-3.17
0.74
•-0.71
1.23
0.25
2.77
0.94
0.87
-1.22
-3.12
0.00
2.51
0.26
2.56
-0.68
0.11
-0.58
-0.78
1.41
-1.88
Lower 95% CL
for mean
62.47
85.48
79.73
76.54
72.47
71.20
77.16
54.51
78.22
55.20
60.93
82.13
59.35
65.09
68.19
60.75
74.25
59.72
65.64
65.64
56.30
68.22
78.3?
61.44
57.42
57.42
57.42
57.42
57.42
57.42
Upper 95% CL
for mean
66.92
90.27
84.52
81.07
77.26
75.94
81.47
58.44
83.01
59.65
65.42
86.98
64.08
69.82
72.68
63.69
80.05
65.52
70.20
70.20
62.67
73.73
82.12
64.82
60.55
60.55
60.55
60.55
60.55
60.55
-------
TABLE 9. AAC-VC COMPARISON OF PREDICTED VERSOS ACTUAL TRANSFER EFFICIENCIES
Observation
number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Observed
value
11.90
15.20
16.50
14.60
13.70
13.00
12.30
11.00
15.70
10.70
9.60
16.40
10.70
13.60
13.50
12.60
15.10
11.70
12.20
11.50
11.20
13.70
15.70
11.70
10.30
11.20
9.50
11.20
11.30
••0.50
Predicted
value
12.73
16.44
15.56
15,01
13.74
13.19
14.07
10.35
14.62
10.91
11.79
15.89
12.17
13.05
13.61
11.23
14.08
10.44
11.70
11.70
11.38
13.14
14.54
11.70
10.76
10.76
10.76
10.76
10.76
10.76
Residual
-0.63
-1.24
0.93
-0.41
-0.04
-0.19
-1.77
0.64
1.07
-0.21
-2.19
0.50
-1.47
0.54
-0.11
1.36
1.01
1.25
0,49
-0.20
-0.18
0.55
1.15
-O.OO
-0.46
0.43
-1.26
0.53
0.53
-0.26
Lower 95% CL
for mean
11.93
15.44
14.72
14.09
12.74
12.22
13.25
9.50
13.65
10.03
11.06
14.92
11.28
12.20
12.73
10.70
13.08
9.50
11.06
11.08
10.38
12.20
13.77
11.08
10.05
10.05
10.05
10.05
10.05
10.05
Upper 95% CL
for mean
13.53
17.45
16.40
15.92
14.75
14.16
14.88
11.2i
15.59
11.79
12.52
16.86
13.07
13.91
14.49
11.77
15.07
11.38
12.33
12.33
12.37
14.08
15.32
12.33
11.48
11.48
11.48
11.48
11.48
11.48
-------
TABLE 10. AAC-GRACO COMPARISON OF PREDICTED VEKSUS ACTUAL TRANSFER EFFICIENCIES
to
oo
Observation
number
1
2
2
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
2*>
30
Observed
value
27.40
39.90
41.70
36.80
33.20
34.60
38.50
25.60
33.20
25.00
26.10
38.40
29.00
31.40
32.40
29.50
38.90
30.10
30.80
29.50
28.70
33.20
35.90
28.10
25.10
26.20
26.20
27.90
28.30
25.70
Predicted
value
23.03
41.58
40.55
35.32
34.57
34.86
35.89
25.81
38.45
24.90
25.94
39.20
29.11
33.14
31.91
29.68
38.0?
28.91
30.98
30.98
28.44
33.35
36.24
28.61
26.63
26.63
26.63
26.63
26.63
26.63
Residual
-0.63
-1.68
1.14
0.97
-1.37
-0.26
2. 60
-0.01
-0.25
0.09
0.15
-0.80
-0.11
1.25
0.38
-0.18
0.88
1.18
-0.68
-1.48
0.25
-0.15
-0.34
-0.51
-1.53
-0.43
-0.4J
l.>6
1.66
-0.93
Lower 95% CL
for mean
26.39
40.05
39.04
33.78
33.05
33.24
34.60
24.45
36.89
23.29
24.32
37.57
27.43
28.53
30.27
28.30
36.13
27.03
29.56
29.56
26.43
31.63
35.06
27.32
25.64
25.64
25.64
25.64
25.64
25.64
Upper 95% CL
for mean
29.67
«.11
42.06
36.86
36.10
36.47
37.18
27.17
40.02
26.52
27.55
40.82
30.79
31. 7S
33.55
31.07
39.88
30.78
32.41
32.41
30.45
35.08
37.43
29.91
27.62
27.62
27.62
27.62
27.62
27.62
-------
as well as providing for worker comfort when considering lower-
ing booth air rates. The regressions in the previous section
cen be used to make reasonable estimates of potential savings.
These savings should be weighed against all costs before a
change is made.
The Graco and FP target configurations also identified
interactions between booth air rate and other variables. These
interactions, while statistically significant, are not con-
sidered large enough to warrant direct practical attention.
It is recommended that the plant management emphasise selection
and maintenance of optimum levels for more critical variables.
The Graco target configuration identified one level of gun
cleanliness as significantly affecting TE. Since this finding
is not consistent across target J.ypes and ic relatively small
when it does appear, it is not considered critical to optimizing
TE. This is not to say that gun cleanliness is unimportant to
the spray painter. Gun cleanliness is one of the few O&M
factors universally stressed by gun manufacturers, spray paint-
ers, and other early participants in the test program. Gun
cleanliness has a profound effect on paint finish, gun life, and
internal gun condition, which were not tested in this program.
Only the aspect of gun cleanliness tested during this experiment
is considered unimportant for AAC spray equipment.
29
-------
SECTION 5
AIR ATCHIZED ELECTROSTATIC SPRAY EQUIPMENT
EQUIPMENT DESCRIPTION
A Graco Model AS-4000 manual electrostatic air spray gun
was was selected for AAE testing. The Model AS-4000 gun was
considered typical of an external-mix, manual electrostatic
spray gun. The sprav gun was equipped with a 177033 air cap,
776976 fluid tip, and a 215864 needle. Paint flow was manually
initiated by opening a valve on the paint supply line. The
spray gun was fixed in open position.
Graco standard black enamel was selected as the test paint.
The paint averaged about 28.7 weight percent solids when cut to
30.4 seconds
-------
OPERATING AMD MAINTENANCE VARIABLES
As listed in Table 11, 19 variables were identified
through interviews and literature that may have potential
to exert an important effect in achieving optimum TE. Seven
of the identified variables were selected for AAE testing
on the basis of: (1) the number of tiaies the variable was
identified for AAE by different sources, (2) the ability to
simulate the variable within the prescribed test methodology,
and (3) the limitation of laboratory time. ?his prior
knowledge enabled us to limit the scope of TE experiments to
only variables of particular interest.
The selected test variables were:
o Restricted atomizing air lines
o Booth air rate (linear velocity)
o Gun cleanliness
o Restricted paint lines
o Fan air (sometimes called horn air or shaping air)
o Tip voltage
o Electrode position
Restricted atomizing air lines can be simulated by de-
creasing the pressure of the air supply to the spray gun. An
air regulator was used for reducing the air pressure to desired
levels. Restricted paint lines were simulated by decreasing the
paint supply pressure in a similar manner.
Booth air rate (linear velocity) was available at only two
levels at this facility, 0.36 m/s and 0.61 Vs (70 ft/min and
120 ft/min) respectively. Gun cleanliness was simulated by
blocking certain air holes in the air cap in a progressively
worse pattern as shown in Figure 2. Fan air was adjusted by
using the adjustment knob on the spray gun, while voltage sup-
plied to the tip was adjusted at the power supply. Electrode
position was set manually as shown in Figure 3.
EXPERIMENTAL DESIGN
An experimental design was deve" ed to accommodate the
limitations of testing while addressing the effects of each
variable as completely as possible.
The first restraint on experimental design as noted
previously was the availability of laboratory time: only
31
-------
TABLE 11. OPERATING AND MAINTENANCE VARIABLES
FOR AAE SPRAY EQUIPMENT*
Atomizing air
Booth air rate
Booth configuration
Conveyor speed
Cure schedule (time, temperature)
Electrode position
Equipment design
Plash off
Gun cleanliness
Gun condition
Gun-to-target distance
Operator error
Paint discharge technique
Paint mass flow rate
Paint characteristics
Restricted air supply
Restricted paint supply
Shaping air (fan air)
Target configuration
*as mentioned by industry sources contacted
32
-------
Level +ai all holes open
Level +li 3 holes plugged
Oi 4 holes plugged Level -It 6 holes plugged Level -ai 8 holes plugged
Figure 2. Air atomized electrostatic air cap (frontal view) showing
selection of test levels for gun cleanliness
-------
FRONT VTEW
SIDE VIEW
V
^WH
Jb
LEVEL 1
7
Jb
LEVEL 0
LEVEL -1
Figure 3. Air atomized electrostatic electrode
position test levels
34
-------
about 30 test runs could reasonably be completed during a
week of testing. The second limitation was the number and
type of simulation levels for each variable.
Table 12 presents the type of variable (quantitative/qual-
itative) and levels to be accommodated in the AAE experimental
design.
A variation of a central composite experimental design was
selected as the most thorough way to examine the effects of these
factors with the fewest number of test runs and still allow for a
regression model to be constructed. The experimental design is
characterized by combining a fractional factorial design portion
with a "star" portion, augmented by replicates. A slight variation
central composite experimental design was constructed for factors
A, C, D, E, F, and G. Five levels were required for factors A,
C, D, and P but only three levels for factors E and G. Thus the
design levels for the star points could not be the same for all
variables. In addition, as in the AAC, the replicates were not
at the traditional center of the design. For pragmatic reasons
the replicates were taken at the extremes in each variable.
Table 13 presents the AAE experimental design. In this
table, the abbreviations "a," "1," "0," "-I," and "-a" denote the
level of each factor to be tested. Level "a" denotes the base
level with a good spray pattern. Level "a" is likely to be dif-
ferent for each factor in each experiment. It remains constant
for a given factor in a given experiment. Level "-a" denotes
the poorest level of a factor to be tested. The intermediate
levels "1," "0," and "-1" are determined along equal spacing
from "a" to "-a" for the particular factor. Factor levels for
AAE testing were determined in pretest trials as described in
the following subsection.
The first 16 test runs in the experimental design are the
fractional factorial portion of the design. When the effects of
several factors are to be studied, a factorial design is usually
the most efficient method to use.* The basic idea of factorial
design is to alter several aspects of a test at a time, but in
such a way that the effects of individual alterations can be
determined. Fractional factorial designs sacrifice some ability
to test for interaction between factors but are able to test for
main effects very efficiently.
*Youden. W. J. and Steiner, E. H., Statistical Manual of the
Association of Official Analytical Chemists, Arlington, Va.,
1982; and Davies, O. L., Design and Analysis of Industrial
Experiments, Great Britain, 1979.
35
-------
TABLE 12. OPERATING AND MAINTENANCE VARIABLES
FOR AAE SPRAY PAINTING EQUIPMENT
Variable
Quant/ No. of
qual. levels
A. Restricted atomizing
air lines Quant.
B. Booth air rate
(linear velocity) Quant. 2
C. Gun cleanliness Qual. 5
D. Restricted paint lines Quant. 5
E. Pan air (shaping air or
horn *ix) Qual.
P. Voltage
G. Electrode position
Quant.
Quel.
3
5
3
36
-------
TABLE 13. AAE EXPERIMENTAL DESIGN
Run number A B
1
2
3
4
5
6
7
8 -1
9 -1
10 1
11 -1
12 1
i3 1
14 -1
15 1
16 -1
17 -a
18 a
19 0
20 0
21 0
22 0
?.3 0
24 0
25 0
26 0
27 0
28 n
29
30
31
32 1
33 1
34 1
Variable
C D
-1
0
0
a
-a
0
0
0
0
0
0
0
0
a
a
a
a
a
a
0
0
0
0
-a
a
0
0
0
0
0
0
G
0
0
0
0
•1
1
0
0
0
0
1
1
1
1
1
1
-1
-1
0
0
0
0
0
0
0
0
-a
a
C
0
0
0
0
0
0
0
0
0
0
0
•1
1
Where:
A
B
C
D
E
P
G
Restricted air lines—test at 5 levels: a,1,0,-1,-a
Booth ait rates—test at 2 levels: 1,-1
Gun cleanliness—test at 5 levels: a,1,0,-1,-a
Restricted paint lines—test at S levels: a,1,0,-1,-a
Pan air—test at 3 levels: 1,0,-1
Voltage—teat at 5 levels: a,1,0,-1,-a
Electrode position—test at 3 levels: 1,0,-1
37
-------
Runs 17 through 28 in Table 13, are the "star" portion of
the experimental design. This portion of the experiment tests
the effects of variables at the extremes of their range (for the
system under test, at "-a" and "a"). The star design broadens
the range of information gathered in the test. The star portion
of the design allows extra degrees of freedom in order to assess
lack of fit.
The final six runs o£ the experimental design are repli-
cates. Replicates are provided at the base condition of all
variables to provide a measure of the precision of the test.
AAE TEST PERFORMANCE
AAE testing was conducted from February 13 to February 17,
1984. Spray equipment was set up on February 13 and initial
spray pattern was checked. Some difficulty was encountered in
establishing a good spray pattern for base levels. The fluid
tip and valve seats were replaced in the spray gun, and the
spray pattern improved. Base levels ("a") for each variable
were established as described in the Test Method (Appendix A).
Deteriorated levels ("-a") were determined by setting all
factors except one at the base level, then decreasing the level
of the selected variable until a noticeably worse spray pattern
resulted. Deteriorated levels of each variable were determined
in turn. Intermediate levels were calculated evenly between
•a" and "-a" for each quantative variable. The final selection
of test levels is presented in Table 14.
Deteriorated gun cleanliness levels were determined by
progressively plugging more holes in the air cap. Final gun
cleanliness levels are shown in Figure 2.
Electrode position was selected through trial and error
spray pattern checks after alterations in electrode position
were made. Selected electrode positions for AAE TE testing are
illustrated in Figure 3.
The experimental design in Table 13 was followed. Three
blocks of runs were made; all of the runs in a block were of
the same electrode position. Total randomization could not
be accommodated without introducing an unacceptable error in
trying to duplicate the desired electrode position. Pretest
trials demonstrated the inability to assure consistent levels
ot electrode position in a totally random experiment. (Spray
gun design caused straightening of the electrode whenever air
cap changes were made.)
TE testing began February 14 after all documentation and
QA/QC measures were completed. Six tests runs were completed.
On the second day of testing 17 runs were completed, with
38
-------
TABLE 14. LEVELS OP OPERATING AND MAINTENANCE VARIABLES
TESTED ON AAE SPRAY PAINTING EQUIPMENT
_ Variable ____ Test levels -
A. Restricted atomizing air lines* a= 293kPa (20 psig)
1« 218.6kPa (17 psig)
0= 197.9kPa (14 psig)
-1- 177.2kPa (11 psig)
-a= 156.6kPa (8 psig)
B. Booth air rate + , J %
(linear velocity) 1- 0.61m/s (120 ft/min)
-1- 0.36m/s (70 ft/min)
C. Gun cleanliness t See Figure 2
D. Restricted paint lines z *• ISO.TkPa (15.5 psig)
1- 170.3kPa (13.5 psig)
0= ISO.OXPa (11.5 psig)
-1- 149.7kPa (9.5 psig)
-a- 139.3kPa (7.5 psig)
E. Pan air t 1- wide open
0= 1 turn shut
-1-» 2 turns shut
P. Voltage ** »• '2 kV
1- 63 W
0- 54 KV
-1- 45 kV
36 kV
G. Electrode position 1" normal
(See Figure 3) 0- bent tt
-1- clipped off
"Measured at the spray gun.
4Actual booth air rates varied from 100 to 140 fpm for level
•+1" and 50 to 90 fpm for level "-1". Average air velocities
are used in this table.
tDeteriorating gun cleaniness was simulated by blocking air cap holes
as shown in Figure 2.
zMeasured at control panel approximately 20 feet from spray gun.
tFan air (sometijnes called horn air or shaping air) was adjusted by
setting the control knob on the gun wide open, then turning it the
required nuntoer of turns towards the closed position.
**Monitored at power supply.
ttBent down and to the left.
39
-------
the remainder finished on February 16, 1984. All data were
gathered according to the requirements of the Test Procedure
(Appendix A) and QA/QC plan (Appendix B).
All data satisfied the requirements of the outlier analysis;
no TE test runs had to be repeated.
TEST RESULTS
TE's were calculated according to the test plan. Final
results are presented in Table 15. Seme corrections were made
to the original TE data when a QA scan identified several
unusual foil weights. These foils were reweighed and the cor-
rect weights used to recalculate TE values. These corrections
are reflected in Table 15.
STATISTICAL ANALYSIS
Regressions are described for each target type in the same
manner as described previously for air atomized conventional
equipment. A discussion of how co uue the regression equations
is included in the AAC Statistical Analysis section.
For AAE equipment, the variables are designated as follows:
x.arestricted atomizing air lines
x2-booth air rate (linear velocity)
x-agun cleanliness
x.orestricted paint lines
x=fan air (shaping air or horn air)
x -voltage
o
x."electrode position
TE°transfer efficiency
Factors x.,x_ and x_ are qualitative variables and
therefore have duffimy variables associated with them. The
regressions developed for each target type follows.
In the case of air atomized electrostatic- spray equipment,
engineering judgment suggested that the following model terms,
including interactions, should be considered.
o Linear and quadratic in x.
o Linear in x_
40
-------
TABLE 15. AAE TEST RESULTS
TE result
Run number FP VC Graco
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
93.3
90.2
92.4
96.2
92.2
93.1
93.1
94.3
96.5
87.4
98.8
96.2
91.7
96.7
89.6
98.8
102.0
88.4
94.4
93.0
88.4
100.0
96.5
93.0
90.5
96.0
87.0
94.9
96.7
86. 8
92.9
94.0
98.6
95.4
33. 8
37.1
45.3
61.1
48.5
28.1
67.3
40.9
60.5
34.6
f.3.3
29.8
30.5
41.7
49. B
7. .4
50.4
42.0
47.4
45.0
44.5
56.5
3U9
29.6
28.6
44.8
36.3
52.7
76.4
77.3
79.6
75.4
79.8
73.9
56.2
60.2
60.2
72.2
62.8
49.5
77.7
57.6
70.1
54.4
72.8
57.3
52.6
62.3
61.5
78.7
67.4
60.8
67.9
64.0
65.5
70.5
56.7
56.7
54.9
70.0
54.0
69.1
78. 1
78.6
77.7
75.0
77.9
80.8
41
-------
o Dummy variables in x_
o Linear and quadratic in x.
o Dummy variables in x_
o Linear and quadratic in x,
o
o Dummy variables in x_
o Interaction between x_ and x5
Flat Panel Target
TE - 94.72 - 2.27x, + 1 . 56x .,
1 4
+ 1.22xg - 3.05x7d1
All factors are significant only in linear form. No quad-
ratic factors are significant for this gun type and target con-
figuration. The negative coefficient on x. indicates a drop in
TE as air pressure increases. Trie positive coefficient on x.
suggests that TE increases with increasing paint pressure.
Likewise, the positive coefficient on xg indicates that TE
increases with increased voltage. The negative coefficient on
x-d. suggests a significant drop in TE, but only when the
electrode position is at level "-1." (See Figure 3 to visualize
level "-1" compared to the other electrode positions.)
The proportion of overall variance explained by the regres-
sion (R ) is 0.67. This is a low R . It is the result of a
lack of overall variance among test runs for this target con-
figuration. The overall variance of all of the TE determina-
tions for AAE FP was 3.69, only about one and a half TE unit
above the targeted precision of 2.0. When the overall variance
is low, it is difficult to tightly fit a regression model to
account for the small differences from run to run.
The standard deviation of the replicates was 2.070, higher
than for most other cases, but very near the target value
of 2.O.*
The error in the regression due to lack of fit was insig-
nificant with F = 1.42 (0.37 probability).
*CENTEC Corporation, "Development of Draft Standard Test Method
for Spray Painting Transfer Efficiency," for USEPA under Con-
tract No. 68-03-1721, Task 2.
42
-------
Vertical Cylinder Target
The derived regression equation for *AE VC is:
TE = 41.34 - I.11x1 - 2.66x2 - 2
* - 8.20x56, + 6.71x
In this case, several factors are significant, in both
linear and quadratic form. As in the flat panel case, the
negative coefficient on x, indicates that as air pressure
increases, TE drops. Similarly, the negative coefficient on
Xj indicates that as booth air rate increases TE decreases.
Tnis negative effect is moderated by the interaction of booth
air with x_, fan air. The positive coefficient of x-x-
indicates that the rate of change of TE with respect to booth
air depends on the prevailing level of fan air. In particular,
this slope becomes positive when the fan air is at the "wide
open" level. Gun cleanliness, x3, exerts a significant effect
on TE only at the "+1" level. (Figure 2 illustrates the different
levels of gun cleanliness.) Restricted paint lines are quadrati-
cally significant with a positive coefficient, indicating that
an increase in paint pressure also increases TE.
This may suggest that there are interactions or special
effects on TE at certain levels of fan air for this system.
The positive linear coefficient on voltage indicates that as
voltage increases, TE increases. The negative quadratic
coefficient on voltage indicates a nonlinear effect for
voltage increases. This negative coefficient moderates the
positive trend for the higher levels of voltage. Electrode
position was found to be significant only at position "+1" (shown
in Figure 3), but not for other electrode positions.
The proportion of overall variance explained by £he
regression (R ) is 0.92. This is a relatively high R , and
considered indicative of a good fit of the regression.
The standard deviation of replicate runs was 2.356, just
over the target standard deviation of 2.0 for the procedure.
The error due to lack of fit was statistically insignificant
at the 5 percent significance level (F = 2.66),
Graco Target
The derived regression for AAE testing using Graco targets
is:
TE = 66.78 - 0.89x1 - 1.14x2 -
- 0.67x4 + 1.12x -5.62x5d1 -
43
-------
The directional effecto of x., x,, x- (at "+1" level),
x (at "-1" level), and x, are the same aS for the vertical
cylinder AAE case. Three new effects are identified for Graco
targets as compared to VC targets, as follows:
o Restricted paint lines have a small negative linear
effect.
o Pan air is found significant at both the "-1" and "-H"
levels-. Both levels produce poorer transfer efficiency
compared to the "0" level.
o Electrode position is found to have a significant
effect at d1 (level "-1") and at d2 (level "-H").
The proportion of overall variance explained by the re-
gression (R ) is 0.94. This is considered a high value,
indicative of the good fit of the regression. The standard de-
viation of replicate Graco test runs was 1.8606, well within the
2.0 limitation set by the test procedure.
Table 16 presents the values of F and the associated
probability (P) for all variables and interactions found to be
significant.
Tables 17, 18, and 19 present a comparison of predicted
values with observed for the flat panel, vertical cylinder, ?nd
Graco target respectively.
AAE CONCLUSIONS
These regressions illustrate the differences target
configuration can make in TE. Even with these differences,
however, there are fundamental consistencies among the
results. Four variables (restricted air lines, restricted
paint lines, voltage, and electrode position) are significant
for all target types. Three other variables (booth air rate,
gun cleanliness, and fan air) are significant for VC and Graco
target configurations. The consistency of these results across
target types strongly implies that all of the factors tested
for AAE spray equipment have an important impact on TE.
The relative importance of each variable for a certain
target configuration should be given individual consideration by
plant management. It is recommended chat laboratory test runs
be made with plant paint and worXpiece targets to determine
optimum combinations of factor levels that result in acceptable
product finish. The developed regressions should serve as
guidelines in setting up the experimental design for site-specific
TE testing. If such tests are impractical, the regressions may
44
-------
serve as guidelines toward maximizing TE. Care mast be taken
when extrapolating the results for one spray system to another.
Previous test experience indicates that paint characteristics,
spray system characteristics, and target geometry can signifi-
cantly alter TE test results; however, the regressions may be
considered directionally sound for similar spray systems.
45
-------
TABLE 16. AAB F-STATISTICS (F) AND ASSOCIATED PROBABILITY (P)*
Flat Panel
Effect
X1
X2
x3d3
X.
4
x«
F P
22.08 .0053
-
-
13.52 .0143
3.34 .034
x,a2
•J
13.18 .065
Vertical Cylinder
F P
13.84 .0137
34.58 .0020
11.28 .020
16.31 .0099
159.6 .0000
92.1 .0002
16.62 .0076
12.15 .0175
7.68 .039
Graco
F P
5.71 .0038
9.73 .0003
13.85
3.29 .0333
11.93 .0000
6.71 .0018
122.94 .0000
18.77 .0001
28.99 .0000
10.61 .0002
F and P are dimensionless. Refer to Appendix G for a definition
of these terms.
46
-------
TABLE 17. AAC-PP COMPARISON OF PREDICTED VERSUS ACTUAL TRANSFER EFFICIENCIES
Observation
number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Observed
value
93.30
90.20
92.40
96.20
92.20
93.10
93.10
94.30
96.50
87.40
98.80
96.20
91.70
96.70
89.60
98.80
102.00
88.40
94.40
93.00
88.40
100.00
96.50
93. 00
90.50
96.00
87.00
94.90
96.70
96.80
92.90
94.00
98.60
95.40
Predicted
value
91. IS
89.66
96.71
95.22
92.78
94.26
92.11
93.60
94.20
86.61
99.76
92.17
89.73
97.31
69.06
96.65
99.25
90.17
94.71
94.71
91.60
97.82
94.71
94.71
92.27
97.16
91.66
94.71
95.73
95.73
95.73
95.73
95.73
95.73
Residual
2.14
0.53
-4.31
0.97
-0.58
-1.16
0.98
0.69
2.29
0.78
-0.96
4.02
1.96
-0.61
0.53
2.14
2.74
-1.77
-0.31
-1.71
-3.20
2.17
1.78
-1.71
-1.77
-1.16
-4.66
0.18
0.96
1.06
-2.83
-1.73
2.86
-0.33
Lower 95% CL
for mean
89.40
87.76
94. 55
94.20
91.01
92.11
90.34
91.44
92.61
84.46
97.99
90.42
87.57
95.41
86.90
94.75
97.22
88.40
93.71
93.71
89.57
96.05
93.71
93.71
90.24
95.39
90.15
93.71
94.10
94.10
94.10
94.10
94.10
94.10
Upper 95% CL
for mean
92.90
91.57
98.86
96.24
94.54
96.42
93.87
95.75
95.80
88.77
101.52
93.92
91.88
99.22
91.21
98.55
101.28
91.94
95.71
95.71
93.63
99.59
95.71
95.71
94.30
98.92
93.17
95.71
97.35
97.35
97.35
97.35
97.35
97.35
-------
TABLE 18. AAC-VC COMPARISON OP PREDICTED VERSUS ACTUAJ, TRANSFER EFFICIENCIES
oo
Observation
number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Observed
value
33.80
37.10
45.30
61.10
48.50
28.10
67.30
40.90
60.50
34.60
63.80
29.80
30.50
41.70
49.80
70.40
50.40
42.00
47.40
45.00
44.50
56.50
31.90
29.60
28.60
44.80
36.30
52.70
76.40
77.80
79.60
75.40
79.80
73.90
Predicted
value
33.49
40.57
51.46
65.82
54.08
36.38
62.94
37.96
56.30
34.16
65.16
35.74
31.27
42.79
49.24
68.04
43.88
39.44
38.68
38.68
51.52
51.52
26.34
J8.17
24.30
51.13
36.35
54.60
75.86
75.86
75.86
75.86
75.86
75.86
Residual
0.30
-3.47
-6.16
-4.72
-5.58
-8.28
4.35
2.93
4.19
0.43
-1.86
-5.94
-0.77
-1.09
0.55
2.35
6.51
2.55
8.71
6.31
-7.02
4.97
5.55
-6.57
4.29
-0.05
-0.05
-1.90
0.53
1.93
3.73
-0.46
3.93
-1.96
Lower 95% CL
for mean
27.12
34.27
45.20
60.99
47.38
30.24
56.20
31.56
49.76
28.36
58.31
29.62
24.60
36.34
42.93
62.31
38.37
32.53
33.85
33.85
44.22
44.22
19.89
32.80
15.39
31.73
31.73
48.17
71.47
71.47
71.47
71.47
71.47
71.47
Upper 95% CL
for mean
39.86
46.87
57.72
70.65
60.78
42.52
69.67
44.35
62.83
39.96
72.00
41.86
37.95
49.24
55.56
73.77
49.39
46.36
43.50
43.50
58.82
58.82
32.78
43.54
33.22
40.97
40.97
61.03
80.24
80.24
80.24
80.24
80.24
80.24
-------
TABLE 19. AAC-GRACO COMPARISON OP PREDICTED VERSUS ACTUAL TRANSFER EFFICIENCIES
Observation
number
1
2
3
4
5
6
7
6
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Observed
value
56.20
60.20
60.20
72.20
62.80
49.50
77.70
57.60
70.10
54.40
72.80
57.30
52.60
62.30
61.50
78.70
67.40
60.80
67.90
64.00
65.50
70.50
56.70
56.70
54.90
70.00
54.00
69.10
78.10
78.60
77.70
75.00
77.90
80.80
Predicted
value
53.89
63.49
62.38
71.98
65.87
51.71
74.16
60.00
68.99
51.27
74.59
56.88
50.77
63.92
61.95
75.10
65.34
61.79
65.63
65.63
69.37
66.69
55.66
57.10
55.18
71.96
54.40
68.06
77.96
77.96
77.96
77.96
77.96
77.96
Residual
2.30
-3.29
-2.18
0.21
-3.07
-2.21
3.53
-2.40
1.10
3.12
-1.79
0.41
1.82
-1.62
-0.45
3.59
2.05
-0.99
2.26
-1.63
-3.87
3.80
1.03
-0.40
-0.28
-1.96
-0.40
1.03
0.13
0.63
-0.26
-2.96
-0.06
2.83
Lower 95% CL
for mean
51.25
60.43
59.26
69.97
62.85
48.65
71.03
56.94
65.92
48.22
71.47
54.10
47.63
60.86
58.82
72.13
62.62
58.65
62.08
62.08
65.13
63.13
51.90
53.33
52.45
68.82
50. 6b
64.29
75.92
75.92
75.92
75.92
75.92
75.92
Upper 95% CL
for mean
56.53
66.54
65.51
73.99
68.90
54.76
77.29
63.05
72.06
54.33
77.72
59.66
53.91
66.97
65.08
78.06
68.06
64.93
69.19
69.19
73.62
70.25
59.42
60.87
57.90
75. ',0
58.16
71.83
80.00
80.00
80.00
80.00
80.00
80.00
-------
SECTION 6
AIRLESS CONVENTIONAL SPRAY EQUIPMENT
EQUIPMENT DESCRIPTION
In airless spray painting, the paint flows from an orifice
at high pressure and breaks up into spray as it enters the at-
mosphere. Typical paint line pressures are 6900 to 27600 kPa
(roughly 1000 to 4000 Ibs/in ). Airless spraying avoids the
problem of turbulence caused by compressed air, which sometimes
prevents proper deposition of the paint on the workpiece.
Airless spray guns will atomize paint and permit application
into corners and recessed interior areas without the blow
back experienced with air spraying.
Dirt or other small particles can obstruct the flow of
paint through the small orifice, which provides the atomization
in airless spray; therefore, special guns, pumps, hoses, etc.,
are required for airless spray. Use of airless spray eliminates
the need for a hose from the compressor to the spray gun (See
Figure 4).
Droplet sizes in airless spraying are larger than with
compressed air atomizing and consequently coatings applied by
airless spray are heavier and rougher. Airless painting is used
widely to apply zinc primers and other highly pigmented paints
and is especially useful for large objectn.
OPERATING AND MAINTENANCE FACTORS
Airless conventional spraying is an uncomplicated process
with few parameters involved. The paint is supplied at high
pressure to the gun from which it is expelled through a single
orifice. The orifice is designed to shape the spray. Orifices
are designated by the diameter and half-width of the laydown
at 25.4 cm (10 in) target distance. While plugging of one or
more holes in a conventional air spray cap is an operating and
maintenance problem, plugging of the single hole in an airless
cap, while possible, is such an obvious situation that the spray
gun operator always detects and corrects the problem before
proceeding.
Erosion of the orifice with continued use does present a
maintenance problem for gun operation. Obstructed paint supply
50
-------
COMPRESSOR
PAINT
LINE
Figure 4. Airless paint spraying system
51
-------
lines leading to reduced pressure at the gun is also a c; »cern.
Finally, the flow of air in the vicinity of the target is of
interest.
Test variables selected, then, were tip erosion, line plug-
ging, and varying booth air flow (See Table 20). The effect of
these variables on the spray painting operation were respectively
simulated by using orifices of progressively greater diameter,
by reducing the pressure of the paint at the gun, and by reduc-
ing the booth air flow. Tip erosion was tested at three levels;
unused tip with 0.28 mm (0.011 in) diameter, 0.33 mm (0.013 in)
diameter, and 0.38 mm (0.015 in) diameter orifices. Restricted
paint lines were tested at five levels: 9066.9 kPa (1300 Ibs/
in2), 8377.2 kPa (1200 Ihs/in2), n687 6kPa (11GO Ibs/in*),
6997.9 KPa (1000 Ibs/in ), and 6308.3 kPa (900 Ibs/in ).
Booth air rate- was simulated at two levels.
ALC equipment specifications for the test are included
in Appendix '3.
EXPERIMENTAL DESIGN
The experimental design for the airless conventional spray
is shown in Table 21. It is discussed
-------
TABLE 20. LEVELS OP OPERATING AND MAINTENANCE VARIABLES
TESTED ON ALC SPRAY PAINTING EQUIPMENT
Quant/ Ho. of
Factor qual. levels Test levels
B. Booth air rate* Quant. 2 1- 0.61ra/s (120 ft/min)
(linear velocity) -1- 0.36m/s (70 ft/rain)
C. Tip erosion*- Quant. 5t a» 0.28 im (.011 in.) cap
1» 0.28 mm (.011 in.) cap
0= 0.33 mm (.013 in.) cap
-1* 0.38 mm (.015 in.) cap
0.38 iim (.015 in.) cap
D. Restricted paint lines Quant. 5 a- 9J66.9 JcPa (1300 psig)z
1- 8377.2 JcPa (1200 psig)
0- 7687.6 kPa (1100 psig)
-1- 6997.9 kPa (1000 psig)
6308.3 kPa (900 psig)
DUMMY Qual. 3 n/a
•Actual booth air rates varied fran 100 to 140 ft/min for level "+1" and
50 to 90 ft/min for level "-1." Average air velocities are used in this
table.
•K>un cleanliness w\s interpreted as "tip erosion" for this experiment.
Progressively wider tip hole diameters were used to simulate tip wear.
tThe original experimental design called for five levels; in practice
we were orly able to simulate three levels.
zMeasured at gun downstream of all paint filters.
53
-------
TABLE 21. ATjC EXPERIMENTAL DESIGN
Run Number B
1
2
3
4
5
6
7
8
9
10
11 -1
12 1
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27 1
28 1
Variable
C D
-1
-1
1
-1
-1
1
-1
-1
1
1
-1
T
1
1
-1
1
-a
a
0
0
0
0
a
a
a
a
a
a
-1
•1
•1
1
•1
-1
1
.«
1
-1
1
1
-1
1
1
1
0
0
•a
a
0
0
a
a
a
a
a
a
0
C
0
0
Where:
B » Booth air rates—test at 2 levels- 1,-1
C - Tip erosion at 3 levels
D « Restricted paint lines—test at 5 levels: a,1,0,-1,-a
Dummy • Dummy variable not expected to affect TE
54
-------
TABLE 22. ORDRR OF PERFORMANCE OP ALC "t'EST RUNS
23
21
5
25
29
11
2
16
15
28
22
10
17
7
6
13
9
19
26
12
8
27
14
1
24
18
3
4
(38 runs)
55
-------
to reduce some of the planned runs without sacrificing
information. Runs 6, 11, 16, 23, 25, and 28 were dropped
from the design as shown in Table 21.
In the case of airless conventional, the experimental
design allowed for estimation of regression terms of the
following type:
o Linear in x.
o Linear and quadratic in x2
o Linear and quadratic in x3
TEST RESULTS
Tests were run and calculation:* performed in accordance
with the standard test method. Values of transfer efficiency
obtained during testing are shown in Table 23.
STATISTICAL ANALYSIS
Based on the TE test results, regression models were
developed to fit the data. Information on how to use these
regressions is presented in the AAC Statistical Analysis
section of this report.
Variables are named in the regressions that follow according
to the table below:
x.Bbooth air flow
x2=tip erosion
x3=restricted paint lines.
TE=transfer efficiency
Only those variables found to be significant have been
included in the final regression. Tip erosion, x., is a
qualitative factor and therefore has dummy variables associated
with it.
Flat Panel Target
The derived regression equation for ALC testing of flat
panel targets is:
TE = 74.4 - 5.47x2 - 1.94x2
56
-------
TABLE 23. ALC TEST RESULTS
Percent transfer efficiency
Run number FP VC Graco
1
2
3
4
5
7
8
9
10
12
13
14
15
17
18
19
20
21
22
24
26
27
76.6
76.6
63.1
79.5
79.2
77.6
75.9
68.3
66.0
69.4
64.8
6C 5
80.5
77.5
65.5
75.7
74.8
76.4
70.7
67.7
68.3
70.2
13.0
13.6
10.4
13.7
13.8
12.9
>3.2
11.2
10.4
10.6
10.8
10.5
14.8
13.2
10.8
12.7
13.1
13.3
12.5
10.6
11.0
10.9
33.7
33.1
28.4
33.0
33.6
33.4
33.0
26.9
25.9
27.6
27.5
27.5
35.0
34.1
26.7
32.2
34.5
32.4
34.4
27.3
27.3
27.9
57
-------
The only significant variable affecting TE is tip erosion.
The negative coefficient on x-d, implies that tip erosion
at level "1" makes TE go down! The positive sign on x,d1
means that the level "-1" makes TE increase.
The proportion of overall variance explained hy the
regression (R ) is 0.87. This is considered a high value,
indicative of a good fit of the regression. The standard
d.-viation of replicate FP test runs was 1.305, well within the
range of 2.0 specified in the test procedure.* The error due to
lack of fit was insignificant, with F = 2.16 (P = 0.36). (Refer
to Appendix G for a glossary of statistical terms.)
Vertical Cylinder Target
The regression analysis derived for ALC testing of VC
targets is:
TE = 12.90 - 1.40x2 - 0.78X2
Like the FP case, tip erosion was the only significant
variable found to affect TE for airless conventional spray
equipment. In this case the direction of the effect is simi-
larly contingent on selection of level (i.e. dj or dj).
The proportion of overall variance explained by the re-
gression model is 0.91. This indicates a good fit of the
regression.
The standard deviation of replicate VC test runs was
extremely small, at only 0.208. While this standard deviation
is admirable given the test procedure precision of 2.0, it
raises some question as to why the procedure is so repeatable
for this target configuration. The answer lies in the very
small overall standard deviation (only 1.4 across the entire
data set) created by intentional introduction of O&M variables.
The insensitivity of this system to intentional attempts to alter
TE demonstrates why the replicate standard deviation is so
small.
The error due to lack of fit was insignificant at the
0.05 level, with P= 7.76 (P= 0.12).
Graco Target
The regression analysis derived for ALC testing of Graco
targets is:
TE = 33.38 - 3.25x2 - 2.98Xj - 0.26x3
*CENTEC Corporation, "Development of Draft Standard Test Method
for Spray Painting Transfer Efficiency," for USEPA under Contract
68-03-1721, Task 2.
58
-------
Tip erosion is found significant only at the d- {"+1" level)
for this target configuration. This is an overwhelmingly large
effect, indicating that the effect on TB is very different at
this tip diameter than at other tip diameters for this system.
Restricted paint lines are also significant for this case, but
only marginally so. No interaction between factors is noted for
this system.
The proportion of overall variance explained by the regres-
sion (R ) is high at 0.95. This indicates a well fitting
model. The standard deviation of Graco target replicate test
runs is 0.346. Like the previous ALC cases, this extremely
low standard deviation is the result of the insensitivity of
this system to the intentional introduction of O&M factors.
The F statistics and associated probabilities are given in
Table 24 for each effect included in the regression.
TABLE 24. Air F-STATISnCS (P) AND ASSOCIATED PROBABILITIES (P)*
Flat Panel Vertical Cylinder
Effect P P P P
X2d1 11.39 .077 19.99 .0466
X2d2 89.56 0.00 255.33 .0044 383.63 .0000
x3 - - 16.59 .0002
Tables 25, 26, and 27 present a comparison of predicted
and observed transfer efficiency values, along with associated
significance limits, for the flat panel, vertical cylinder, and
Graco targets, respectively.
ALC CONCLUSIONS
Three O&M variables were selected for testing on ALC spray
painting equipment: tip erosion, booth air rate, and restricted
paint lines. In every test case, tip erosion is the overwhelming
variable affecting TB. The only other variable identified as signi-
ficant in any ALC test was restricted paint lines for the Graco
target.
The tremendous response to changes in tip diameter is
indicative of a very strong relationship between selection of
appropriate tip diameter and TE. Tip diameter should be careful-
ly selected. Table 24 shows that the "+1" level displays by far
the most significant effect for all three target types.
*Refer to Appendix G for glossary of statistical terms.
59
-------
TABLE 25. ALC-FP COMPARISON OF PREDICTED VERSUS ACTUAL TRANSFER EFFICIENCIES
Observation
number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
IS
16
17
18
19
20
21
22
Observed
value
76.60
76.60
63.10
79.50
79.20
77.60
75.90
68.30
66.00
69.40
64.80
66.60
80.50
77.50
65.50
75.70
74.80
76.40
70.70
67.70
68.30
70.20
Predicted
value
77.92
77.92
66.99
77.92
77.92
77.92
77.92
66.99
66.99
66.99
66.99
66.99
77.92
77.92
66.99
74.40
74.40
74.40
74.40
66.99
66.99
66.99
Residual
-1.32
-1.32
-3.89
1.57
1.27
-0.32
-2.02
1.31
-0.99
2.41
-2.19
-0.39
2.57
-0.42
-1.49
1.30
0.40
2.00
-3.70
0.71
1.31
3.21
Lower 95% CL
for mean
76.39
76.39
65.61
76.39
76.39
76.39
76.39
65.61
65.61
65.61
65.61
65.61
76.39
76.39
65.61
72.23
72.23
72.23
72.23
65.61
65.61
65.61
Upper 95% CL
for mean
79.45
79.45
68.36
79.45
79.45
79.45
79.45
68.36
68.36
68.36
68.36
68.36
79.45
79.45
68.36
76.56
76.56
76.56
76.56
68.36
68.36
68.36
-------
TABLE 26. ALC-VC COMPARISON OF PREDICTED VERSUS ACTUAL TRANSFER EFFICIENCIES
Observation
number
1
2
3
4
5
6
7
a
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Observed
value
13.00
13.60
10.40
13.70
13.80
12.90
13.20
11.20
10.40
10.60
10.80
10.50
14.80
13.20
10.80
12.70
13.10
13.30
12.50
10.60
11.00
10.90
Predicted
value
13.52
13.52
10.72
13.52
13.52
13.52
13.52
10.72
10.72
10.72
10.72
10.72
13.52
13.52
10.72
12.90
12.90
12.90
12.90
10.72
10.72
10.72
Residual
-0.52
0.07
-0.32
0.17
0.27
-0.62
-0.32
0.48
-0.32
-0.12
0.08
-0.22
1.27
-0.32
0.08
-0.20
0.20
0.40
-0.40
-0.12
0.28
0.18
Lower 95% CL
for mean
13.20
13.20
10.42
13.20
13.20
13.20
13.20
10.42
10.42
10.42
10.42
10.42
13.20
13.20
10.42
12.44
12.44
12.44
12.44
10.42
10.42
10.42
Upper 95% CL
for mean
13.84
13.84
11.01
13.84
13.84
13.64
13.64
11.01
11.01
11.01
11.01
11.01
13.64
13.84
11.01
13.35
13.35
13.35
13.35
11.01
11.01
11.01
-------
TABLE 27. ALC-GRACO COMPARISON OF PREDICTED VERSUS ACTUAL TRANSFER EFFICIENCIES
o\
K)
Observation
number
1
2
3
4
5
6
7
3
9
10
11
12
13
14
15
16
17
IB
19
20
21
22
Observed
value
33.70
33.10
28.40
33.00
33.60
33.40
33.00
26.90
25. 9C
27.60
27.50
27.50
35.00
34.10
26.70
32.20
34.50
32.40
34.40
27.30
27.30
27.90
Predicted
value
33.38
33.38
26.88
33.90
33.38
33.90
33.33
27.40
26.88
27.40
26.88
27.40
33.90
33.64
27.14
32.85
33.89
33.37
33.37
27.66
27.66
27.66
Residual
0.31
-0.28
1.51
-0.90
0.21
-0.50
-0.38
0.50
-0.98
0.19
0.61
0.09
1.09
0.45
-0.44
-0.65
0.60
-0.97
1.02
-0.36
-0.36
0.23
lower 95% CL
for mean
32.77
32.77
26.20
33.25
32.77
33.25
32.77
26.89
26.20
26.89
26.20
26.89
33.25
33.08
26.61
31.87
32.91
32.58
32.58
27.02
27.02
27.02
Upper 95% CL
for mean
33.99
33.99
27.56
34.55
33.99
34.55
33.99
27.91
27.56
27.91
27.56
27.91
34.55
34.20
27.67
33.83
34.87
34.16
34.16
28.30
28.30
28.30
-------
Prom the data generated during this test program, very
little can be said about the effects of other variables on TE
for ALC spray systems. The response of TE to tip erosion is
so dramatic that it may obscure other potentially important
variables.
63
-------
SECTION 7
AIRLESS ELECTROSTATIC SPRAY EQUIPMENT
EQUIPMENT DESCRIPTION
A Graco Mo<'sl AL-4000 was selected as the ALE spray equip-
ment for TE testing. The AL-4000 is operated like conventional
airless spray equipment except the spray is electrically charg-
ed. The electrical charge is an attractive agent pulling the
paint towards the nearest ground, the target. Electrical power
is supplied at a controlled voltage on the electrode at the gun
tip. Fluid flows through the gun at high pressure and is
atomized through a carbide tip. The atomized paint picks up an
electrical charge as it is sprayed past the charged electrode.
The spray pattern of ALE equipment is determined primarily by
tip orifice size. Fluid flow cannot be adjusted at the gun
(as it can in conventional and conventional electrostatic
equipment); it is either full on or full off.
Graco standard black enamel was used as the test paint. It
was cut to 25.5 seconds on a Shell #3 cup at 25*C. A 16 L
(4 gal) batch of paint was mixed and stored in a 20 L (5 gal)
Graco paint pressure pot. This batch was not enough to complete
all ALB testing and was made up on the second and third days of
testing. The paint was kept in a constant temperatjre booth
along with the paint pump (Model 207-707, K3D 30:1), stirrer,
viscosity measurement equipment, and som* supply lines. All
paint supply lines were insulated.
ALE tests were conducted in the Dynaprecipitator water wash
booth described in Section 4. Booth characteristics were
identical to earlier test runs, with only two air speeds avail-
able.
Foil weights were determined on Precise laboratory scales
accurate to 0.01 g. Height-percent-solids samples weie weighed
on 0.0001 g accuracy scales.
A forced-air, gas-fired oven was used for curing weight-
percent-solids samples and TE samples. The oven was cleaned
daily to prevent contaminants from adhering to the samples. All
samples were cured at 171.1"C (340*F) for 20 minutes. This
is a more severe cure than for previous experiments. It was
54
-------
instituted to ensure a complete cure for the heavier laydown of
paint expected for this equipment type. Trial and error weight-
percent solids determinations were made to document the point of
assured complete curing.
The mass flow meter described in Section 3 was used for
paint ma»s flow determinations. The test method presented
in Appendix A was strictly adhered to for ALB testing, as were
the QA/QC requirements of the test.
ALB equipment specifications for this test series are
included in Appendix P.
OPERATING AND MAINTENANCE FACTORS
Variables had been previously identified through interviews
and literature search that were considered to have an important
effect in achieving optimum TE for ALE equipment. These 14 var-
iables are presented in Table 28. Five variables were selected
for ALE testing on the basis of the number of times it was iden-
tified for ALE by different sources.- the ability to simulate the
variable within the prescribed test methodology, and finally,
the limitation of laboratory time. The five selected test
variables were:
o Booth air rate (linear velocity)
o Tip erosion (substituted for gun cleanliness)
o Restricted paint lines
o Voltage
o Electrode position
A dummy variable was also included to provide a measure of
th«» inherent error in the experiment.
EXPERIMENTAL DESIGN
An experimental design was developed to accommodate the
limitations of testing while addressing the effects of each
variable as completely as possible.
As before, the first restraint on experimental design
was the availability of laboratory time: only about 30 test
runs could be reasonably expected during a week of tescing.
The second limitation was the number and type of simulation
levels for each variable. Only two booth air rates were
possible in the test laboratory, while three levels of fan
-------
TABLE 28. OPERATING AND MAINTENANCE VARIABLES
FOR ALE SPRAY EQUIPMENT*
Booth air rate
Booth configuration
Cure schedule (time, temperature)
Paint discharge technique
Equipment design
Flash off
Gun cleanliness
Gun condition
Gun-to-target distance
Operator error
Paint mass flow rate
Paint characteristics
Restricted paint supply
Target configuration
*as mentioned by industry sources contacted
66
-------
air were achievable, and ifive or more levels of some other
variable could be simulated. Table 29 presents the type of
variable (quantitative/qualitative) and levels to be accom-
modated in the experimental design.
Table 29 shows the use of a dummy variable. This variable
represents the effect of a totally unrelated action on TE. If
the data analysis shows any significant effect for the dummy
variable it is indicative of some type of problem with the test
method or test performance.
Table 30 presents the ALE experimental design. In this
figure, the abbreviations "a," "1," "0," "-I," and "-a" denote
the level of each variable to be tested. Level "a" denotes the
base level with a good spray pattern. Level "-a" denotes the
poorest level of a factor to be tested. The intermediate levels
"1," "0," and "-1" are determined along equal spacing from "a"
to "-a" for the particular variable. Variable levels for ALE
testing were determined in pretest trials as described in the
following subsection.
The first 16 test runs in the experimental design are the
fractional factorial portion of the design. When the results of
several variables are to be studied, a factorial design is
usually the nost efficient method to use.* The basic idea o£
factorial design is to alter several aspects of a test at a
time, but in such a way that the effects of individual alter-
ations can bos determined. Fractional factorial designs sacri-
fice some ability to test for interaction between variables but
are able to test for main effects very efficiently.
Runs 17 through 26 in Table 30, are the "star" portion of
the experimental design. This portion o: the experiment tests
the eftects of variables at the extremes of their range (for the
system under test, at "-a" and "a"). The star design broadens
the range of information gathered in the test. The star portion
of the design allows extra degrees of freedom in order to assess
lack of fit.
As in the case of previous designs, the design used hare
entails a central composite design for variables C, D, P, «md G.
However, G contains only 3 levels while C, D, and ? contain 5
levels. As a result, the design is not a standard central
composite design.
*Youden, W. J. and Steiner, E. H., Statistical Manual of the
Association of Official Analytical Chemists, Arlington, Va.,
1982; and Davies, O. I.. , Design and AnalyaTa of Industrial
Experiments, Great Britain, 1979.
esig
7T9
67
-------
TABLE 29. EXPERIMENTAL VARIABLES SELECTED FOR
TESTING ALF. SPRAY EQUIPMENT
Factor
ID Factor description
Quant./
qual.
No. of
test levels
B Booth air rate (linear
velocity) Quant.
C Gun cleanliness (tip erosion) Quant.
D Restricted paint lines Quant.
P Voltage Quant.
G Electrode position Qual.
Dummy Dummy action or variable Cual.
2
5
5
5
3
2
68
-------
TABLE 30. ALE EXPERIMENTAL DESIGN
Run number B C
1 -1 -1
2 1
3 -1
4 -1
5 1
6 1
7 -1
8 1
9 -1
10 1
11 1
12 -1
13 -1
14 1 -1
15 1 1
16 -1 1
17 1 -a
18 1 a
19-0
20-0
21 0
22 0
23-0
24-0
25 0
26 0
27 a
28 a
29 a
30 a
31 a
32 a
Variable
0
0
-a
a
0
0
0
0
0
0
a
a
a
a
a
-1
-1
-1
1
-1
1
1
1
1
1
-1
1
-1
-1
1
-1
0
0
0
0
0
0
-a
a
0
0
a
a
a
a
a
a
_G
-1
1
1
1
0
0
0
0
0
0
0
0
-1
-1
-1
1
-1
1
-1
1
1
1
-1
1
-1
1
-1
1
-1
0
0
0
C
~ 1
1
0
0
0
0
1
1
1
1
1
1
Where:
B = Booth air rates—test at 2 levels: 1,-1
C » Tip Erosion—teat at 5 levels: a,1,C,-1,-a
D = Restricted paint lines—test at 5 levels:
a,1,0f-1,-a
F =• Voltage—test at 5 levels: a,1,0,-1,-«;
G * Electrode position—test at 3 levels: 1,0,-1
Dummy » Dummy variable not expected to impact TE
b9
-------
The last six runs of the test design are replicates.
Replicates are provided at the base condition of all variables
to provide a measure of the precision of the test.
ALE TEST PERFORMANCE
ALE testing began February 27, 1984. Equipment set up,
target assembly and hanging, foil cutting and preweighing, and
other preparatory tasks were completed earlier in the week. The
paint was adjusted to desired specifications in a 20 L (5 gal)
paint pot. Once the paint was adjusted, all equipment and lines
were checked for proper installation and freedom from obstruc-
tion. The mass flow meter was calibrated and zeroed. Mass flow
calibration was double checked against anatomized paint capture
and found to be within 0.4 percent of the meter reading, as
required.
Base level ("a") for each variable was determined by
setting the equipment according to Graco experience with the
test paint and spray painting system. Some adjustments were
necessary to provide a good spray pattern without excessive
laydown. Final base levels were confirmed by a visual spray
pattern check. Base levels thus determined are shown in
Table 31.
Deteriorated levels were selected by setting all factors
except the subject variable (for each variable in turn) at
the base level. The subject variable was changed until a sig-
nificantly worse spray pattern could be discerned. Th-.» spray
pattern was checked by spraying onto a paper target for 5 to 6
seconds, and observing the resulting pattern. Deteriorated
variable levels ("-a") thus determined are shown in Table 31.
Intermediate levels were calculated to be evenly spaced
from the base level ("a") *:o the deteriorated level ("-a").
Intermediate levels are also shown in Table 31.
Electrode position was similarly defined. Base level was
with the electrode in normal position. Deteriorated level
("-a") was selected with the electrode clipped off. An inter-
mediate level was decided as a bent electrode. All electrode
position levels are shown in Figure 5. (Tip orientation was
vertical for actual testing.)
Although gun cleanliness had been selected as an ALE
experimental variable, a partially blocked tip could not be
simulated. Any blockage affixed to the tip was blown out by
the pressure of the paint during spraying.
70
-------
TABLE 31. LEVELS OF OPERATING AMD MAINTENANCE VARIABLES
TESTED ON ALE SPRAY PAINTING EQUIPMENT
Factor
Quant/ No. of
qual. levels
Test levels
B. Booth air rate*
Quant.
1= 0.61m/s (120 ft/min)
-1= 0.36m/s (70 ft/min)
C. Tip erosiont
Quant.
a= 0.28 inn (.011 in.) cap
1= 0.28 mm (.011 in.) cap
0= 0.33 mn (.013 in.) cap
-1= 0.38 mm (.015 in.) cap
0.38 mm (.015 in.) cap
D. Restricted paint lines Quant.
a= 9066.9 kPa (1300 psig)z
1= 8032.4 kPa (1150 p*,ig)
0= 6997.9 kPa (1000 psig)
-1= 5953.4 XPa (850 psig)
-a= 4929.0 kPa (700 psig)
F. Voltage*
Quant.
G. Electrode position
Qual.
a= 72 kV
1= 63 kV
0= 54 kV
1= 45 kV
36 kV
1= normal
0= bent ++
-1= clipped off tt
*Actual booth air rates varied frcm 100 to 140 fpm for level "+1" and
50 to 90 fpm for level B-1." Average air velocities are used in this
table.
1Gun cleaniness was interpreted as "worn tip" for this experiment.
Progressively wider .±p hole diameters were used to simulate tip
wear.
+The original experimental design called for 5 test levels; in
practice we were only able to simulate 3 levels.
zMeasured at gun downstream of all paint filters.
ttKonitored at power supply.
•Hfient as shown in Figure 5.
ttElectrode cut off at plane of cap.
71
-------
FRONT
SIDE
Level +a and +1: Normal electrode position
SIDE
Level 0: Bent electrode
FRONT
SIDE
Level -a and -1: Electrode cut off
Figure 5. Airless electrostatic air cap showing
test levels for electrode position
72
-------
To salvage the variable, it was decided to look at another
identified variable instead. The only ether tip factor identified
for ALB equipment was tip erosion. Abrasive paints can erode
the tip orifice after prolonged use. To simulate tip erosion,
tips at a variety of diameters were obtained and checked for
spray pattern. Only three tips gave acceptable spray patterns,
at 0.28 mm, 0.33 mm, and 0.38 mm diameters. With larger tips,
the paint laydown was too heavy to avoid running; smaller tips
were not avialable.
As in previous experiments, the booth air rate could only
be controlled to two levels. Voltage and restricted paint lines
were each simulated at five levels, as shown in Table 31.
Variaole level selection was completed on February 27, 1984.
ALE test runs were started on February 28, 1984. Weight-percent-
sol.xds samples were taken. The results were in close agreement,
and testing began in randomized order according to the test plan.
Nine test runs were completed the first day of ALE testing.
One run was thrown out due to a timer malfunction; one run was
deleted because the booth water wash was not on; and one run was
eliminated because grounding wires had not been attached to the
foils (even though the flat panel was grounded).
Paint had to te added and adjusted for the second day of
testing. All preparatory steps were taken, but on the first run
the mass flow meter totalizer stuck. Mass flow measurements
were lost, and the run was repeated immediately after repair of
the malfunctioning switch. After eight runs were completed, the
laboratory experienced a 2-1/2-hour power failure. When power
was restored, all start-of-test QA/QC measures were repeated
before resuming testing. Final weight-percent-solids determina-
tions were made after 15 runs were completed. The morning and
evening weight-percent-solids determinations agreed nicely, but
the power failure sample was several weight-percent higher.
The power failure sample had not been stirred during the power
failure, and probably was not adequately stirred prior to sampl-
ing. This weight percent solids was omitted from TE calculations
as a suspect sample.
Paint was added and viscosity adjusted for the final day of
ALE testing. All prescribed preparatory steps were taken ac-
cording to the test plan (Appendix A) and QA/QC plan (Appendix B).
The rest of the experiment was completed without incident on
March 1, 1984.
TEST RESULTS
Tests were run and calculations performed
-------
TABLE 32. ALE TEST RESULTS
Percent transfer efficiency
Run number FP VC Graco
91.5 47.5 83.6
* 87.8 43.2 64.6
3 93.9 42.2 61.0
4 89.8 68.8 77.6
5 83.3 44.1 58.0
6 93.3 65.1 77.2
7 96.1 76.2 74.5
8 90.2 60.2 70.8
9 90.9 69.4 76.3
10 93.7 48.2 64.3
11 94.8 60.0 70.0
12 87.6 68.0 71.9
13 90.8 66.1 71.9
14 84.7 48.0 63.7
15 90.6 71.9 74.2
16 88.5 66.4 72.1
17 91.8 53.3 63.7
18 89.6 56.7 67.7
19 88.6 66.2 75.1
20 91.9 57.0 68.3
21 90.5 55.3 68.3
22 84.3 51.0 65.8
23 89.2 44.1 59.4
24 93.2 74.9 76.2
25 88.3 47.3 62.5
26 88.8 65.9 72.3
27 91.7 78.6 78.2
28 92.5 77.8 79.1
29 89.1 73.3 74.2, 76.
30 90.4 75.9 79.3
31 92.6 80.8 78.7
32 93.4 76.6 76.8
33 78.2*
*An extra replicate using only the Graco targets was made for
Grace's own purposes. The data is included here for com-
pleteness.
74
-------
STATISTICAL ANALYSIS
The terminology shown below is used in the regressions
that follow:
x.=booth air rate (linear velocity)
x.=tip erosion
x3=restricted paint lines
x.=voltage
x-=electrode position
TE=transfer efficiency
In the case of the airless electrostatic, the following
linear, quadratic, and interaction effects were chosen for
the regression model.
o Linar in x.
o Linear and quadratic in x-
o Linear and quadratic in x.
o Linear and quadratic in x.
o Dummy variables in x_
o Interaction between x. and x.
o Interaction between x_ and x.
o Interaction between x- and x^
o Interaction between x. and x.
A discussion of how to use the regression equations is
presented in the AAC Statistical Analysis section of this
repo.rt. The derived regression for each target type follows.
Flat Panel Target
TE » 90.30 - 1.12x2 + 1.37x4 - 0.68x^3
Only linear effects are significant for ALE testing of flat
panel targets. Tip erosion, x2, has a negative effect on TE.
The positive coefficient
-------
The proportion of overall variance explained by the re-
gression (R ) is 0.28. This is the poorest case of all equip-
ment types and target configurations tested. The raw data was
reviewed to locate any test errors contributing to this un-
usually low R , but no experimental source was found. The low
R may be attributed to the low overall variation in this test
series. The variation of TE over all of the experimental com-
binations was only 3.0. This value is barely above the standard
deviation of the test procedure (2.C). The lack of variance
demonstrates the insensitivity of the system to O&M factors. The
standard deviation of replicated test runs was 2.287, high and
not quite within the specified range of the test procedure. The
error due to lack of fit is insignificant, with P » 0.38.
Vertical Cylinder Target
TE = 57.31 - 3.77x1 + 2.85x2 - 1.55x3
+ 7.02x, - 4.66x_d1 + 8.62x,.d_
4 51 3 I
More than twice as many variables are significant for VC
testing than were identified for FP testing. The negative
coefficient on x. indicates that TE decreases with increasing
booth air rates. The positive coefficient or. tip erosion indi-
cates that as tip diameter decreases, TE increases. But neither
the booth air rate nor the tip erosion trends are constant
because of the interaction between the two. The coefficient of
the interaction is negative. Thus the negative effect of booth
air is enhanced at the high level of tip erosion but is moderated
at the low level of tip erosion. TE is adversely affected by
increasing restrictions in the paint lines (x3). However this
negative trend is not constant, due to the interaction with tip
erosion. Increasing voltage ..ends to increase TE, dramatically.
The magnitude and direction of the effects of different levels of
electrode position changes with the selection of electrode
position. Figure 5 shows the various test levels for electrode
position.
Two interactions are significant for this case, tip erosion
with booth air and tip erosion with restricted paint lines.
Each effect acts in a different direction. Nevertheless, it is
clear that tip erosion is the overwhelming factor for this
case.
The proportion of overall variance explained by the regres-
sion (R ) is 0.95, a respectable value. The standard deviation
due to repeats is 2.55, slightly over the 2.0 value specified
in the test procedure. The error due to lack of fit is insigni-
ficant, with F= 2.16 (P- 0.20).
-------
Graco Target
TE - 69.05 - 2.83x1 -
- G.84xJ + 4
The Graco target configuration identified the most signifi-
cant O&H variables for ALE testing. Like the other rases, in
creasing booth air rates (x. ) causes a drop in TE. Restricted
paint lines (x.) also cause a drop in TEr while increasing
voltage (x.) raises TE linearly and causes it to slightly drop
quadratically. Electrode position is significant at tne d2
("+1") level only. At this level TE is increasing with changes
in electrode position. Apart from the linear effects, inter-
actions between x. and x_, x. and x_, and x. and x, complicate
the system. The linear trends described above are distinct but
the rates of change of TE with respect to x. , x-, and x.
are not constant. As an example, the booth air rate effect is
negative but is moderated at the high level of x_, restricted
paint lines.
The proportion of overall variance explained by the regres-
sion (R > is a modest O.B3. The standard deviation of rep-
licate test runs on Graco targets was 1.274, well within the
2.0 specified by the test procedure. The error due to lack
of fit was insignificant, with F» 3.04 (P=» 0,14). Table 33
gives the value of the P-statistic and associated probability
for each effect of significance.
Tables 34, 35, and 36 presents compariaona of predicted
and observed values of transfer efficiency for each experimental
condition for the flat ptnel, vertical cylinder, and Graco
targets respectively.
ALE CONCLUSIONS
ALE test results sho^-d the most difference in discrimi-
nation among target C'.v.cigurations. Only two variables were
identified as significant for the flat panel target (tip
erosion and voltage), and these were only marginally signifi-
cant. The Graco and vertical cylinder targets identified
four and five significant variables, respectively, not includir.g
several interactions between v& -tables.
Voltage, booth air rate, restricted paint lines, and
electrode position were significant factors for Graco and VC
targets types. These results are consistent with findings from
AAE and ALC experiments: where electrostatic forces are involv-
ed, voltage, booth air rate, and electrode position are impor-
tant to establishing optimum TE. The higher the voltage and the
77
-------
TABLE 33. ALE F-STAT1STICS (P) AND ASSOCIATED
PROBABILITIES (P)
Effect
X1
x2
X3
X4
S5d1
X5d2
X1X2
X1X3
X2X3
X3X4
x2
Flat Panel
F P
6.64 .05
-
4.85 .07
-
-
-
6.42 .05
-
-
- -
Vertical
F
54.31
22.94
9.53
191.31
10.09
30.86
8.61
-
7.00
-
-
Cylinder
P
.0007
.0049
.0272
.0000
.0246
.0026
.0325
-
.0457
-
_
Gzaco
F P
117.16 .
-
111.94 .
166.26 .
-
72.09 .
7.67 .
-
30.94 .
32.82 .
10.30 .
0000
-
0000
0000
-
0000
0006
-
0000
0000
OC01
78
-------
TABLE 34. ALE-FP COMPARISON OF PREDICTED VERSUS ACTUAL TRANSFER EFFICIENCIES
-j
so
Observation
number
1
2
3
4
5
6
7
8
9
10
It
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Observed
value
91.50
87.80
93.90
89.80
83.30
93.30
96.10
90.20
90.90
93.70
91.80
87.80
90.80
84.70
90.60
63.50
91.80
89.60
88.60
91.90
90.50
84.80
89.20
93.20
88.30
88.80
91.70
92.50
87.10
90.40
92.60
93.40
Pradicted
value
69.39
88.45
90.69
*2.13
87.16
91.19
93.42
93.42
89.89
92. 1 3
90.69
91.19
87.16
89.39
89.89
88.45
91.41
89.17
89.00
91.59
90.29
90.29
87.55
93.03
90.29
90.29
90.62
90.62
90.62
90.62
30.62
90.62
Residual
2.10
-0.65
3.20
-2.33
-3.86
2.10
2.67
-3.22
1.00
1.56
4.10
-3.59
3.63
-4.69
0.70
0.04
0.38
0.42
-0.40
0.30
0.20
-5.99
1.64
0.16
-1.99
-1.49
1.07
1.87
-3.52
-0.22
1.97
2.77
Lower 95% CL
for mean
87.08
06.20
88.78
90.08
84.94
88.98
91.12
91.12
38.52
90.08
88.73
88.98
84.94
87.08
88.52
86.20
89.73
87.63
86.96
89.27
P9.25
8*.?5
85.2
9t. o::
89.25
89.25
88.72
88.72
88.72
86.72
88.72
88.72
Upper 95% CL
for mean
91.70
90.70
92.49
94.16
89.37
93.40
95.73
95.73
91.27
94.18
92.59
93.40
89.37
91.70
91.27
90.70
93.09
90.72
91.01
93.90
91.33
91.33
89.86
95.04
91.33
91.33
92.51
92.51
92.51
92.51
92.51
92.51
-------
TABLE 35. ALE-VC COMPARISON OP PREDICTED VERSUS ACTUAL TRANSFER EFFICIENCIES
oo
o
Observation
number
1
2
3
4
5
G
7
8
9
10
11
12
13
14
15
16
1?
18
19
20
21
22
23
24
25
26
27
28
29
30
31
J2
Observed
value
47. M)
43.20
42.20
68.80
44.10
65.10
76.20
60.20
69.40
48.20
60.00
68.00
68.10
48.00
71.90
66.40
53.30
56.70
66.10
57.00
55.30
51.00
44.10
74.90
47.30
65.90
78.60
77.80
73. 3C
75.90
80.80
76.60
Predicted
value
47.86
42.90
44.63
75.18
42.93
70.22
68.95
57.95
68.07
51.73
57.20
68.10
67.32
50.98
70.25
67.34
52.49
54.60
64.17
57.98
53.54
53.54
47.04
75.11
48.88
62.16
77.28
77.28
77.28
77.28
77.28
77.28
Residual
-0.36
0.29
0.56
-6.38
1.16
-5.12
7.24
2.24
1.32
-3.53
2.79
-0.10
0.77
-2.98
1.64
-0.94
0.80
2.09
2.32
-0.98
1.75
-2.54
-2.94
-0.21
-1.58
3.73
1.31
0.51
-3.98
-1.38
3.51
-0.68
Lower 95% CL
for mean
43.52
38.90
37.40
71.01
39.15
66.15
64.65
53.82
63.83
47.58
53.05
64.07
63.20
46.86
68.14
63.18
49.09
51.29
63.58
54.25
50.85
50.85
43.45
71.28
46.38
59.84
74.80
74.80
74.60
74.80
74.8J
74.80
Upper 95% CL
for mean
52.20
46.91
45.87
79.34
46.72
74.30
73.26
62.09
72.30
55.88
61.36
72.12
71.43
55.09
72.37
71.51
55.88
57.90
67.76
61.71
56.23
56.23
50.63
78.84
51.38
64.49
79.77
79.77
79.77
79.77
79.77
79.77
-------
TABLE 36. ALE-GRACO COMPARISON OF PREDICTED VERSUS ACTUAL TRANSFER EFFICIENCIES
Observation
number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Observed
value
83,60
64.60
61.00
77.60
53.00
77.20
74.50
70.80
76.30
64.30
70.00
71.90
71.90
63.70
74.20
72.10
63.70
63.70
75.10
68.30
68.30
65.80
59.40
76.20
62.50
72.30
78.20
79.10
76.20
79.30
78.70
76.80
Predicted
value
74.27
65.34
62.99
82.74
60.59
73.81
76.75
70.77
73.17
64.77
71.71
73.71
74.11
60.43
74.35
69.36
65.41
67.01
77.25
66.51
66.21
66.21
62.09
74.91
66.21
70.92
77.93
77.93
77.93
77.93
77.93
77.93
Residual
9.32
-0.74
-1.99
-5.14
-2.59
3.38
-2.25
0.02
3.12
-0.47
-1.71
-1.81
-2.21
3.26
-0.15
2.73
-1.71
0.68
-2.15
1.78
2.08
-0.41
-2.69
1.28
-3.71
1.37
0.26
1.16
-1.73
1.36
0.76
-1.13
Lower 95* CL
for mean
70.36
61.42
59.39
78.48
56.65
69.63
72.77
66.84
69.83
61.08
67.78
70.08
69.85
56.32
71.81
65.14
62.81
64.37
73.78
62.65
64.08
64.08
57.16
70.44
64.08
68.35
75.35
75.35
75.35
75.35
7£.35
75.35
Upper 95% a,
for mean
78.18
69.25
66.59
87.01
64.53
77.99
60.73
74.70
76.51
68.47
75.63
77.33
78.37
64.53
76.89
73.58
68.01
69.66
80.72
70.36
68.35
68.35
67.02
79.38
68.35
73.48
80.50
80.50
80.50
80.50
80.50
80.50
-------
lower the booth air rate, the better TE is likey to be. Thus,
ALB spray painting equipment should be maintained to supply the
maximum allowable voltage to the tip. Periodic checks of power
supply are recommended to assure tip voltage remains at the
desired level. Booth air rate should be kept to the lowest
level acceptable for safety, environment, and worker comfort.
The effect of the position of the electrode in the atomized
paint field is less clear, appearing significant in some cases
and insignificant in other similar cases. It seems prudent,
however, to maintain the electrode position well into the atom-
ized paint field. Trimming the electrode is not recommended.
Restricted paint lines have a significant effect on TE for
Graco and VC target types. This is a shared phenomenon with
other equipment types. Pressure of the paint supply to the
spray gun should be monitored to avoid degeneration through
clogging or other restrictions. If the paint pressure is not
monitored, trie operator may notice a loss of spray quality, but
he is likely to take an inappropriate action to remedy the
problem. This situation is espei:'.ally true for air-atomized
spray systems where the operator may adjust the fan air or the
atomizing air to counteract the effects of lower paint pressure.
It is equally applicable for ALE spray systems.
82
-------
SECTION 8
COMPARISON OP TARGETS
BACKGROUND
The Draft Standard Test Method (Appendix A) specifier two
sets of targets for spray painting in each test run. These
targets are described in detail in Appendix A. The test targets
consisted of a set of foil-covered aluminum vertical cylinders
(VC) mounted in certain positions inside a wooden frame, and a
set of foil strips mounted at certain spacing on a large flat
stainless steel panel (PP). Both targets were suspended from an
overhead conveyor for the test. The VC targets were designed to
be somewhat representative of smaller, more open <*nd intricate
substrates. The PP targets were designed to be representative
of large, relatively flat and closed substrates. The test
results from a single transfer efficiency determination include
a VC result and a PP result. These results have quite different
values.
During the test program at Graco, a third set of targets
(called Graco targets) were painted at the same conditions as
the Draft Standard Test Method targets. These targets consisted
of a set of ten 15.24 cm (6 in) wide metal panels mounted
15.24 cm (6 in) apart, and hanging 121.92 cm (48 in) long. The
TE results from the center six cargets were averaged to obtain a
single TE value. The TE value obtained for this uarget type was
different from the values obtained for VC or PP targets.
This chapter evaluates the transfer efficierry character-
istics of all three target types for four equipment types to
determine if any of the designs has special advantages over
other targets for future testing.
COMPARISON OP FACTORS IDENTIFIED AS SIGNIFICANT
Table 37 presents a summary of the variables identified as
significant for each target type and each equipment type. The
Graco target configuration was the most sensitive, identifying
23 significant O&M variables (cr interactions) over all equip-
ment types. VC targets carr.e in a close second by identifying
19 significant variables, followed by FP targets at only 13
significant variables.
83
-------
TABLE 37. COMPARISON OF SIGNIFICANT FACTORS IDENTIFIED
BY THREE TARGET CONFIGURATIONS
Eauipment
type
ALE
R squared
ALC
R squared
AAE
R squared
AAC
VC
Booth air
Tip Eros.
Paint lines
Voltage *
Elect, pos.
Booth air x tip
Tip x paint lines
0.95
Tip eros. *
0.91
Air lines
Booth air
Gun cleanliness
Paint lines
Fan air
Voltaqe *
Electrode pos.
Booth air x fan air
0.92
Air lines *
(Booth air-close)
Paint lines
Fan air
Target con-
figuration
Graco
Booth air *
Paint lines *
Voltage *
Elect, pos. *
Booth air x tip
Tip x paint lines
Paint lines x volt.
0.83
Tip eros. *
Paint lines
0.95
Air lines
Booth air
Gun cleanliness
Paint lines
Fan air
Voltage *
Electrode pos.
0.94
Air lines *
Booth air
Gun cleanliness
Paint lines
Fan air *
FP
Tip eros. (marg)
Voltage (marg)
0.28
Tip eros. *
0.87
Air lines *
Paint lines
Voltage
Electrode pos.
0.67
Air lines *
Booth air
Paint lines *
Fan air *
air lines x booth air
booth air x
fan
R squared
0.79
0.96
0.99
* Strong response, overriding factor influencing TE
(marg) Marginally significant response
84
-------
The variables identified by the Graco targets match up
fairly consistently with those identified by VC's; FP targets
presented some anomolies.
Graco targets had the highest correlation coefficients,
averaging 0.92, followed closely by VC at 0.89 and FP at 0.70.
WORTH ASSESSMENT OF THREE TARGET CONFIGURATIONS
A Worth Assessment Model* was constructed to evaluate the
relative merits of each target type. Six criteria were selected
for this evaluation:
1. High correlation coefficient (ability to fit
mathematical models)
2. Target discrimination (ability to identify
significant effects)
3. Ease of fabricating the target
4. Ease of transporting/storing the target
5. Ease of use during testing
6. Target cost
Each of the targets was ranked from 0 (low) to 1 (high) for
these criteria. The rank was multiplied by weighting factors
and summed to generate a score. Several different weighting
factor combinations were used in calculations to compare the
effects en the final score.
In every case, the Graco target configuration scored high-
est. The Graco target scored consistently higher in almost all
categories than VC or FP targets. The Graco targets were easier
to handle, provided the best sensitivity to significant factors,
and demonstrated a very good correlation coefficient.
The worth assessment scores for evenly weighted criteria
were:
Graco 0.79
VC 0.50
FP 0,50
This case is the closest competition between target types
Detailed computer printouts of this analysis, is shown in Table
38.
*CENTEC Corporation, ""Worth Assessment Model," computer
software, Copyright 1979.
85
-------
TABLE 38. WORTH ASSESSMENT MODEL COMPARING TARGET CONFIGURATIONS
(1) VC target
Factor name
Ranking Selection description
Weight
Value
1 Correlation coefficient 1.0
2 Target TE discrimination 0.8
3 Ease of fabrication 0.3
4 Ease of transport/storage 0.5
5 Ease of use during test 0.0
6 Cost of target 0.5
Very high R squared
Identifies many factors
Difficult to make
Moderately difficult to T*S
Difficult to use and handle
Moderate cost
0.170
0.170
0.170
0.170
0.170
0.150
Score
0.17000
0.12750
0.04250
0.08500
0.07500
0.07500
0.50000
(2) Graco target
Factor name
Ranking Selection description
Weight
Value
00
Oi
1 Correlation coefficient 0.8
2 Target TE discrimination 1.0
3 Ease of fabrication 0.8
4 Ease of transport/storage 0.8
5 Ease of use during test 0.8
6 Cost of target 0.8
High R squared
Identifies most factors
Fairly easy to fabricate
Fairly easy to trans. & store
Fairly easy to use and handle
Relatively inexpensive
,170
,170
,170
,170
,170
,150
Score
0.12750
0.17000
0.12750
0.12750
0.12750
0.11250
0.79250
(3) FP target
Factor name
Ranking Selection description
Weight
Value
1 Correlation coefficient 0.8
2 Target TE discrimination 0.3
3 Ease of fabrication 1.0
4 Ease of transport/storage 0.3
5 Ease of use during test 0.3
6 Cost of target 0.5
High R squared 0.170 0.12750
Identifies a few factors 0.170 0.04250
Easy to fabricate 0.170 0.17000
Difficult to trans. & store 0.170 0.04250
Very inconvenient to use 0.170 0.04250
Moderate cost 0.150 0.07500
Score
0.50000
-------
APPENDIX A
DRAFT STANDARD METHOD FOR SPRAY PAINTING
TRANSFER EFFICIENCY OPERATIONS AND MAINTENANCE TESTING*
1. SCOPE
1.1 This method covers testing to determine the effects of
certain operating and maintenance factors on transfer
efficiency. Four types of spray equipment, air
atomized conventional (AAC), airless conventional
(ALC), air atomized electrostatic (AAE), and airless
electrostatic (ALE) are to be tested.
1.2 The factors selected for testing and the levels of each
factor to be tested are summarized in the experimental
design matrix for each type of spray equipment (Subtask
Re. ort, Tables 5, 6, 7, and 8).
1.3 This test program is estimated to take 4-5 weeks of
laboratory work.
1.4 This method is applicable only to solvent or wafer-
borne coatings applied in a single pass. The same
coating shall be used for all tests in this program.
2. APPLICABLE DOCUMENTS
2.1 ASTM Standards:
• D-1200-70 Viscosity of Paints, Varnishes, and
Lacquers by Ford Viscosity Cup
• D-2369-81 Standard Test Method for Volatile Content
of Coatings
• D-1005-51 Measurement of Dry Film Thickness of
Organic Ccatings
*Many conventional industrial units are used throughout the
test procedure to accommodate participating laboratories and to
minimize conversion errors on site. Metric conversions are
made as required as shown in the conversion list at the front
of the report.
87
-------
• ASTM D1212-79 Measurement of Wet Film Tnickness of
Organic Coatings
• ASTM D2353-68 Flow Rating of Organic Coatings Using
the Shell Flow Comparator
• ASTM D1475-6U Density of Paint, Varnish, Lacquer,
and Related Products
2.2 ANSI/IEEE Metric Practice
3. SUMMARY OF METHOD
3.1 A battery of specially designed targets are covered
with preweighed, labeled foil, then spray painted in a
single pass under rigidly controlled conditions as
specified in the test matrix. The foils are removed
from the targets, cured, and weighed. The net weight
gain is divided by the weight of paint sprayed at the
targets to yield a single transfer efficiency
determination.
3.2 The battery of targets is composed of 2 sets of 4
targets each. The first set of targets consists of 4
foils mounted in prescribed positions on a large steel
plate. The mean weight gain for these 4 foils is used
to calculate the transfer efficiency. This target con-
figuration is intended to be representative of large,
relatively flat industrial workpieces. The second set
of targets consists of 4 foils mounted on widely spaced
vertical cylinders. The mean weight gain for these 4
foils is used to calculate the transfer efficiency.
This target configuration is designed to be repre-
sentative of smaller, more intricate and open
industrial workpieces.
3.3 A transfer efficiency determination shall be made for
each set of conditions shown in each test matrix, ex-
cept that runs will be performed in randomized order
within each matrix.
3.4 base conditions ("a") shall be established through a
set of pre-test runs to determine levels of each
factor at good spray conditions. Deteriorating levels
{"1,0,-1,-a") of each factor will be determined from
the base levels. The base level and deteriorating
levels of each factor shall be determined prior to
beginning each test matrix.
88
-------
4. TEST TARGETS
4.1 Test targets consisting of a set of 1-1/4-inch diameter
aluminum cylinders and a large stainless steel flat
pan.^1, configured as shown in Figure A-l or Figure A-2,
shall be used for this test.
5. TEST APPARATUS
5.1 Spray painting booth, preferably back-drawn with
100-fpm linear air velocity at the plane of the tar-
gets or, if not available, any booth meeting regula-
tions for the type of spray system being tested may be
used. The same spray booth shall be used for all tests
of a series. The spray booth must be large enough to
accommodate che prescribed targets. The spray booth
must be equipped with a conveyor system capable of
carrying the test panels past the spray equipment at
the desired speed, and capable of at least 2 linear air
rates.
5.2 Four complete systems (AAE, ALE, AAC, and ALC) for
spray painting application, including spray gun,
paint supply pot, power supply (if electrostatic),
air supply lines, paint supply lines, power cables
(if electrostatic), regulators, and pressure gages
shall be used in this test.
5.3 Scales of suitable size and accuracy shall be used for
paint mass flow rate determinations. Laboratory scales
of suitable size and accuracy shall *e used for
weighing test foils. Accuracy of 0.01 percent is
recommended as a minimum accuracy for scales.
5.4 Foil, mounted to cover vertical cylinder and flat test
panels as shown in Figure A-l and Figure A-2 shall be
used. Six-inch wide 1.5-mil medium temper alloy foil
shall be used for covering the test panels. The shiny
side of the foil shall always face out.
5.5 A standard 10-minute stopwatch with 0.1-second accuracy
shall be used.
5.6 Tape measure, graduated in 1/16 of an inch, 10 feet
long, such as a rigid carpenters' rule, may be used.
5.7 Aluminum foil dish, 58 mm in diameter by 18 mm high
with a smooth bottom surface shall be used.
5.8 Syringe, 5 ml, capable of dispensing the coating under
test at sufficient rate that the specimen can be
dissolved in solvent shall be used.
89
-------
Target movement toward gun
hi-t film
I lucl in-sj
vO
o
conveyor
Foil Number
4321
Foil Number
4321
Vertical Cylinder
Target (VC)
1
6"
C"
*"*'
h"
6"
6"
M
V
C
o
ID
U
•A
d"
Flat Panel Target (FP)
Figure A-l. Target Configurations for Air Atomized
Conventional and Electrostatic Spray Guns
-------
Target movement toward qun
conveyor
vo
\
\
1
.0"
1
•
x •
M
U
>
-------
5.9 Forced draft curing cwen, sufficient to hold a com-
plete set of test foils and aluminum dishesr shall
be used.
5.10 Wet film measurement gage.
5.11 Thermometer, with suitable range for spray and cure
conditions, accurate to 0.2°F shall be used.
5.12 Anemometer, with suitable range for booth linear
velocity, accurate to 3 percent of reading shall be
used.
5.13 Test Notebook, a bound test notebook containing the
test procedure, data sheets, reference methods, and
QA/QC Plar shall be provided to the laboratory by
CENTGC.
6. PROCEDURE AND CALCULATIONS
6.1 Perform calibration of the platform scale once per week
or each time that it is moved and leveled, whichever
occurs more frequently. Perform calibration of the
laboratory scale once every test series. Calibrate all
pressure gages per standard operating procedure prior
to test.
6.2 Select test equipment for first test series. Using
Data Sheet 1, document the test equipmpnc specifica-
tions. Be suio to check all information and sign the
form. Each data sheet shall be double checked by a
second party, either engineer or technician, and signed
off.
92
-------
Data Sheet 1
Test Equipment Specifications
Test Date: Test No.: Data by/Checked by;
A. Weight Percent Solids Measurement Equipment
1. Laboratory Scales
a. Manufacturer
b. Model No. ~
c. Serial No. ~
d. Capacity, g ~
e. Rated accuracy, g ~
2. Foil Dishes
a. Type
b. Size
3. Syringe
a. Type
b. Capacity, mL ~
4. Solvent Type "
B. Conveyor Speed Measurement Equipment
1. Rule
a. Type
b. Graduations
2. Electronic Timer
a. Type
b. Manufacturer
c. Model No.
d. Serial No.
e. Rated accuracy, s
C. Mass Flow Measurement Equipment
1. Platform Scales
a. Manufacturer
b. Model No.
c. Serial No.
d. Capacity, kg
e. Rated accuracy, g
2. Stopwatch
a. Manufacturer
b. Model No. _
c. Serial No. __
d. Rated accuracy, s
D. Target Foil
1. Type
2. Nominal Thickness, mils
3. Temper
E. Wet Film Measurement Equipment
a. Manufacturer
b. Model No.
93
-------
6.3 Select coating. The same coating shall be used for all
tests in this projram. Using Data Sheet 2- document
the paint characteristics. Paint characteristics shall
be documented daily, at each addition of paint, and at
other times as requested by the CENTEC engineer or GRACO
representative. Again, check your information and sign
the form.
Dita Sheet 2
Paint Specifications
Test Date:
Test No.:
Data by/Checked by:
1. Paint Type
2. Resin Type
3. Manufacturer
4. Manufacturer's Paint ID No.
5. Lot No.
6. Color
7. Recommended Cure Schedule
8. Viscosity (uncut)
9. Reducing Solvent
10. Vol. of Solvent Put into
Vol. Paint
11. Viscosity - Spray (cut)*
12. Wt./Gallon - Spray
13. Wt. Solids - Spray
14. Resistivity or Conductance
min. 9 "F
sec.* Ford Cup 8 °F
(vol) solvent in
(vol) paint
see.*
Ford Cup g
Ibs/qal
«A
*Use ASTM D-2353-68, ASTM D-1200-70, or ASTM D-3794 part 6.
94
-------
6.4 Set up paint supply eqii'pment and platform scale.
Using Data Sheet 3, document the paint supply equip-
ment specifications. Be sure to check your informa-
tion and sign the form.
95
-------
Data Sheet 3
Paint Spray and Peripheral Equipment Specifications
Test Date: Tesc No.: Data by/Chkd by:
A. Paint Supply Tank
1. Type
2. Manufacturer
3. Model No.
4. Serial No.
5. Rated Capacity, gal
B. Paint Spray Equipment
1. Type
2. Manufacturer
3. Model No.
4. Serial No.
5. Rated Capacity, cc/min
6. Air Cap
7. Fluid Tip
8. Needle
C. Paxnt Spray Booth
1. Type
2. Manufacturer
3. Model No.
4. Serial No.
5. Rated Capacity, cfm
D. Conveyor
1. Type
2. Manufacturer
3. Model No.
4. Serial No.
E. Forced Draft Oven
1. Type
2. Manufacturer
3. Model No.
4. Serial No.
F. Paint Heaters
1. Type
2. Manufacturer
3. Model No.
4. Serial No.
96
-------
6.5 For electrostatic spray equipment only, ground paint
supply equipment and platform scale per Figure A-3.
NOTE: In accordance with Section 9-8 of NFPA 33 for
fixed electrostatic apparatus, measure resistance of
equipment to ground (conveyor frame) to insure resist-
ance is less than 1 x 10& Ohm.
6.6 Using a small glass jar with an airtight lid, take
paint grab sample from paint pot. ASTM D-3925-81 pro-
vides a good standard practice guide for paint sampling.
Record test series number on label of jar.
6.7 Measure weight solids from paint sample. Use syringe
weight difference technique as described in A.STM
D-2369-81. Document the cure oven bake sch-adule and
temperature on Data Sheet 4. Be sure you use the cure
schedule recommended by the manufacturer on Data Sheet
2. Record raw data and results on Data Sheet 5.
Paint weight percent solids should be determined before
each test series, at the start of each test day,
periodically between tests, and at the end of each test
day. The participating laboratory shall store all weight
percent solids samples until notified by CENTEC that the
data analysis is complete.
97
-------
nc A
I
Paint
Suppl
Tank
K Digital electronic
platform scale
Arrangenent B
Electrostatic Equipment
Gi
oindir.g Cable
gMflMHgggkggj
Paint
I_
Supply
Tank
.^Electrical Insulation Block*
r
•Slock -ust be capaole of preventing current flow from supply tank to
grour.a t.-.rougn t^e platfor— scale.
Figure A-3. Set-up for Paint Supply Equipment and
Platform Scales
98
-------
Data Sheet 4
Equipment Operating Conditions
Test Date: Test No.: Data by/Chkd by:
A. Paint Spray Equipment
1. Paint Pressure at Paint Pot, psig
2. Paint Pressure at Spray Gun, psig
3. Atomizing/Turbine Air Pressure at
Spray Gun, psig
4. Operating Voltage, kV
5. Disk or Bell Speed, rpm
a. With Paint Applied
b. Without Paint Applied
6. Shaping Air for Bell, psig
7. Paint Temperature at Paint Pot, °F
8. Gun to target distance, cm
9. Pump Setting
B. Paint Spray Booth
1. Ambient Temperature, °F
2. Relative Humidity, %
3. Air Flow Velocity, fpm
4. Air Flow Direction
C. Target Parameters
1. Average Wet Film Thickness, mils
2. Average Dry Film Thickness
3. Vertical Paint Coverage, cm (in)
4. Target Height, cm (in)
5. % Vertical Coverage
6. Resistance to Ground, Ohm
D. Forced Draft Oven*
1. Cure Time, minutes
a. Foil Dish (sample)
b. Target Foil
2. Cure Temperature, °F
a. Foil Dish (sample)
b. Target Foil
E. Paint Heaters
1. Temperature In, °F
2. Temperature Out, °F
F. Conveyor Speed Setpoint, fpm (cm/sec)
*Same cure schedule as foils.
99
-------
Data Sheet 5
Weight Solids Test Data & Results
Test Date:
Test No.:
Data by/Chkd by:
1. Syringe Weight
a. Full, g
b. Empty, g
c. Net Wet Paint, g
2. Dish Weight
a. After Drying, g
b. Empty, g
c. Net Dry Solids, g
3. % Weight Solids (2c/lc)
Sample
A
Sample
B
Average
A3
NOTES:
1. Actual Cure Schedule
min.
Refer to ASTM 2369-81, Procedure B of "Standard Test Method for
Volatile Content of Coatings."
100
-------
6.8 Set up the paint spray equipment. Using Data Sheet 3,
document specifications for the paint spray equipment
and spray booth used in this test. Check your infor-
mation and sign the data sheet.
NOTE; Equipment selection, equipment
condition, paint selection, target con-
figuration, and operating conditions have
a substantial effect on transfer efficiency.
Care should be taken to use the same booth
and spray equipment, paint, targets, and
operating conditions as specified for the
run in the test matrix (Data Sheet 4, Sec-
tions A, B, C, and 6a, Sections Id, Ic,
3, and 4) from test to test for comparable
results.
6.9 Set up the conveyor speed measuring equipment. This
equipment may consist of photoelectric cells or limit
switches used in conjunction with an automatic digital
timer. Alternatively, the conveyor speed may be nea-
sured using timing marks (chalk marks) on the conveyor
in conjunction with a hand held stopwatch. Figure A-4
shows the permissible methods for conveyor speed mea-
surement. Using Data Sheet 6a, record the horizontal
distance between the photo cell or limit switch on/off
positions.
6.10 Determine base level of each test factor which will
produce a reasonably good spray pattern and finish.
If base level has already been determined for test
series, proceed to 6.13. The CENTEC engineer in
agreement with the laboratory representative shall
determine "reasonably good spray pattern and finish."
Base level shall be determined only once for each
test series. Base level is denoted by "a" in the
test matrix.
6.11 Determine deteriorating levels of each test factor to
be examined in this test series. Selection shall be
niad<» by reducing the level of each factor to a point
where an obvious impact on spray pattern and finish
is noted. Again, the CENTEC engineer in agreement
with the laboratory representative shall determine
the level where spray pattern and finish is obviously
poor. This level is the poorest value of each
factor. It is denoted by "-a" in the test matrix.
101
-------
METi:02 A
Tarqet
E.'.ictronic
Timer
e ,
n
A • Stationary photoelectric cell or limit switch
3 - Stationary photoelectric cell or limit switch
C • Moving plate of KJIOWH width
METHOD B
Known Distance
Conve"or
£ • Fixed timing nark
r « Moving tvsunq nark
Target.
Target
=
Target
IF
G
Stop Hatch
Figure A-4 • Permissible Methods for Measuring
Conveyor Speed
102
-------
6.12 Levels "1," "0," and "-1" shall be selected at even
absolute spacing from the value of each variable "a"
to "-a."
6.13 The value of "a", "1", "0", "-l"r and "-a" shall remain
fixed for each variable through a test series.
6.14 Set up targets in accordance with Figure A-l or A-2, as
appropriate. Target configuration, material, and
spacing is critical. Scavengers are metallic, as is
the FP target. Cut 6-inch-wide aluminum foil strips to
required length for each target. Label each foil strip
with the appropriate nomenclature. (Nomenclature is
shown on Table A-l.) Weigh each foil strip and record
value on foil and on Data Sheet 6b. Check your infor-
mation and sign the data sheet.
6.15 Attach foils to the vertical cylinder and/or flat panel
targets as shown on Figure A-5 or A-6, as appropriate.
Perform resistance check to verify adequacy of ground-
ing. Per NFPA 33 Section 9-8, resistance shall be less
than 1 x 10* ohms.
6.16 In accordance with Figure A-l or A-2, attach shim stock
to scavenger in order to measure wet film thickness.
6.17 Adjust all equipment operating parameters, i.e., gun to
target distance, paint pot pressure, turbine air pres-
sure, etc., to base values. Set factor levels to values
required for this test run in the matrix. Record equip-
ment operating parameters on Data sheet 4. Check your
data and sign the data sheet. NOTE: In accordance with
Section 9-7 of NFPA 33 for tised electrostatic apparatus,
the gun to target distance shall be at least twice the
sparking distance.
NOTE; Equipment selection, equipment
condition, paint selection, target con-
figuration, and operating conditions have
a substantial effect on transfer efficiency.
Care should be taken to use the same booth
and spray equipment, paint, targets, and
operating conditions (Data Sheet 4, Sec-
tions A, B, C, and 6a, Sections Id, Ic, 3,
and 4) from test to test for comparable
results.
6.18 Check spray equipment and parameters to assure they
are correct for this run.
103
-------
TABLE A-l. NOMENCLATURE FOR SPRAY PAINTING
TRANSFER EFFICIENCY TESTS
Each test foil will be labeled in 5 segments as follows:
1. Spray Equipment Type
Air atomized conventional : AAC
Airless conventional : ALC
Air atomized electrostatic : AAE
Airless electrostatic : ALE
2. Target Configuration
Flat Panel : FP
Vertical Cylinder: VC
3. Target Position: 1, 2, 3, or 4.
4. Test Serirs Identifier (letter or number)
Example: AAC-FP2-12
where AAC = air atomized conventional spray equipment
FP2 « the second flat panel target
12 • test rur. identifier
104
-------
f: rt:-al Cylinder
Direction cf -vrar
Hold edgo of foil in place against
cylinder »hilc wrapping leading edge
arour.d cylinder.
roil
II | '
HI j
i. '.-.'far vortis.ii cylinder taracts with cylinders nounted-on target bracket
iSco riqurc ; inc J i . Wrap so the leadinq edge foms a seaw away
frcn t:-.o cir:rtion of srrav.
"Cri="
"Crip"
"Grip"
"Grip-
As leading odTC overlaps starting
oilgo, solidly "grip" foil into piaco
by tjraspinc foil-covered cylinder.
Secure foil on cylinder by gripping
the length of the cylinder. Foil will
have a uniformly wrinkled surface.
Figure A-5. Vertical Cylinder Wrapping Technique
105
-------
FLAT PANEL TAR&ZT
Flat Panel Target
7oil-Peady for
attachment -
Double-sided tape
Figure A-»>. Flat Panel Foil Attachment Technique
106
-------
Data Sheet 6a
TE Test Data and Results
Test Date: Test No.: Data by/Checked by:
A. Weight Percent Solids (from Data Sheet 5) A3
B. Total Solids Sprayed
1. Paint Spray Plow Rate
a. Beginning Weight, g
b. End Weight, g
c. Time Between Weighings/ s
d. Flow Rate, g/s Bid
2. Conveyor Speed
a. Distance Between Marks, cm
b. Time Between Narks, s
c. Speed, cm/a B2c
3. Total Effective Target
Width, cm* 15.24 63
4. Total Solids Sprayed at Each
Target, g
(A3 x Bid x B3/B2c) B4
5. Micromotion-metered paint
mass flow rate, g/s Bid"
* Total effective target width is six inches per foil on flat
panel target (on 6" centers), and six inches per cylinder
on vertical cylinder target (also on 6* centers). Six
inches • 15.24 cm.
107
-------
6.19 For electrostatic spray equipment, measure the gun tip
operating voltage (with lines full of paint, but gun
not operating). Adjust to desired voltage and record
on Data Sheet 4.
6.20 Check conveyor clock, stopwatch, micromotion meter and
platform scale to ensure that all have been zeroed
(reset) and that the scales are in the tare mode.*
6.21 Turn on conveyor. As the leading edge of the first
scavenger passes in front of the gun, turn on paint
spray equipment and initiate flow; simultaneously,
start stopwatch and record scale reading.
6.22 As the trailing edge of the last scavenger passes in
front of the gun area, stop stopwatch and paint spray
flow simultaneously. Turn off conveyor. Record
platform scale, conveyor clock, micromotion meter
flow rate, and stopwatch readings on Data Sheet 6a.
Check the data and sign the data sheet.
6.23 Measure wet film thickness on shim plate and record on
Data Sheet 4. line C-l.
6.24 Remove foils from targets, making sure no tape has
stuck to the targets and no paint is lost. Securely
attach coated foils to oven racks so all painted
surfaces are exposed for uniform drying. Spring
clips or tacks may be used to.mount wet targets on
racks. Insert racks in oven And bake at recommended
schedule per Data Sheet 2. Flash time (the time be-
tween spraying and getting the targets into the oven)
should be kept to a minimum. Set oven timer per
recommended schedule.
6.25 Remove foils from oven and record actual bake schedule
on Data Sheet 4. Weigh foils and record weight on each
foil and on Data Sheet 6b. After weighing, store foils
in appropriately labeled plastic bags, i.e., bags that
have test run number identified. The laboratory shall
retain all samples until data analyses are complete.
Check all data for correctness and completeness. Both
the engineer and technician must check and sign all
data sheets before proceeding.
Replicates ot each test run shall be made immediately
after the original run, if required.
*Du.ing 10 tasts, ch^ck micromotion meter vs manual deter-
minations. If within precision requirements (see QA/QC Plan),
use only micromotion meter thereafter.
108
-------
Data Sheet 6b
IE Test Data and Results
Test Date:
Test No.:
Data by/Checked by:
C. Total Solids on Target
Flat Panel Target
Foil Weight After Drying, g:
Foil f 1 2
Foil Weight Before Spraying, g:
Total
D. Vertical Cylinder Target
Foil Weight After Drying, g:
Foil * 1 2
Net Dry Solidsf g:
Total
Foil Weight Before Spraying, g:
E. Transfer Efficiency (by weight]1
Flat Panel Target
Vertical Cylinder Target
Net Dry Solids, g:
1. TL =
(Net Dry Solids, g) x 100%
("Total Solids Sprayed at Each Target, g" fron Data Sheet 6a) x (4 targets)
109
-------
6.26 Perform TE calculations as indicated on Data Sheet 5,
6a, and 6b using the weight solids determined for
the test series. Document results on Data Sheet 6b,
noting that each transfer efficiency observation is
the mean of the transfer efficiency for 4 foils.
6.27 Repeat above steps (6.2 through 6.26) for each test
run.
6.28 Be sure all data sheets have been properly completed,
checked, and signed.
6.29 Record transfer efficiency in appropriate column on
Data Sheet 7. When roughly 70 percent of the runs
in a series are complete, the CENTEC engineer shall
transmit the TE results to Dr. Ray Myers at 703-
961-5638. Dr. Myers shall perform an outlier
analysis and respond to the engineer within 24
hours. Outlier runs will be repeated as resources
allow.
6.30 To proceed with the next run in a series, go to
6.10. To begin a new test series, go to 6.1.
6.31 CENTEC shall retain all original data sheets and the
test notebook.
110
-------
Where:
Data Sheet 7
Air Atomized Conventional Test
FACTOR TE Result
Run Number A B C D E PP VC
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
-1
-1
-1
-1
1
1
1
1
1
1
1
-1
-1
-1
-1
1
-a
a
0
0
0
0
0
0
a
a
a
a
a
a
-1
-1
-1
1
-1
1
1
1
-1
-1
-1
1
1
1
-1
1
-1
-1
1
1
-1
-1
1
1
1
1
1
1
1
1
-1
-1
1
-1
-1
1
»1
-1
1
1
-1
1
1
-1
1
1
0
0
-a
a
0
0
0
0
a
a
a
a
a
a
-1
1
-1
-1
-1
-1
1
-1
1
-1
1
1
-1
1
1
1
0
0
0
0
-a
a
0
0
a
a
a
a
a
a
1
-1
-1
-1
-1
-1
-1
1
-1
1
1
-1
1
1
1
1
0
0
0
0
0
0
-1
1
1
1
1
1
1
1
A = Restricted air lines—test at 5 levels: a,l,0,-l,-a
B = Booth air rates—test at 2 levels: 1,-1
C B Gun cleanliness—test at 5 levels: a,l,0,-l,-a
D = Restricted paint lines—test at 5 levels: a,l,0,-l,-a
E = Pan air—test at 3 levels: 1,0 ,-1
Ml
-------
Where:
Data Sheet 7
Airless Conventional Test
FACTOR TE Result
Run Number BCD Dummy FP VC
1 -1 -1 -1 -1
2 1 -1 -1
3 -11-1 -1
4 -1-11 -1
5 -1 -1 -1 1
6 11-1-1
7 1-11-1
8 1-1-1 1
9 -111-1
10 -11-1 1
11 -1-11 1
12 111-1
13 11-1 1
14 -111 1
15 1-11 1
16 111 1
17 1 -a 0 0
18 1 a 0 0
19 -1 0 -a 0
20 -1 0 a 0
21 100-1
22 100 1
23 1 a a 1
24 1 a a 1
25 1 a a 1
26 1 a a 1
27 1 a a 1
28 1 a a 1
B » Booth air rates—test at 2 levels: 1,-1
C = Gun cleanliness—test at 5 levels: a,1,0,-1,-a
D = Restricted paint lines—test at 5 levels: a,l,0,-l,-a
Dummy = Dummy variable not expected to affect TE
112
-------
Data Sheet 7
Run Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Air Atomized Electrostatic Test
B
TE Result
FP VC
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
a
a
0
0
0
0
0
0
0
0
0
0
9
-1
1
-1
1
-1
1
-1
1
-1
1
-1
1
-1
1
-1
1
-1
-1
1
1
-1
-1
1
1
-1
-1
1
1
1
1
1
1
1
1
-1
1
-1
1
-1
1
-1
1
1
-1
1
-1
1
-1
1
-1
0
0
a
-a
0
0
0
0
0
0
0
0
a
a
a
a
a
a
-1
-1
1
1
1
1
-1
-1
-1
-1
1
1
1
1
-1
-1
0
0
0
0
-a
a
0
0
0
0
0
0
-1
-1
1
1
1
1
-1
-1
1
1
-1
-1
-1
-1
1
1
0
0
0
0
0
0
-1
1
0
0
0
0
1
1
1
1
1
1
-1
-1
1
1
-1
-1
1
1
-1
-1
1
1
-1
-1
1
1
0
0
0
0
0
0
0
0
-a
a
0
0
a
a
a
a
a
a
-1
1
-1
1
1
-1
1
-1
1
-1
1
-1
-1
1
-1
1
0
0
0
0
0
0
0
0
0
0
-1
1
1
1
1
1
1
1
Where:
A
B
C
D
Z
F
Restricted air lines—test at 5 levels: a,l,0,-l,-a
Booth air rates—test at 2 levels: 1,-1
Gun cleanliness—test at 5 levels: a,l,0,-l,-a
Restricted paint lines—test at 5 levels: a,l,0,-l,-a
Fan air—test at 3 levels: 1,0,-1
Voltage—test at 5 levels: a,l,0,-l,-a
G = Electrode position—test at 3 levels: 1,0f-l
113
-------
APPENDIX B
QUALITY ASSURANCE/QUALITY CONTROL PLAN
SENSITIVITY STUDIES ON THE EFFECTS ON TRANSFER EFFICIENCY
OF IMPROPERLY MAINTAINED OR OPERATED SPRAY PAINTING EQUIPMENT
CENTEC Corporation
Reston, Virginia 22090
CONTRACT NO. 68-03-1721, Task 1
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
APPROVAL SIGNATURES:
Charles H. Darvin
EPA Project Officer
(Original signed by Guy Sims)
Guy F. Slines
EPA Quality Assurance Officer
Edward H. Comfort
Quality Assurance Officer
Kerri C. Kennedy
CENTEC Sr. Project Engineer
JANUARY 19AA
114
-------
TABLE OP CONTENTS
Paqe
SECTION 1 INTRODUCTION B-3
SECTION 2 PROJECT DESCRIPTION B-4
SECTION 3 PROJECT ORGANIZATION AND RESPONSIBILITY . . B-6
SECTION 4 QA OBJECTIVES FOR MEASUREMENT DATA IN
TERMS OP PRECISION, ACCURACY, COMPLETE-
NESS, REPRESENTATIVENESS AND
COMPARABILITY B-8
SECTION 5 SAMPLING PROCEDURE B-10
SECTION 6 SAMPLE CUSTODY B-11
SECTION 7 CALIBRATION PROCEDURES, ANALYTICAL
PROCEDURES AND FREQUENCY B-12
SECTION 8 DATA REDUCTION, VALIDATION, AND REPORTING . B-13
SECTION 9 INTERNAL QUALITY CONTROL CHECKS B-15
SECTION 10 PERFORMANCE OP SYSTEM AUDITS B-16
SECTION 11 PREVENTATIVE MAINTENANCE B-17
SECTION 12 SPECIFIC ROUTINE PROCEDURES TO ASSESS
DATA PRECISION, ACCURACY AND
COMPLETENESS B-1 8
SECTION 13 CORRECTIVE ACTION B-21
SECTION 14 QUALITY ASSURANCE REPORTS TO MANAGEMENT . . B-22
J15
-------
Sect ion 1
Revision No. Original
Date December 1983
Page 1 of 1
SECTION 1
INTRODUCTION
This quality assurance/quality control (OA/OC) plan assures
collection of high quality data and insures consistent quality
control measures for data developed under "Sensitivity Studies
on the Effects on Transfer Efficiency of Improperly Maintained
or Operated Spray Painting Equipment," Contract No. 68-03-1721.
Under this contract, CENTEC Corporation will be conducting tests
using a draft standardized method to determine the effect of
operating and maintenance parameters on transfer efficiency (TE),
116
-------
Sect ion 2
Revision No. Original
Date December 1983
Page 1 of 2
SECTION 2
PROJECT DESCRIPTION
Sensitivity studies on the effects of TE of improperly maintained
or operated spray painting equipment will be conducted in this
test program. A draft standardized TE method will be used for
all tests in this program. The draft standard TE test method
consists of passing a prescribed set of preweighed targets in
front of spray equipment under rigidly controlled conditions in
an industrial laboratory spray booth. The cured painted targets
are weighed, and the original weight subtracted from the final
weight to obtain the net dry solids deposited on the targets.
The net dry solids is divided by the total solids sprayed at the
targets, which is then multiplied by 100 percent to determine TE.
A complete description of the draft standard TE test method is
contained in Appendix A of the Subtask Report for this contract.
Four types of spray equipment will bp tested during this program:
air atomized conventional (AAC), airless conventional (ALC), air
atomized electrostatic (AAE), and airless electrostatic (ALE).
Each type of equipment has an individual experimental design.
Five operation and maintenance (OtM) factors have been selected
for testing on conventional spray equipment. These factors
include booth air rates, atomizing air pressure, fan air, paint
pressure, and cleanliness of the spray gun. Qualitative factors
(booth air rate and fan air) will be tested at a minimum of
117
-------
Sect ion 2
Revision No. Original
Date December 1983
Page 2 of 2
levels each, while quantitative factors (atomizing air, cleanli-
ness, and paint pressure) will be tested at five levels. Levels
will be selected for testing based on an original set-up with a
good spray pattern. For electrostatic guns these factors will
be tested as described, except two more factors, tip voltage and
electrode position, will also be tested. Tip voltage will be
tested at five levels, while electrode position will be tested
at three. Six replicates are provided for each equipment type.
The four experimental designs are planned to provide enough data
to support development of a response surface and regression
equations to describe the response surface. A complete descrip-
tion of the experimental design is included in the Subtask
Report for this contract.
Negotiations are underway with r.n industrial laboratory to
arrange the test program. Testing is scheduled to begin in
February H34, and will last approximately one month.
113
-------
Secti on
Revision No.
Date January
Page""""" 1 or
SECTION 3
PROJECT ORGANIZATION AND RESPONSIBILITY
This project is administered through CENTEC Corporation structure,
as shown in Figure B-1. Day-to-day test program activities will
be managed on-site by a CENTEC Senior Project Engineer in direct
contact with CENTEC QA management personnel.
At the test site, the CENTEC engineer is responsible for imple-
menting QA throughout the test program. The engineer conducts
onsite evaluations to verify the degree of implementation,
assures that appropriate QA records are kept, provides QA
direction to the laboratory staff, and reports regularly to the
Project Manager on the status of QA.
The Project Manager, Ed Comfort, is the Quality Assurance
Officer. He continuously monitors the implementation of QA and
provides feedback to the CENTEC engineer onsite and to CENTEC
management. QA records Jeep1- by the engineer (onsite) and by the
Quality Assurance Officer (offsite) serve as resources for
preparing reports and documenting adherence to QA procedures
and specifications.
-------
CHAIRMAN AND
PRESIDENT
PAUL S. MINOR
(Director)
VICE PRESIDENT
AND
CHIEF TECHNICAL OFFICER
CHARLES S. MATHENY
(Director)
VICE PRESIDENT
FINANCE AND
ADMINISTRATION
ROBERT D. SMITH
VICE PRESIDENT
PROCESS SYSTEMS
CHARLES M. ROWAN
VICE PRESIDENT
CENTEC APPLIED
TECHNOLOGIES
ROBERT SCHAFFER
VICE PRESIDENT
MANAGEMENT SYSTEMS
CURT GRINA
QUALITY ASSURANCE
OFFICER
ED COMFORT
SENIOR PROJECT
ENGINEERS
K. KENNEDY
C. ROBERTS
Figure B-l. Project Organization as Related to Corporate Structure
-------
Sect ion 4
Revision No. Original
Date December 1983
Page 1 of 2
SECTION 4
QA OBJECTIVES FOR MEASUREMENT DATA IN
TERMS OF PRECISION, ACCURACY, COMPLETENESS,
REPRESENTATIVENESS AND COMPARABILITY
For each manor measurement parameter specific QA objectives for
precision, accuracy, and completeness are required. These
objectives are detailed in Table B-1.
Care must be taken to assure that all measurements are repre-
sentative of the media and conditions b<»inj measured. Proven
techniques or methods are therefore use3 for all measurements.
Data quality objectives are based on accuracy and precision of
each measurement parameter, as established in Table B-1. Data
integrity will be validated throuoh a series of inspections and
tests described later in this plan.
121
-------
Table B-1. Spray Painting Transfer Efficiency Precision,
Accuracy and Completeness Objective
Itoaaurenent Paranetar
(Matted)
s^t
s Grounding
• Vbltaga
s units
s Distance-length
a Tine (Stopwatch.
tUer)
• tot Pilis Thickness
• Dry nim Thickness
s VlacoBity (Ford cup)
* teslstivity
• Pressure
s Temperature
• Linear Air Velocity
(rotating vem or
haatsd wire anenoej
s Density
S Mt% SollC*
s> Paint Stapling
Reference
Matted
IfflR Std 12-1972
AKSt/IEES Std 142-1972
ANSI C2
IEEE Std 4-1918
ASTM IS 380-76/
1BEB Std 26»-1976
{SM AST* 1200-70)
AS™ D-1212-79
AS™ 0-1005-51(1079)
AS« D 1200-70(1976)
AS™ D 2353-68(1978)
ACEIH Mecoenndad
Practice. Section 9*
ter)
AiHM D 1475-60(1980)
ASM D 2369-81
ASM 0 3925
Exuerlnental Conditions
Laboratory conditions
Laboratory conditions
Laboratory conditions
Laboratory conditions
Laboratory conditions
laboratory oondltiona
Laboratory conditions
Laboratory conditions
TMt conditions
In accordance with Mm
Precision
(Std. Deviation)
lab scale 0.01 g
plat, seals 5 g
™ "
0.05 HV
—
1/32 in
O.ls
0.265 •!!
(2%)
+0.1 mil
1.5s
0.1 MQ
5%
0.1«C
33 (3%)
«C.001g/W.
(1.5%)
—
Accuracy
lao scale *0.01 g
plat, seals «,5 g
^™ *
*0.1 W
^
1/64 In
0.2s
0.85 mil
2%
2s
0. 1 MU
Air atomlMd «0.5 kPa
Airless 0.5 kPs
0.1-C
±3%
0.002 gyW.
4.7%
—
CotRDletaneas
100%
1CO%
190%
100%
100%
100%
110%
100%
100%
100%
100%
100%
100%
100%
100%
s Condition in Container ASTM D 3011-1 — —
•Industrial \tentllatlon - A Manual oC
Practice. American Oonfervnoe at GovemBntal Induatxial Hygenlats, 1972.
-------
Sect ion 5 ,
Revision No. Original
Date December 1983
Page 1 of 1
SECTION 5
SAMPLING PROCEDURE
A description of the sampling procedure is provided in the
Subtask Report, Appendix A, Draft Standard Test Method for Spray
Painting Transfer Efficiency. The draft standard test method
includes:
• A description of the test method, including references
to standard methods
• Figures illustrating specific operations
• Description of sampling and test equipment
• Data sheets
• Other special conditions and considerations in
performing the test
-------
Section §
Revision No. Original
Date December 1983
Page 1 of 1
SECTION 6
SAMPLE CUSTODY
Sample custody procedures are addressed in Draft Standard Test
Method for Spray Painting Transfer Efficiency, Subtask Report
Appendix A. The CENTEC engineer and laooratory technician will
check and sign all data sheets. The laboratory will retain all
weighed foils, as described in the draft test method, until the
data analysis is complete. CENTEC will retain all original data
sheets.
1241
-------
Sect i on 7
Revision No. Original
Date December 1983
Page 1 of 1
SECTION 7
CALIBRATION PROCEDURES, ANALYTICAL PROCEDURES AND FREQUENCY
Calibration procedures, analytical procedures* end frequency
requirements are included in Draft Standard Test Method for
Spray Painting Transfer Efficiency, Subtask Report, Appendix A.
125
-------
Sect icn 8
Revision No. Original
Date December 1983
Page 1 of 2
SECTION 8
DATA REDUCTION, VALIDATION, AND REPORTING
8.1 GENERAL
Data will be collected at the test laboratory under the guidance
of a CENTEC engineer. The data will be collected and documented
according to the requirements of the Draft Standard Test Method
for Spray Painting Transfer Efficiency (Subtask Report, Appendix
A). Equations for reducing the data are also contained in the
draft standard test Method. Figure B-2 shows the responsible
parties for each data validation and reduction step.
8.2 DATA REDUCTION, VALIDATION, AND REPORTING
Data reduction will be performed using standard statistical
practices as described in Draft Standard Test Method for Spray
Painting Transfer Efficiency, Subtask Report, Appendix A. Any
data generated by test runs with known discrepancies in perform-
ance will be labeled as suspect for later evaluation. Duplicate
data for all suspect runs will be obtained whenever resources
permit.
For each experimental design, the reduced data will be subjected
to a series of t tests using studentized residuals to evaluate
outliers. This evaluation will be performed onsite when 75 per-
cent of a test series is complete. Outliers will be replaced by
duplicate runs as resources permit. Any remaining outliers will
be eliminated from the data set where possible without rendering
the data set useless.
-------
CENTEC ENGINEER ON-SITE
CENTEC PROJECT MANAGER
DATA COLLECTION
TE, SD, COV
CALCULATIONS
OUTLIER
EVALUATION
STATISTICAL
ANALYSIS AND
MODELING
Figure B-2. Data Validation Resposibilities
-------
Sect ion 9
Revision No. Original
Date December 1983
Page 1 of 1
SECTION 9
INTERNAL QUALITY CONTROL CHECKS
Internal quality control checks are incorporated into the
experimental design and Draft Standard Test Method for Spray
Painting Transfer Efficiency (Subf.sk Report, Appendix A).
These checks include a battery of replicates for each type of
spray equipment to be tested. Calibration requirements also are
specified in the Subtask Report, Appendix A. All data is
subjected to two inspections for error (by the CENTEC engineer,
and by a laboratory representative), with concurring signatures
required on each data sheet.
128
-------
Sect ion 10
Revision No. Original
Date December 1983
Page 1 of 1
SECTION 10
PERFORMANCE OF SYSTEM AUDITS
The performance of the TE tests will be monitored constantly as
described in Draft Standard Test Method for Spray Painting
Transfer Efficiency, Subtask Report, Appendix A.
129
-------
Sect ion 11
Revision No. Original
Date December 1983
Page 1 jt 1
SECTION 11
PREVENTAT1VE MAINTENANCE
Certain preventative maintenance (PM) procedures must be followed
to keep downtiine to a miniimjm. Kost PM practices are recommended
by the manufacturer to the spray equipment user. These practices
include keeping the spray equipment and spray area clean, handling
equipment carefully to avoid damage, and using appropriate
equipment for the given job. These general practices must be
observed to prevent inadvertent deterioration of spray equipment
condition and to minimize downtime.
In addition to these PM practices, extra electrodes and air caps
should be kept on hand. Ample supplies for performing TE tests
should be available to avert, shortages. These include foil,
paint, tape, solvent, and others outlined in the draft standard
TE test method.
130
-------
Sect ion 12
Revision No. Original
Da t e December 1983
Page 1 of 3
SECTION 12
SPECIFIC ROUTINE PROCEDURES TO ASSESS DATA PRECISION,
ACCURACY AND COMPLETENESS
After the spray painting system is operational, performance
audits will be conducted to assure continued acceptable pre-
cision and accuracy during testing. It is the nature of the
experimental design for this program that TE results cannot be
tested for outliers until three-quarters of a test series is
complete. To minimize the likelihood of obtaining poor TE
results prior to outlier analyses, performance audits are re-
quired twice daily for each major measurement contributing to TE:
• Net solids on target, g
• Conveyor speed, cm/*
• Paint weight fraction solids
• Paint mass flow rate, g/s
• Effective target width, cm
These measurements are subject to the precision, accuracy, and
completeness criteria in Table B-l. They will be audited for
precison and accuracy at the beginning and completion of each
test day. Periodic audits also may be conducted during the test
day as deemed appropriate by either the laboratory technician or
CENTEC engineer on site. Performance audit requirements are
detailed in Table B-2 and in the Draft Standard Test Method for
Spray Painting Transfer Efficiency.
131
-------
Sect ion 12
Revision No. Original
Date December 1983
Page 2 of 3
TABLE B-2. PERFORMANCE AUDIT REQUIREMENTS
Measurement When
Parameter (units) Performance Audit Method Raquired
Net solids on target(g) o Measure known control weight A,B,C,D
Conveyor speed (CT/S) o Blank run using electronic
timer A,B,C,D
o Chalk mark and stopwatch
Paint weight fraction o Conduct duplicate analyses A,B,C,D
solids per ASTM 2369 at manufac-
turers recommended cure
schedule
Paint mass flow rate (g/s) o Spraying, using stopwatch A,B,C,D
and scales
Effective target width o Ruler or tape measure C
A » Start of each day
B • At change of paint or spray equipment
C • As requested by lab technician or eng.
D « End of each day
132
-------
Sect ion 12
Revision No. Original
Da t e December 1983
Pane 3 of 3
The precision and/or accuracy of the total measurement system will
be documented at least twice daily. Problems identified by the
performance audit will be corrected before continuing with the
test program.
Completeness requirements are audited continuously and automati-
cally by the dual check off procedures required on each data
sheet in the draft standard test method.
133
-------
Sect ion 1J
Revision No. Original
Date December 1983
Page 1 of 1
SECTION 13
CORRECTIVE ACTION
Performance audits are required twice daily for each major
measurement contributing to TE. Should any measurement not meet
the precision or accuracy requirements laid out in Table B-l,
corrective action must be taken. Corrective action includes
recalibration, repair, or replacement of the measurement system
in question. The CENTEC engineer on site is responsible for
initiating the appropriate corrective action, with concurrence
from the participating required in writing in the next QA report
to management.
Corrective action may also be taken to replace data identified
as erroneous by the required data outlier analysis. The CENTEC
Project Manager is responsible for initiating corrective action
to replace outlier data.
Other corrective action may be taken at the request of onsite
CENTEC or laboratory personnel whenever suspect or undocumented
conditions occur. The CENTEC engineer is responsible for all
such corrective actions.
-------
Sect i on 14
Revision No. Original
Date December 1983
Page 1 of l
SECTION 14
QUALITY ASSURANCE REPORTS TO MANAGEMENT
The CENTEC engineer on site will report daily via telephone to
the CENTEC Project Manager regarding the results of all perform-
ance audits, measurement system accuracy, and measurement system
precision. Significant OA problems and recommended solutions
will be discussed. Brief records of these reports will be kept
for later inclusion in the final test report OA section.
135
-------
APPENDIX C
AAC TEST EQUIPMENT AND PAINT SPECIFICATIONS
Data Sheet 1
Test Equipment Specifications
Test Datei Test No.: Data by/Checked by:
-. 5...,w /M,'..,; "^-f l*( r
A. Height Percent Solids Measurement Equipment
1. Laboratory Scales
a. Manufacturer »reei««
b. Model NO. 240-21
C. Serial NO. .-.574227
d. Capacity, g 0-300/0-1000 CRAMS
e. Rated accuracy, g .PIG t»fff\atinr>/.\e,
7. Poil Dish*s
a. T>pe Aluminum
b. Size 5Bar round
3. Syringe
a. Type Class
b. Capacity, nL 5
4. Solvent Type Xvi. v
B. Conveyor Speed Measurement Equipment
1. Rul»
•. Type conventional
b. Graduations 1/32*
2. Electronic Timer
a. Type Elactro-Heehanieal Digital
b. Manufacturer
c. Model No.
d. Serial No.
e. Rated accuracy, s
C. Mass Plow Measurement Equipment
1. Platform Scales
a. Manufacturer y»«« now n«t«r. mn-nmntion
b. Model No. C12*T with DIP *t Indicator
c. Serial No. .2425
d. Capacity, kg o to IQOQ e/mn. .68 to 13 jtG/Min.
e. Rated accuracy, g »i - .4% B»«I»M
2. Stopwatch
a. Manufacturer ^gnm Praeiaion Produeta. Inc.
b. Model No. _^
e. Serial No. _=
d. Rated accuracy, s .ei
C. Target Poil
1. Type Alum. Alley 1145-
2. Nominal Thickness, ailo 1.5
3. Temper *«diua
E. Wet Pi1» Measurement Equipment
a. Manufacturer Cardce 0-4 Mils
b. Model No. Precision Direct Readme.
P. Dry Film Measurement Equipment
a. Manufacturer DsFelske Corp.
b. Model NO. Poaiteeter 2000 .1 mils Accuracy
136
-------
6.3 Select coating. The eanc coating shall be used for all
teats in this program. Using Data Sheet 2, document
the paint characteristics. Paint characteristics shall
be documented daily, at each addition of paint, and at
other times as requested by the CENTEC engineer or GRACO
representative. Again* check your information and sign
the torn.
Data Sheet 2
Paint Specifications
Test Datei
Test No. i
AfV (s
Data by/Che Aed by:
O-v / K'O^
1. Paint Type
2. Resin Typs
3. Manufacturer
4. Manufacturer's Paint ID No.
5. Lot No.
«. Color
7. Recommended Cure Schedule
8. Viscosity (uncut)
9. Reducing Solvent
10. Vol. of Solvent Put into
Vol. Paint
il. Viscosity - Spray (cut)*
12. Ht./Gallon - Spray
13. Ht. Solids - Siray
14. itosistlvity o.- ' (inductance
fr f f*f,i *>' if—ff /*. •»•> £'**
Black Enanal
-------
Data Sheet 3
Paint Spray and Peripheral Equipment Specifications
Teat Datei
Teat No.t
ft'V Lf
Data by/Chkd bys
A. Paint Supply Tank
Type
Manufacturer
Model No.
Serial No.
Rated Capacity, gal
B. Paint Spray Equipment
1. Type
2. Manufacturer
. Model No.
. Serial No.
. Rated Capacity, cc/aUn
. Air Cap
. Fluid Tip
. Needle
C. aint Spray Booth
Type
Manufacturer
Model No.
Serial No.
. Rated Capacity, cfn
D. onveyor
E.
*o»
BO76O5 (DKO. Ho.)
Manufacturer
Model No.
Serial Ho.
orci*d Draft Oven
Type
Manufacturer
M— A ^ | Ujk
nCySV 1 1*0 •
Serial No.
aint Heaters
»«li«nc« Zltctrie Co.
Manufacturer
Model No.
Serial No.
1I»41 Cet»troll«r 1-3M1-A Own
n/a
138
-------
APPENDIX D
AAE TEST EQUIPMENT AND PAINT SPECIFICATIONS
Data Sheet 1
Test Equipment Specifications
Test Datei Test No. t Data ^Checked by:
~ - 2.3L
A. Height Percent Solids Measurement Equipment
1. Laboratory Scales
a. Manufacturer pr»ci«»
b. HoJel No. 340-?1
e. Serial No.
d. Capacity, g o-ioo/o-iooo CBAMS
e. Rated accuracy, g .oir. i^.oiutinn/.ir
Foil Dishes
Almdnia
b. Size Sflpa round
3. Syringe
•• Type
b. Capacity, mL
4. Solvent Type
B. Conveyor Speed Measurement Equipment
1. Rule
b. Graduations i/n-
2. Electronic Timer
a. Type EI«ctro-M«eh«nie«l Dioital
b. Manufacturer »»rn«.
b. Model No. CHAT with DIP itt
e. Serial No.
d. Capacity, kg a tn 1000 c/mn. .KB «a 13 u/Mia.
• . Rated accuracy, g »i - .4% •».•»<,»
2. Stopwatch
a. Manufacturer ammu »y«p«««nn 9*04.,***. jj,c
b. Model NO. -
e. Serial No. -
d. Rated accuracy, s .01
D. Target Foil
1. Type MUB. Mlov 1145-0
2. Nominal Thickness, mils
3. Temper mdiia
E. Wet Film Measurement Equipment
a. Manufacturer earteo o-4 nil*
b. Model No. Fraction Diraet
F. Dry Film Measurement Equipment
a. Manufacturer n»r«i»ko Corp.
b. Model No. ": »n«etor 2000 .1 »xl» Aeeuraev
139
-------
6.3 Select coating. The same coating shall be used for all
tcsti in thii program. Using tat Sheet 2, document
the paint characteristics. Paint characteristics shall
be documented daily, at each addition of paint, ar.d at
other times as requested by the CEHTEC vnginser or CRACO
representative. Again, check your information and sign
the torn.
Data Sheet 2
Paint Specifications
Test Date t
J
Test No.i
MC-11
1. Paint Type
2. Resin Type
3. Manufacturer
4. Manufacturer's Paint ID No.
5. Lot No.
6. Color
7. RecoKBsndeJ Cure Schedule
S. Viscosity (uncut)
9. Reducing Solvent
10. Vol. of Solvent Put Into
Vol. Paint
11. Viscosity - Spray (cutI*
12. Ht./Gallon - Sprey
13. Wt. Solids - Spray
.i. Resistivity or Conductance
Data by/Checked by:
•lack &«•»! 5PTito»e> CUD t *P
\/
(vol) solvent in
(vol1 paint
CUP
Ibt/oal
yt.
nn.
•Use ASTM D-23SJ-68, ASTN D-1200-70. or ASTM D-3794 part 6.
140
-------
Data Sheet 3
Paint Spray and Peripheral Equipment Speeifieationa
Teat HO.i Data by/Chkd byt
Test Datei
a. Paint Supply Tank
Type
Hanuf acturer
Model Ho.
Serial No.
Rated Capacity,
gal
jCji.
B. Paint Spray Equipment
1. Type
2. Manufacturer
3. Model No.
4. Serial No.
5. Rated Ctpacity, ec/min
(. Air O?
7. Fluid Tip
8. Needle
C. Paint Spray Booth
1. Type
2. Manufacturer
3. Model No.
4. Serial No.
5. Rated Capacity, cln
D. Conveyor
1. Type
2. Manufacturer
3. Model No.
4. Serial Ho.
C. Forced Draft Oven
1. Type
2. Manufacturer
3. Model No.
4. Serial No.
r. Paint Heater*
1. Type
2. Manufacturer
3. Model No.
4. Serial No.
tiff. J.lf\t7t3
Lit/ S
5 f - 7/.C
****>>
B0760S (Dun. He.)
IUli*nc« El«etric Co.
m-^.fc y j Drive
O*«n
14 i
-------
APPENDIX E
ALC TEST EQUIPMENT AND PAINT SPECIFICATIONS
Test Date:
2 -ii-s-i
Data Sheet 1
Test Equipment Specifications
Test No. I
by^Checked by:
Model No.
F. Dry Film Measurement Equipment
a. Manufacturer
b. Model No.
240-21
Aluminum
58mm round
conventional
1/32'
A. Height Percent Solids Measurement Equipment
1. Laboratory Scales
a. Manufacturer
b. Model Ko.
c. Serial No.
d. Capacity, g
e. Rated accuracy, g
2. roil Dishes
a. Type
b. Size
3. Syringe
a. Type
b. Capacity, «L
4. Solvent Type
B. Conveyor Speed Measurement Equipment
1. Rule
a. Type
b. Graduations
2. Electronic Timer
a. Type
b. Manufacturer
c. Model No.
d. Serial No.
e. Rated accuracy, a
C. Mass Plow Measurement Eqjipment
1. Platform Scales
a. Manufacturer f
b. Model Ko.
c. Serial No.
d. Capacity, kg
e. Rated accuracy, g *i - .«%
2. Stopwatch
a. Manufacturer
b. Model No.
e. Serial No.
d. Rated accuracy, s
D. Target roil
1. Type
2. Nominal Thickness, • is
3. Temper
t. Wet Film Measurement Equipment
a. Manufacturer
b.
El«etro-?larti«nieal Dieital
Prccuion Scientific Co.
69333
flaw jtfT;
f1irrril>Ct ion
with P10 Bt Indicator
O *n 100O e/Htn. .68 to 13 KC/Mifl.
Inc.
-fll.
»\um. fclloy 1K5-0
1.5
Madtuo
Carclco 0-4
142
Praciaien Direct R»«dlna
OtFeliko Corp.
Pooiteetor 2000 .1 miIt Accuracy
-------
6.3 Select coating. The same coating shall be used for all
tests in this program. Uaing Data Sheet 2, document
the paint characteristics. Paint characteristics ahall
be documented daily, At each addition of paint, and at
other tines as requested by the CCNTEC engineer or CRACO
representative. Again, check your information and sign
the form.
Data Sheet 2
Paint Specifications
Test Datei
Test No.i
1. Paint Type
2. Resin Type
3. Manufacturer
4. Manufacturer's Paint ID No.
5. Lot No.
6. Color
7. Recommended Cure Schedule
8. Viscosity (uncut)
9. Reducing Solvent
10. Vol. of Solvent Put into
Vol. Paint
11. Viscosity - Sprsy (cut)*
12. Wt./Gallon - Spray
13. Mt. Solids - Spray
14. Rtaistivity or Conductance
Dsta by/Checked by:
•lack Ena»»l (Craco •077-001)
(.'A//763 *-
M»li«nc«
210-3159
I: H 12
4. -*»-
Black
min. 9 3»9 »f
MilUOVO 66 S«C. I2ZAHN » 77'P
sec.i Ford Cup •
(vol) solvent in
(vol) paint
Ibs/aal
•Use ASTM D-23S3-68, AST* D-l200-70, or ASTM 0-3794 part 6.
143
-------
Data Sheet 3
Paint Spray and Peripheral Equipment Specifications
Teat Datet Teat No.i
Data by/Child byi
A. Paint Supply Tank
1 Type
Manufacturer
Model No.
Serial No.
Rated Capacity, gal
B. Paint Spray equipment
1. Type
2. Manufacturer
I. Model No.
4. serial No.
S. Rated Capacity, cc/aun
6. Air Cap
7. Fluid Tip
8. Needle
C. Paint Spray Booth
1. Type
2. Manufacturer
1. Model No.
4. Serial No.
5. Rated Capacity, cfn
0. Conveyor
1. Type
2. Manufacturer
3. Model No.
4. serial NO.
B. Forced Draft Ov«n
1. Type
2. Manufacturer
3. Model No.
4. Serial No.
F. Paint Heatera
1. Type
2. Manufacturer
3. Model No.
4. serial No.
Dvt.jtsrnpicll.or IUt»r Wash
B07605 CPwo. No.)
OV»rh«»d
»>li«ne« Electric Co.
Mltmjtk V.S Driva
Air C«« Ovtn
SvBt»m« Connanv
. .-ntroller 1-2341-* Or n
n/a
n/a
n/a
144
-------
APPENDIX F
ALE TEST EQUIPMENT AMD PAINT SPECIFICATIONS
Data Sheet 1
Te«t Equipment Specifications
Test Dstei
*• x
A. Height Percent Solids Measurement Equipment
1. Laboratory Scales
a. Manufacturer PT«.
b. Model No. J4p. 31
c. Serial No.
d. Capacity, g
e. Rated accuracy, g .me p..Biu^M/.ie »..niut<,»
2. Foil Dishis
a. Type fcluainuBi
b. Site iBm round
3. Syringe
a. Type
b. Capacity, mL _S
4. Solvent Type _J
B. Conveyor Speed Meaaurement Equipment
1. Rule
Type e
Graduations
2.
C. Haaa
1.
lectronic Timer
Type Il»ctro-H»ehanic«l Digital
Manufacturer frtemcn Scientific Co.
Model No. 1*330
Serial No. -
Rated accuracy, a .1 »te.
flow Measurement Equlpnent
letters Scales
Manufacture! iu«« MB» M»»-«T. mrl.»-^n~.
Model No. P?*f yritfc C1° *"
Serial No. Mas
Capacity, kg fl_t
Rated accuracy, g «i .
2. topvateh
Haimia. ACtUiTtteT ^^gQ^*B^^2*B£n^^£f^^Q£££^£tl) t XllCe
Model No. -
Serial No. -
Rated accuracy, a .01
D. Target r>! 1
1. Type Mi». Alloy 1145-C
2. NO»I.-» Thickness, mils i.i
3. Temper Mtdiua
t. Wet rilm Meaaurement Equipment
a. Manufacturer cardce 0-4 MI•
b. Model NO. »r«ei«ioq Pjrjrt P"J;'"
r. Dry rilin Measurement Equipment
a. Manufacturer Dtr«i«ke Corp.
b. Model No. >o»it«ctor 2000 .1 «il» Accuracy
145
-------
6.3 Select coating. The same coating shall be used for all
tests in this progran. Using Data Sheet 3, docunent
the paint characteristics. Paint characteristics shall
be documented daily, at each addition of paint, and at
other times as requested by the CENTEC engineer or GRACO
representative. Again, check your Information and sign
the torn.
Data Sheet 2
Paint Specification!
Test Datei
Test No.i
&iP 7-3
Beta by/Cheeked byt
1. Paint Type
2. Resin Type
3. Manufacturer
4. Manufacturer's Paint 10 No.
5. Lot No.
6. Color
7. Recommended Cure Schedule
I. Viscosity (uncut)
9. Reducing Solvent
10. Vol. of Solvent Put into
Vol. Paint
11. Viecoeity - Spray (cut)*
12. Mt./Gallon - Spray
13. *t. Solids - Spray
14. Resistivity or Conductance
»l«cfc Enaaal (Grace 1077-0011
Alkyd •«••
mtliance
21C-31S3
•lack
min. a
0* MC. I2ZM1N ff 71
sec.t Pord Cup •
(voll eel vent In
. Itt't -t"
(vol) paint
»ee.>
Ford Cup f *r
.•',-o
•Use ASTH D-23S3-68, ASTN 0-1200-70, or ASTN P-3TV4 pert t.
146
-------
Data Sheet 3
Paint Sorav and Peripheral Equipment Specifications
Test Date:
Teat .No. t
J£L_
Data by/Child Oyi
A. Paint Supply Tank
Type
Manufacturer
Model No.
Serial No.
Rated Capacity, gal
•aflfe
«3gP
B. Paint Spray Equipment
1. Type
2. Manufacturer
1. Model No.
4. Serial No.
5. Hated Capacity, ce/min
7. Fluid Tip
B. Needle
C. Paint Fpray Booth
». Type
2. Manufacturer
1. Model No.
4. Serial No.
5. Rated Capacity, eta
D. Conveyor
1. Type
2. Manufacturer
). Model He.
4. Serial No.
E. Forced Draft Oven
1. Type
2. Manufacturer
>. Model No.
4. Serial No.
F. Paint >i.ater»
1. Type
2. Manufacturer
3. Model No.
4. serial No.
XP"~
jC^^U^o
^^B-^arAwT
•O7605 (Bug. Bo.)
O»«rh«»d
laetric Co.
V.S t>riv«
rare.
•ft
147
-------
APPENDIX G
Glossary of Statistical Terms
Regression
The procedure of fitting a model to a set of data using the method of
least squares. The product is a prediction equation for predicting a
f""nendent response as a function of independent "input" variables.
Residuals
The error in fit of a regression equation. The residual is the difference
between the observed response and a predicted response from the regression
model.
t-tests
t-tests are used in the present context to test the hypothesis that a
regression coefficient is zero. The t-statistic Is a ratio
regression coefficient
C " standard error of coefficient
Small values of t are evidence of a coefficient that does not differ
significantly from zero.
F-tests
F-tests are used in a manner very similar to the t-tests. For a specific
regression coefficient, and thus for a particular variable, the F-
statistlc represents the ratio of the variance explained by the variable
being tested to the variance attributed to experimental error.
148
-------
Significance level (p value)
The significance level is used in Che context of significance testing.
If a regression coefficient (or a model variable) is significant at the
0.02 level, the value 0.02 is the probability of obtaining a t-statistic
(or F) as large as that observed, when in fact the model variable, plays
absolutely no role in the system. In other words, a p valuo is the prob-
ability of obtaining such information due to chance alone. Clearly, a
small p value is evidence of a strong model term or variable.
Standard Deviation
A standard deviation is a measure of spread in a statistical distribution
or a set of data. Given x., x_, ..., x , observations in a set of data
ii n
and x, the mean, the sample standard deviation is given by
R?- (Coefficient of Determination)
The coefficient of determination R2 is a measure of quality of fit of
fitted model. The statistic R2 is defined as
R2 o variation tn response explained by model
variation in response observed
149
-------
Confidence Limits on Mean Response
In a regression context, Che 95% confidence limits on the mean response
represent "bounds" around the fitted regression that are defined such
that
"we are 95% confident that the mean response, at rhe data
locations in question, is covered by the bounds."
Lack of Fi'. Test
The lack of fit test is an F-test for ascertaining whether or not a
fitted model is adequate. The test essentially tests for the signifi-
cance of higher order terms in the egression. If the F-statistic is
nonsignificant, the conclusion is that there is no evidence that a more
complicated model would improve the regression.
Dummy Variables
The use of dummy variables is a standard way of accommodating "categories"
in a regression situation that also contains the ordinary continuous
variables.
------- |