United States
Environmental Protection
Agency
Air and Energy Engineering
Research Laboratory
Research Triangle Park NC 27711
K^
1
Research and Development
EPA/600/S2-85/064 Aug. 1985
Project Summary
Destruction of VOCs by a
Catalytic Paint Drying (IR)
Device
C. David Cooper
Volatile Organic Compounds (VOCs)
are emitted from a number of different
types of sources, one of which is paint
drying and curing ovens. A device that
catalytically oxidizes fuel to generate
infrared (IR) radiation recently has been
introduced by SUNKISS Thermoreac-
tors. Inc., as a new technology for paint
drying and curing. During its operation,
the device also oxidizes some of the
paint solvents that are evaporated in
the oven, thus reducing overall emis-
sions of VOCs into the atmosphere.
A system was constructed at the Uni-
versity of Central Florida (UCF) to ex-
perimentally test the SUNKISS device,
to quantify its VOC destruction capabil-
ities. The system consists of flow con-
trol and measuring devices, analytical
equipment, and aim3 chamber in
which the SUNKISS device was
mounted. Three paint solvents—
hexane, methyl ethyl ketone (MEK),
and toluene—were dispersed in air
streams that flowed through the cham-
ber at various rates. The solvent de-
struction efficiency of the device was
shown to be a strong function of the air
residence time in the chamber, but was
independent of VOC concentration.
Hexane and MEK behaved similarly,
while toluene was more reactive. Ob-
served VOC destruction efficiencies
ranged from less than 20 to over 50%.
Some slight degree of fuel non-
combustion was observed, as was a
very small amount of CO generation.
This Project Summary was devel-
oped by EPA's Air and Energy Engineer-
ing Research Laboratory, Research Tri-
angle Park, NC, to announce key
findings of the research project that is
fully documented in a separate report
of the same title (see Project Report or-
dering information at back).
Introduction
" Volatile Organic Compounds (VOCs)
are emitted from a number of different
types of sources, one of which is paint
drying and curing ovens. In conven-
tional ovens, items are dried by hot air
or by electric infrared (IR) lamps. The
solvents that evaporate are discharged
to the atmosphere with the dryer efflu-
ent air unless that air is routed through
a Final Control Device (FCD), such as an
incinerator. Solvent concentrations in
the oven exhaust air are typically 100-
1000 ppm.
The amounts and types of VOC emit-
ted from any process depend greatly on
the type of paint, the type of process,
and the configuration and rate of air
flow through the paint booth, flash
dryer, and oven. Many VOCs are photo-
chemically reactive and can contribute
to high oxidant levels in the ambient air.
As a result, many large paint drying and
curing operations have been required to
reduce emissions of VOC to the atmos-
phere.
Recently, a novel type of paint drying
device was introduced commercially. It
oxidizes fuel gas on a catalyst and emits
IR radiation in wavelengths adsorbed
by paint pigments. A number of such
devices are placed in the oven and ori-
ented toward the painted parts to dry
and cure the pairit. According to tests by
the manufacturer, the SUNKISS units
achieve significant energy savings com-
-------
pared to conventional hot air or IR (elec-
tric) lamp ovens. Recent installations of
the SUNKISS system include several
automotive, bus, and truck painting fa-
cilities in the U.S. and Canada.
Another benefit of the device is that
some of the solvent vapors in the oven
are oxidized on the catalytic surface dur-
ing the drying process. Because of this
concurrent VOC destruction, total mass
emission rates of VOC from the oven
can be reduced substantially. The de-
gree of VOC destruction was of interest
to the U.S. EPA and the American Elec-
troplaters' Society because of the po-
tential benefits of the device to the envi-
ronment and to industry.
The main objective of the study con-
ducted at UCF was to quantify the VOC
destruction efficiency of the SUNKISS
device under controlled laboratory con-
ditions. Secondary objectives were (1)
to test the effect on the VOC destruction
efficiency of a painted object placed in
the oven, and (2) to test the effective-
ness of recent engineering modifica-
tions (i.e., a higher pressure fuel jet and
thicker catalyst pad) on the VOC de-
struction efficiency.
Experimental Equipment and
Procedures
System Description
The experimental apparatus, con-
structed specially for this project, is
shown schematically in Figure 1. Basi-
cally, the equipment consists of air flow
measurement and control devices, a
solvent saturator system, a reaction
chamber, and sampling and analytical
apparatus. House air, filtered through
glass wool, silica gel, and activated car-
bon, flowed through at least two ro-
tameters in parallel. A measured rate of
air, bubbled through liquid solvent in a
double saturator system, produced a
saturated stream of solvent vapor in air
at 0°C. That stream, diluted by another
measured flow of pure air to create a
stream of air at a desired concentration
and desired total flow rate, was fed into
the reaction chamber.
The reaction chamber is a welded
steel box enclosing about 1 m3 of space.
The SUNKISS device was mounted on
the side opposite the air inlet, and an
exhaust stack was mounted on the top
on the inlet end of the chamber. The
SUNKISS unit has an internal 60 cfm
(1,700 l/min) fan that provided some
mixing of the air in the chamber, and
was augmented by a 200 cfm (5,700
l/min) fan mounted coaxially with the air
flow inlet. The SUNKISS fan blew cham-
ber air across the front face of the cata-
lyst pad; some of the VOC was ab-
sorbed and oxidized on the surface of
the catalyst.
Several sample lines and thermocou-
ples were at strategic locations in the
chamber. Samples were withdrawn by
a vacuum pump through a gas sam-
pling valve mounted on an ANTEK 2400
chromatograph equipped with a flame
ionization detector. Samples were also
routed to a Thermoelectron Model 48
CO analyzer.
Flow Measurement and Con-
trol
Air flow was measured and con-
trolled, using manually controlled ro-
tameters, each calibrated carefully with
either an accurate wet test meter or an
accurate dry gas meter. UCF house air
first was passed through a filter package
consisting of a layer of glass wool, fol-
lowed by granular activated carbon, sil-
ica gel, and glass wool. The air was then
split into four possible streams, the flow
rate of each of which could be individu-
ally measured and adjusted. Piping per-
mitted the main air flow rate to bypass
the solvent saturator system, while one,
two, or three of the other streams were
routed through the saturator. Thus, al-
most any combination of total air flow
rate (2-8 cfm; 57-227 l/min) and solvent-
in-air concentration from 200 to 2000
ppm could be created and maintained
during the operation.
Solvent Saturator System
The solvent saturator system pro-
duced a reproducible and reasonably
Exhaust
, Thermoreactor
CO
Analyzer
Vacuum Pump
Figure 1. Schematic flow diagram of system.
2
-------
accurately known concentration of VOC
in air. A wide range of concentrations
could be produced by blending a stream
of solvent-free air with an indepen-
dently controlled stream of completely
saturated air.
Primary air was passed through a
dual saturator system consisting of two
1000 ml Erlenmeyer flasks, Teflon tube
bubblers, and a copper cooling coil. Sat-
uration of 40-50% was obtained in the
first stage which operated at 25°C. The
air then was passed through copper
coils submerged in a 10°C water bath
and then into a second saturator, sur-
rounded by a 0°C water/ice bath. In the
second stage, 100% saturation was
achieved due to additional vapor/liquid
contact, and probable condensation of
VOC picked up in the first saturator. Di-
lution air was metered through a sec-
ond calibrated rotameter, bypassing the
saturator system, and was blended with
the saturated stream to achieve des'--pd
concentrations. A mixing tank ensured
complete mixing prior to the combined
stream's flowing to the reaction cham-
ber.
Reaction Chamber
The chamber was designed to en-
close about 1m3 of space. The chamber
was built from 11-gauge steel welded at
all seams. Thus, it was structurally
sound, leakproof, able to withstand all
possible operating temperatures, and
did not have any potential for self-
emission of fumes (as plywood or com-
posites might). The SUNKISS unit was
mounted at the end of the chamber, op-
posite the air entrance.
The inlet gas (the premixed stream of
air with a low concentration of VOC) en-
tered through a diffuser-type entrance
near the bottom of the chamber and di-
rectly across from the SUNKISS unit.
Both the total inlet gas flow rate (2-8
cfm; 57-227 l/min) and the inlet VOC
concentration (from about 200 to 2000
ppm) could be independently con-
trolled. Inlet and outlet gases were mon-
itored for concentration and tempera-
ture (sample ports built into the
chamber enabled samples to be taken
from within the chamber).
Test Procedure
A typical experimental run is de-
scribed below. All instruments (i.e., gas
chromatograph, Co analyzer, chart
recorder) were allowed sufficient time
for warmup and stabilization. The ap-
ropriate solvent was poured into the
aturators, and all fittings were checked
for leaks. The GC column temperature
was adjusted (based on experience in
this study) to provide the most consis-
tent and accurate results for each sol-
vent used.
The individual stream flow rates to
achieve desired solvent concentrations
and total inlet air flow rates were ad-
justed via inline rotameters. To ensure
representative results, steady state con-
ditions had to be maintained in the test
chamber. The chamber average resi-
dence time, tp (chamber volume/air flow
rate), varied for the different flow rates.
At each different flow rate, sample test-
ing was delayed a period of 4 x tR.
Each time a sample was analyzed for
VOC, three or four samples at 1 minute
intervals were analyzed, and the aver-
age of the peak heights on the chart
recorder was allowed to represent the
true VOC concentration. Chamber inlet
and outlet samples were taken. The
mean of the outlet solvent peaks was
compared to the mean of the inlet tank
peaks, and both were recorded. CO was
also measured. Propane peaks were
also observed and recorded. VOC de-
struction efficiencies were calculated
based on inlet air molar flow rate x VOC
inlet concentration vs. estimated outlet
gas molar flow rate (including propane
combustion products plus any en-
trained air) x outlet VOC concentration.
The above procedure was repeated at
least three times for each specific test
(each air flow rate and concentration),
and a final destruction efficiency was
calculated as the average of all the repli-
cations.
Results and Discussion
Initial Operating Tests
The SUNKISS unit was initially tested
with the SUNKISS unit on with only
fresh air (no VOC) entering the cham-
ber. A VOC response was observed on
the GC, which implied that some un-
combusted propane fuel might be es-
caping into the chamber. Introduction
of some hexane into the inlet air and
observation of two separate peaks on
the GC confirmed that propane gas was
indeed entering the chamber. Dis-
appearance of one peak when the
propane fuel valve was shut further
proved that one peak was propane.
Recognition of the problem of propane
non-combustion led to more detailed
testing during the solvent destruction
efficiency tests. In later tests, the per-
cent propane combustion was calcu-
lated to be 98-99.8%, depending on the
total air flow rate. Also, modifications of
the SUNKISS device during the latter
part of this research partially alleviated
this problem of propane non-
combustion. Quantitative propane re-
sults will be presented later.
Effect of Solvent Concentration
Several tests were conducted at 170
1/min (6 cfm) air flow, but at different
inlet concentrations of hexane in air. Ba-
sically, chamber inlet and outlet (and
some interior point) concentrations
were measured to determine the per-
centage destruction efficiency of the
SUNKISS unit. Because the total molar
flow rate out of the chamber was not
equal to the inflow rate of solvent laden
air (the fuel gas plus entrained ambient
air also entered the chamber), the VOC
destruction efficiency was calculated
as:
n0C0 - neCe
n0C0
neCe
nnC0
(1)
where
n0 = inlet molar flow rate of stream of
air with VOC
C0 = inlet concentration
ne = total molar flow rate of exhaust
gases
Ce = exit concentration
Equation (1) properly accounts for the
dilution effect; i.e., that VOC concentra-
tions in the outlet stream will be some-
what lower than in the inlet due solely
to dilution by the addition of other
gases.
It was determined from data supplied
by the manufacturer that the propane
injection rate was about 0.087 gmole/
min. Furthermore, based on data from
the manufacturer, the ambient air en-
trained by the inlet jet of the fuel (the
primary air) was calculated to be about
0.133 gmole/min. Complete propane
combustion produces 7 moles of gases
for 6 consumed. Thus, the total molar
flow of gases exiting the chamber wa.s
equal to the sum of the moles of solvent
laden air flowing in, plus 7/6 times the
propane flow, plus the entrained ambi-
ent air flow. In other words, the molar
flow rate out was equal to moles of sol-
vent laden air in, plus 0.23 gmoles/min.
Molar expansion due to oxidation of the
VOC was ignored.
Several individual solvent destruction
tests were carried out using hexane in
air. All replications of individual tests
were quite reproducible. Certain con-
-------
40 p
30
I
,2
i
Q
20
10
®
©
©
- ©
©
Figure 2.
500 1000 1500 2000
Intel Hexane Concentration, ppm
Hexane destruction efficiency vs. inlet hexane concentration.
2500
centrations of hexane-in-air and certain
air flow rates into the chamber were
replicated as many as five or six times.
The destruction efficiency for each test
never varied more than 2 percentage
points from the mean. The results are
shown in Figure 2.
Effect of Chamber Residence
Time
After it was shown that destruction
efficiency was essentially independent
of inlet hexane concentration, the effect
of inlet air flow rate was investigated.
For a fixed volume chamber, the volu-
metric air flow rate into the chamber is
related to the inverse of the average res-
idence time. The average residence
time in a CSTR is actually best calcu-
lated by the volume of the chamber, di-
vided by the exhaust (as opposed to the
inlet) gas rate. In these experiments, the
exhaust gas volumetric flow rate was
higher than the measured air inflow rate
due to the fuel and primary air inflow,
and due to the volume expansion of the
gas in the chamber caused by increased
temperature. The ideal gas law was
used to calculate the outlet exhaust vol-
umetric flow.
Effect of Solvent Type
In addition to hexane, two other com-
mon paint solvents were tested: methyl
ethyl ketone (MEK) and toluene. In es-
ence, the MEK destruction efficiencies
nearly duplicated those observed with
hexane. The MEK data trended a slight
bit lower than hexane and showed a
slight bit more scatter, but the same es-
sential behavior was evidenced. The re-
sults for toluene showed significantly
more solvent destruction than with
either hexane or MEK. All the destruc-
tion efficiencies are plotted vs. resi-
dence time in Figure 3.
Painted Object Tests
As part of a planned attempt to more
closely simulate conditions in a real
oven, a painted (black) metal plate was
placed in the test chamber during the
toluene tests. The plate measured ap-
proximately 40 by 65 cm, and was
placed normal to the IR from the cata-
lyst pad about 40 cm from the device.
The plate simulated a painted part being
cured. Destruction efficiencies were in-
creased only about 2 percentage points
over what had previously been ob-
served for any given solvent at any
given air flow rate.
Two similar steel plates (one was
painted and the other was not) were
placed in the oven. To each plate, a ther-
mocouple was attached to measure sur-
face temperature. At the start of the test,
a steady stream of air was flowing
through the oven, and the SUNKISS
unit was turned on. The temperature of
the painted plate's surface increased
quickly to a significantly higher temper-
ature than that of the unpainted plate.
At steady state, the painted plate was
about 40°C hotter than the unpainted
plate, which essentially achieved the in-
side chamber air temperature.
Effect of SUNKISS Device Mod-
ifications
During the course of this research, the
manufacturer modified the basic device
to increase solvent destruction effi-
ciency and to decrease residual (un-
burned) propane concentrations. The
modifications were basically: (1)
smaller fuel jets with an increased fuel
supply pressure, and (2) a thicker cata-
lyst pad with more catalyst surface area.
Each modification was tested in the UCF
experimental test system.
To install the new system, the old cat-
alyst pad and the old fuel injection ori-
fice were replaced with the new ones.
The propane fuel pressure was in-
creased from 11 in. H20 to 2 psig (3 to 14
kPa). Hexane was used as the solvent,
and several air flow rates were tested
(all at an inlet hexane concentration of
1000 ppm). Three replications were
made at each of six flow rates, and hex-
ane destruction efficiencies were deter-
mined. Also, propane and CO concen-
trations were measured at each inlet air
flow rate.
The results of the above tests showed
a slight (but non-negligible) improve-
ment in VOC destruction. In addition,
propane non-combustion with the new
system (new pad-new jets) was reduced
considerably at all flow rates, compared
with that observed with the old system.
CO concentrations were lower at the
high air flow rates, but higher at the low
air flow rates.
In calculating the destruction efficien-
cies, the increase in dilution effect due
to the new jets was accounted for. The
new jets were designed to pass the
same molar rate of fuel into the
SUNKISS device, but at a higher veloc-
ity. The higher velocity entrains more
ambient air in through the back of the
catalyst pad and permits more com-
plete combustion of propane. However,
-------
the additional air adds to the total volu-
metric flow of gases through the cham-
ber, thus diluting the concentrations of
VOC, propane, and CO somewhat com-
pared with the original system.
Estimates of the air entrainment rate
of the newly modified system were ob-
tained from SUNKISS, Inc. It had been
estimated previously that, with a pro-
pane supply pressure of 11 in. H20
(3 kPa) and with the old jets, about 0.113
cfm (about 0.13 gmoles/min) of ambient
air was drawn into the chamber. With
the new jets, at 2 psig (14 kPa), the ambi-
ent air entrainment rate was about 0.663
cfm or 0.78 gmoles/min. At the lower air
flow rates into the chamber (2, 3, or
4 cfm: 57, 85, or 113 l/min), there is sig-
nificant dilution due to the additional
primary air brought in with the new sys-
tem. (Note that the calculations of de-
struction efficiencies that follow depend
60 h
on the above estimates of air entrain-
ment rates.) The outlet air flow rates
from the chamber were not measured in
this work.
After accounting for the additional air
dilution due to the new jets, the VOC
destruction efficiencies were calculated
for the new jets—new pad system. Also,
because chamber residence times at
each inlet air flow rate were decreased
due to the additional primary air flow,
the residence times were also recalcu-
lated. The fairest comparison is to plot
the efficiencies against gas residence
time as shown in Figure 4. This plot
shows that VOC destruction efficiencies
did improve with the new system.
To fairly compare propane combus-
tion under the new system vs. the old
system, the molar flow rates of propane
out of the chamber were calculated
under each system. The rate of propane
50
40
.5
.G
i
o
§
30
20
10
© Hexane
Q Toluene
A. MEK
I
I
I
Figure 3.
2468
Residence Time ft R). minutes
VOC destruction efficiency vs. gas residence time.
10
12
into the SUNKISS device was assumed
constant at 0.10 gmole/min. The percent
of propane combusted was calculated
for each system and compared for each
inlet air flow rate.
To fairly compare CO production
under each system, the CO generated
was expressed as a percentage of all the
carbon atoms combusted. That is, first
the molar rate of hexane that was de-
stroyed in the chamber was calculated.
Then, that rate was multiplied by the
number of carbon atoms per molecule
of hexane (six). Next, the molar propane
combustion rate was multiplied by the
number of carbons in propane (three).
The two numbers were summed to give
a total carbon atom combustion rate.
The molar flow rate of CO exiting the
chamber was calculated and divided by
the total carbon atom combustion rate
to give the CO generation percentage.
The propane and CO results are given in
Table 1.
Conclusions and Recommenda-
tions
The solvent destruction capability of
the SUNKISS thermoreactor has been
tested under laboratory conditions. This
catalytic device has been shown to be
effective in oxidizing VOC in air in a
completely mixed chamber with de-
struction efficiencies primarily depen-
dent on air flow rates (residence times)
in the chamber. Observed destruction
efficiencies ranged from less than 20%
to above 50%. The destruction efficien-
cies were essentially independent of
VOC concentration, but were higher for
toluene than for either hexane or MEK.
These destruction efficiencies were sig-
nificantly lower than those typical of a
final control device (such as a vapor in-
cinerator), but may be high enough to
contribute significantly to an overall
VOC control program.
Some slight degrees of fuel non-
combustion and a very small amount of
CO generation were observed, particu-
larly at the lower total air flow rates. The
IR generated by the device is absorbed
by a painted surface resulting in higher
surface temperatures than for un-
painted steel, thus enhancing the drying
and curing process. The newer
SUNKISS system (thicker pad, smaller
fuel jets, and higher fuel pressure) re-
sulted in better VOC destruction effi-
ciencies and in more complete propane
combustion and is recommended over
the older system. The SUNKISS system
can dry and cure painted objects and
-------
I
.§
I
I
%
0.07
0.05
0.06
0.06
0.05
0.05
Based on molar flow rates of propane into and out of the chamber.
Based on total moles of carbon atoms combusted.
help reduce VOC emissions to the
atmosphere. It is worthy of considera-
tion as part of any proposed operation
that involves solvent evaporation inside
enclosures.
•&U. S. GOVERNMENT PRINTING OFFICE-.1985/559-111/20646
-------
C. Cooper is with the University of Central Florida, Orlando, FL 32816.
Charles H. Darvin is the EPA Project Officer (see below).
The complete report, entitled "Destruction of VOCs by a Catalytic Paint Drying(IR)
Device," (Order No. PB 85-215 333/AS; Cost: $ 10.00, subject to change) will be
available only from:
National Technical Information Service
5285 Port Royal Road
Springfield. VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S2-85/064
OOOC329 PS
U S ENVIR PROTECTION AGENCY
REGION 5 LIERARK
230 S DEAR8CRN STREET
CHICAGO IL
------- |