CHARACTERIZATION OF OZONE EMISSIONS FROM
AIR CLEANERS EQUIPPED WITH OZONE
GENERATORS AND SENSOR AND FEEDBACK
CONTROL CIRCUITRY.
Mark A. Mason1, Leslie E. Sparks2, Scott A. Moore3, Ivan Dolgov4, Richard B. Perry5. (1) U.S.
EPA, Office of Research and Development, National Risk Management Research Laboratory,
Research Triangle Park, NC 27711, (919) 541-4835, mason.mark@epa.gov: (2) National Risk
Management Research Laboratory, Research Triangle Park, NC 27711, (919) 541-2458,
sparks.les@epa.gov; (3) National Risk Management Research Laboratory, Research Triangle
Park, NC 27711, (919) 541-5104, moore.scott@epa.gov; (4) National Risk Management
Research Laboratory, Research Triangle Park, NC 27711, (919) 541-2116,
dolgov.ivan@epa.gov; (5) National Risk Management Research Laboratory, Research Triangle
Park, NC 27711, (919) 541-2721, perry.richard@epa.gov.
ABSTRACT
Ozone emission rates of several consumer appliances that are marketed as indoor "air treatment"
or "air purification" systems were determined in a well-characterized 30 m3 environmental
chamber. Additional tests were conducted in a research test house to determine the effects of
appliances on research house ozone concentrations and to evaluate the performance of an ozone
generator feedback control system. Ozone emission rates determined in the- chamber studies
ranged from 0.5 to >150 mg h"1 for one appliance at 50% relative humidity (RH). Ozone
emission rates increased as RH decreased for all models tested. Nitrogen dioxide (NO,)
generation rates ranged from 0.06 to 0.16 times the ozone emission rate and varied among
models of different manufacturers. When chamber-derived ozone emission rates were input into
the RISK model, the model provided reasonable estimates of the time history of ozone
concentrations that were measured in a research house. The research house experiments
demonstrated that these types of devices are capable of producing ozone concentrations above
accepted health guidelines. Ozone concentrations in the research house stayed below the stated
control level of 50 parts per billion (ppb) when models equipped with a feedback control system
were operated in the control mode.
INTRODUCTION
Ozone in the indoor air is due to infiltration, active transport of polluted outdoor air by heating,
ventilation, and air-conditioning systems (HVAC), and indoor sources that generate ozone such
as office equipment and machines1*2'3'4. Consumer appliances marketed as air treatment devices
or air purifiers that intentionally produce ozone are an additional important source, due to their
potential to create high ozone concentrations5 and exposures6.
Ambient and workplace levels of ozone are regulated by various government organizations. For
example, the Food and Drug Administration (FDA) requires output of medical devices to be no
more than 0.05 part per million (ppm), and the Occupational Safety and Health Administration
(OSHA) requires that workers not be exposed to an average of greater than 0.10 ppm for 8 h 7'8.
-------
The maximum 8 h average for ambient air, stipulated in the Environmental Protection Agency's
(EPA's) National Ambient Air Quality Standard, is 0.08 ppm9.
The often-stated rationale for introduction of ozone to indoor air is that ozone improves air
quality by controlling or reducing biological and/or chemical contaminants. Though treatment of
drinking water with ozone is effective against many biological contaminants, available evidence
indicates that treatment of contaminated surfaces with airborne ozone has little impact on fungal
spores, even at ozone concentrations that are considered highly toxic to humans10. The reactions
of ozone with ambient pollutants have been well characterized and, for most organic compounds,
reaction rates are too slow to significantly reduce indoor air concentrations11. On the other hand,
ozone reacts with many unsaturated volatile organic compounds (VOCs) at rates that are capable
of reducing indoor contaminant levels, but reaction products are aldehydes, ketones, and organic
acids11'12. Heterogeneous reactions between ozone and indoor surfaces may also result in
increased concentrations of formaldehyde and other aldehydes13. Recent mouse bioassay
experiments indicate a potential for formation of strong airway irritants from the reaction of
ozone with alpha-pinene, a common indoor air contaminant14. Thus, though ozone readily reacts
with unsaturated VOCs, reaction products may degrade rather than improve indoor air quality.
According to one manufacturer's sales brochures, some ozone generating appliances are capable
of producing ozone at rates of 320 mg h"1. Some models of these devices are now equipped with
a sensor and controller that is advertised to maintain in-room ozone concentrations at no more
than 0.05 ppm15. However, there is little data available that demonstrates the performance of the
control system. Also, there is limited data available that evaluates our ability to predict ozone
exposure from carefully conducted performance tests. To better understand, the potential risk due
to use of these appliances, we conducted chamber experiments to characterize ozone emission
rates of selected appliances and investigated the effectiveness of the feedback control system in
preventing the appliance from creating in-room ozone concentrations that exceed 0.05 ppm.
Experiments were conducted in EPA's 30 m3 indoor air quality research chamber and research
house. In pilot studies, we characterized the behavior of ozone in the test chamber and
investigated the ozone emission rates from a single appliance across the range of ozone generator
set points and at three levels of relative humidity (RH). We determined the ozone emission rates
for seven additional appliances at selected set points and determined ozone emission rates at
three levels of RH for one model of each appliance. In the research house, we characterized the
performance of units equipped with the ozone sensor-controller circuit and measured ozone
concentrations in the research house that resulted from operation of selected appliances.
Nitrogen oxide (NOX) emission rates were determined from chamber tests, and NOX
concentrations were monitored in the research house experiments.
METHODS
Characterization of the Test Chamber and Pilot Investigation of Factors
Affecting Ozone Emission Rate
Ozone loss rates to chamber surfaces were determined at three levels of RH and three airflow
rates. Pilot tests were conducted to investigate the range of ozone emission rates and effects of
RH on ozone generation rate.
-------
Determination of Ozone Deposition Rates in the Large Chamber
To determine ozone deposition velocity in the chamber, we dosed the chamber with ozone and a
non-reactive gas, and then monitored the decay of ozone and the non-reactive gas after the
sources were removed. The ozone deposition rate was determined from the apparent increase in
chamber air exchange rate, determined from the ozone concentration decay data, compared to the
air exchange rate determined from concentration decay data of the non-reactive compound16.
Tests were conducted at flows of 30 and 150 m3 h"1 (1 and 5 air changes per hour, ACH) and at
near-static conditions. For the tests conducted at 1 and 5 ACH, ozone was generated by passing
dry air through a commercial NOX generator (Teco Model 100) and injected into the chamber at a
constant rate. The ozone insult was stopped after steady state ozone concentrations were
observed for several hours. A non-reactive tracer compound [sulfur hexafluoride (SF6)j was
periodically injected from a cylinder into the inlet air stream of the chamber. At the chamber
outlet, we monitored the concentration of ozone with an ultraviolet (UV) photometer (TECO
Model 49) by method ASTM D 5156-9517 and monitored the concentration of SF6 with a gas
chromatograph (GC) equipped with automated gas sampling valves and an electron capture
detector (BCD).
For the near-static chamber tests (sealed chamber), ozone was generated by remote activation of
an ozone-generating appliance inside the chamber for a short period of time. SF6 was manually
injected into the chamber with a syringe. A fan mounted from the ceiling continuously mixed the
chamber airi Ozone and SF6 concentrations in the chamber were monitored by sampling through
6 mm OD (1A in.) Teflon tubes placed into the center of the chamber through sampling ports. Air
exchange in the sealed chamber was primarily due to leakage into the chamber as air was
removed through sampling. The chamber is shown in Figure 1.
Figure 1. EPA's room sized indoor air test chamber.
Ozone Monitor
at Air Cleaner Inlet
Chamber Inlet
Air Ozone Monito
Clean Air In
-------
Calculation of Deposition Velocity (v^
The mass balance for a well-mixed chamber is given by Equation 1.
Equation 1. Mass balance equation for the chamber in the absence of an ozone source.
AC —CxO — C-xv x A
U\^ \_^ *** y/ \^ •*• V j ^ /3L:., t
dt V
where: C = concentration of the contaminant (mg m"3); Q = air flow through the chamber
(m3 h"1), determined from the slope of the In of the concentration vs. time plot of the tracer
compound (SF6); yd - deposition velocity (m h"'); ^.^ = area of the sink (57 m2); and V
volume of the chamber (30 m3).
The solution to this equation is:
Equation 2. Solution to Equation 1 .
-(— +vrfx4iLL)«
v v
where: £\ = the initial ozone concentration at the start of the decay (the time that the ozone
insult was terminated) (mg m"3) and t= time from the start of the decay (h).'This expression may
be rewritten as:
Equation 3. Substitution of apparent air exchange into Equation 2.
where Napparent is the apparent air exchange rate (h"1) determined by fitting a straight line to the
ln(C) versus time plot. This equation may be rearranged to determine ozone deposition velocity.
Equation 4. Determination of deposition velocity.
(apparent) ~ xl
vd =
•"•sink
Pilot Investigations of Air Cleaner Performance
The ozone generation rate controller mechanism of the appliance used in the pilot tests, herein
referred to as Manufacturer 1, model A, unit (0) [Ml, A (0)], consists of a control knob
surrounded by three concentric rings. Each ring is marked by dial settings that identify the square
footage of the floor area of the space to be treated. One small, or one or two large ozone
generating plates are installed to treat areas from 20 to 1600 ft2. Pilot experiments were designed
to investigate ozone emission rates as a function of dial setting, RH, fan speed, repeatability at
the same ozone generator set point, stability of ozone emission rate, and NOX generation rate. For
-------
the pilot tests, we tested the appliance in a one small-plate configuration at 20, 60, and 100 ft2, in
a one large-plate configuration at 50, 400, and 800 ft2, and in a two large-plate configuration at
200 and 900 ft2 and at foil scale, with the dial turned beyond the 1600 ft2 setting. Test duration
ranged from 2 to 66 h. A total of 13 pilot tests and one replicate test were conducted. Additional
tests at intermediate set points were conducted later to provide increased confidence in the fit of
ozone emission rate versus set point. The ozone sensor controller was placed in the manual mode
(deactivated) for the pilot tests.
Testing Protocol for Determination of Ozone Emission Rates in the Large Chamber
The appliance was placed on a small table near one wall of the chamber (see Figure 1) facing the
opposite wall. The power cord of the appliance was connected to a cord that passed through a
port in the chamber wall. The chamber was purged with clean, humidified air at the conditions
specified for the test and the chamber was periodically dosed with SF6 from an automated
system. Depending upon ozone generator dial set point, the air exchange rate for the chamber
was set to either 5 or 1 ACH so that the steady state ozone concentration would be within the
calibration range of the ozone monitors. Ozone, NOX, and SF6 concentrations were monitored at
the chamber outlet duct. The ozone concentration was also monitored from a sampling line that
terminated just behind the appliance. Temperature and RH were continuously monitored with
sensors located inside the chamber. A test was started by plugging the appliance power cord into
a monitored 1 10-V AC supply and continued until steady state ozone concentrations were
observed for at least 1 h. The chamber control data system collected 5 min averages of chamber
temperature, RH, and flow. A second data system collected ozone, NOX, and power monitor
output as 1 min averages. Chromatography software implemented on a personal computer
(Chrom Perfect) collected SF6 data from the GC. All data were compiled in spreadsheets, along
with daily zero and span results for each instrument. The ozone emission rate (ER0Z) was
determined from the mean ozone concentration averaged over the period of steady state
concentration by Equation 5.
Equation 5. Calculation of ozone emission rate from chamber test data.
where: ERoZ= ozone emission rate (mg h"1), N = chamber air exchange rate (h"1), C = mean
chamber ozone steady state concentration (mg m"3), vd = mean deposition velocity of ozone
determined from the static tests (m h"1), A = chamber interior surface area (m2), and V =
chamber volume (m3).
Chamber Characterization of Selected Appliances and Appliance
Performance in the Research Test House
Following the pilot tests, we characterized the performance of seven appliances that included two
models (A and B) from two manufacturers (Ml and M2). Two units of each model were tested,
except for M2, model A. For the seven appliances, in the chamber, we characterized ozone
emission rates at low, medium, and high ozone generation set points, and low, medium, and high
RH levels. In the research house we characterized the dead band for sensor-controller equipped
appliances (Ml, models A and B), and characterized ozone concentration gradients due to
constant operation at selected ozone generation rates. The test matrix for the chamber tests is
-------
presented in Table 1.
Table 1. Test matrix for large chamber ozone emission rate determinations.
Manufacturer, Model
(Unit ID)
M1,A(1)
Ml, A (2)
M1,B(1)
Ml, B (2)
M2,A
M2, B (1)
M2, B (2)
Total
LowRH (30%)
High 03
High03
MedRH (50%)
High03
Low, Med, High O3 (1)
High03
Low, Med, High 03(2)
Low, Med, High O3<3)
High03
Low, Med, High O3 <4)
HighRH (70%)
High03
High03
# tests
1
3
3
3
3
3
3
19
(l) For the Ml, model A, all tests with 2 large plates installed: Low = ozone generator at 200, Medium = ozone
generator at 900 ft2, High = ozone generator at full scale (FS).
<2) For the Ml, model B, Low = ozone generator at 40, Medium = ozone generator at 320, High = ozone generator at
640 ft2.
(3)For all M2 units, Low = lowest setting on purification dial (1), medium = mid-scale point (4), High = highest set
point (8).
Characterization of Ozone Generating Air Cleaners in the Research Test House
The research test house (Figure 2) is an unoccupied, unfurnished, 1200 ft2 single-story ranch
house that provides a semi -realistic environment for evaluation of chamber emissions data. We
conducted several tests with selected appliances and compared observed concentrations with
those predicted from an indoor air quality model (RISK)18. We also determined the deposition
velocity for ozone inside the house and the penetration factor for outdoor ozone as described
below.
Figure 2. EPA's research test house.
- Local ion: Ozone generator for velociiy deposition exp
The ozone concentration in the test house is a function of ozone emission rate from the source,
ozone deposition velocity, penetration of outdoor ozone into the test house, and loss of ozone
-------
from the test house due to air exchange rate. Equation 6 is the mass balance equation for the
house.
Equation 6. The mass balance equation for ozone in the research house.
JS~1
Vx— — =ESozone+CozoneouxQxp—CxQ-Cxv(f xA
at
where: V = volume of the test house (m3), ERozone = ozone emission rate of the appliance (mg h"
'), Cozoneout = outdoor ozone concentration (mg m"3), Q = volumetric flow leaving the house (m3h"
'), p = penetration factor, C = indoor ozone concentration (mg m"3), vd = deposition rate of ozone
to surfaces in the test house (m h"1), and A = indoor surface area for deposition (m2). Q is
determined from the product of air exchange rate (N) and volume (V), where N is air exchange
rate (h"1) measured by the decay rate of SF6. All other factors, except p and vd, are measured.
Determination of Deposition Velocity and Penetration Factor for the Test House
When the ozone generator is running and indoor ozone concentrations are high relative to the
outdoor concentration, we assume that vd A/V» pCozoneout. The model for the data is:
Equation 7. Model for the test house determination of deposition velocity.
dt V V
Tests were conducted by placing an ozone generating appliance [Ml, A, (0)] at the end of the
kitchen counter facing the hallway with the output of the generator directed towards the air return
grill in the ceiling of the hall (see Figure 2). The ozone generator and fan were set to the highest
set points and the unit was operated for 24 to 72 h dosing periods. Tests were conducted with the
heating and air-conditioning (HAC) system on and furnace fan on continuously and also with the
furnace fan and HAC system off. Floor fans were placed throughout the house to distribute air in
the absence of the furnace fan. For all tests, ozone and NOX were monitored in the den, the master
bedroom and outdoors.
Investigation of Ozone Generators in the Test House
Tests to evaluate our ability to predict ozone concentration from chamber emissions tests were
conducted with the sensor-controller of the Ml units deactivated. Additional tests were
conducted with the sensor-controller activated to determine its effectiveness in maintaining
indoor ozone concentration. All tests were conducted with the HAC system on, with the
thermostat set to 70° F. The HAC fan was on continuously to promote well-mixed conditions.
Two tests each were conducted with the Ml model A, sensor off and on, ozone generator control
set at 1200 ft2; one test was conducted with an Ml model B with sensor on and ozone generator
control set at full scale (>1600 ft2); and one test was conducted with an M2 model B with ozone
generator control set at full scale.
-------
RESULTS AND DISCUSSION
Characterization of Ozone Interactions with Chamber Surfaces
At flow rates of 30 m3 h'1 and RH of 30 and 70%, vd determinations yielded 0.07 and 0.03 m h"1,
respectively. Replicate tests conducted at flow rates of 150 m3 h"1 and RH of 50% yielded vd of
0.1 and 0.01 rn h"1. Of the four static experiments that were conducted in the large chamber, the
first and the last tests were conducted after an intervening flush-out with clean, humidified air
and represent conditions where the chamber surfaces were not deactivated by previous ozone
exposure. Deposition velocities of 0.041 and 0.043 m h"1 were determined for the unconditioned
chamber and 0.022 and 0.024 m h"1 for the conditioned chamber. The average of eight deposition
experiments was 0.043 m h"1 with a standard deviation of 0.029. These deposition velocities are
consistent with values determined by several researchers and summarized by Cano-Ruiz et al. for
several aluminum and stainless steel chambers'9. The impact of the uncertainty of vd on ozone
emission rates determined in the chamber, estimated as ± 2 standard deviations of average
deposition velocity, ranges from 2% at 5 ACH (150 m3 h;1) to 11% at 1 ACH (30 m3 fr1).
Ozone Emission Rates Determined in Chamber Tests
Ozone Emission Rate as a Function of Dial Set Point
In the pilot tests we observed ozone emission rates of <0.5 to 148 mg h"1 at 50% RH over the
range of plate configurations and set points for the PTA. We also determined that the emission
rate held constant (±10%) over 2 to 72 h test periods. The relationship between set point and
ozone emission rate at 50% RH is shown in Figure 3. Note that the full-scale data are plotted at
1600 ft2, which is the highest numbered set point. Average ozone emission rate at full scale is
somewhat higher than that at 1600 ft2, but the difference is not significant. The data demonstrate
the non-linearity of the control dial and show that that generator set points for the Ml model A
for the same square foot of treatment area (ozone generator control dial settings as square feet of
floor area) utilizing different ozone generator plate configurations, may result in very different
ozone emission rates.
Figure 3. Ozone emission rate as a function of generator set point for the appliance [Ml, A, (0)]
used in pilot tests.
1
OSmallPlate H1 Large Plate A2 Urge Plates
Generator Set Point (ft2)
The ozone emission rates determined at 50% RH in the test chamber for two additional Ml
-------
appliances are presented in Figure 4. The model A data are generated with two plates installed.
Figure 4. Ozone emission rates determined in a large chamber at selected dial set points for two
models of Manufacturer 1 appliances at 50% RH, in the manual mode (ozone sensor-controller
not activated).
-0
D
D
_CL
Generator Set Point (ft2)
1,ModelB QMI.ModelA
Ozone Emission Rates for M2 Models A and B
The ozone generator dials of the Manufacturer 2 appliances are scaled from 1 to 8. The ozone
emission rates determined in the chamber at generator set points of 1, 5, and 8 are presented in
Table 2.
Table 2. Ozone emission rates for two models of M2 appliances.
Model and ID
B(l)
B(l)
B(l)
A(l)
A(l)
A(l)
A (2)
Generator Set Point
1
5
8
1
5
8
8
Chamber RH (%)
49,7
50.8
49,7
48.5
48.7
50.3
49.7
Ozone Emission Rate (mg h"1)
9.1
13.7
14.0
17.5
33.7
35.2
26.1
Repeatability of Chamber Tests
For the tests conducted after the pilot studies, three tests were conducted using the same
appliance at the same set point and chamber conditions, and three tests were conducted with two
appliances of the same model at the same set point and chamber conditions. The percent
difference (absolute difference between replicate emission rate determinations/mean) averaged
2.7% for the tests conducted with the same appliance and averaged 16.5% for tests conducted
with different appliances of the same model.
Effect ofRHon Ozone Emission Rate
The effect of RH on ozone emission rate for three models is shown in Figure 5. The linear least
squares fit (Excel) is provided for each model. The data were generated at the maximum
emission rate for each model. The data are consistent across models in that ozone emission rate
-------
decreases with an increase in RH.
Figure 5. Effect of RH on ozone emission rate.
= -1,952Sx +247.33
Chamber RH (•/,)
• Ml. Model A
IM1, ModelB
A M2, Model B
NOX Emission Rate as a Function of Ozone Generation Rate
The relationships between NOX and ozone emission rates are presented in Figure 6 for the Ml
and M2 appliances, A linear least squares fit is shown for the pooled Ml and pooled M2 data
sets. The relationship between NOX generation rate and ozone generation rate is:
ER(NOx)=0.16xER(ozone) + 0.36 for the M2 and ER(NOx)=0.06xER(ozorie) + 0.44 for the Ml
appliances, with R2 for the linear fit of 0.846 and 0,956, respectively. Thus, the M2 appliances
appear to generate more NOX per amount of ozone generated than the Ml appliances.
Figure 6. NOX emission rate as a function of ozone emission rate.
§ 8.00
60 100
Ozone Emission Rate (mg/h)
* M1, Model A
X M2 Model B
• M1, ModelB
——Regression All M1
M2 Model A
Regression All M2
Stability of Ozone Generation Rates
The stability of ozone generation rates was determined as the standard deviation of ozone
emission rate determined after the test chamber reached steady state conditions. The stability
10
-------
was typically in the range of 3 to 8%, and variations in emission rate tracked the fluctuations in
test chamber RH. Therefore, we suspect that the apparent variability of ozone emission rate is
due primarily to the fluctuations of RH in the test chamber. Some variation of apparent emission
rate was observed in the chamber test of the M2, model B that did not appear to be related to RH.
A similar fluctuation was observed in the test house (see Figure 9). We have no explanation for
this behavior.
Effect of Appliance Fan Speed on Ozone Emission Rate
The ozone emission rates observed for the tests conducted at low and medium fan speeds were
bounded by the emission rates observed for replicate tests at the highest fan set point. We
conclude that ozone emission rate is not particularly sensitive to fan speed, and we recognize that
our test procedure does not have the precision to elucidate a relationship, if one exists.
Characterization of Ozone Deposition and Penetration in Research House
The ozone deposition rate for the test house was determined by operating the ozone generator for
several hours when there was no outdoor ozone. Because the ozone generation rate was known
from the pilot experiments and the air exchange rate was measured, the ozone deposition rate
could be determined from the steady state solution of Equation 7. The ozone deposition rate was
found to be 1.5 h"1 (4 x 10"4 sec"1). This value is somewhat lower than the values that range from
1 x 10~3 to 7 x 10"4 sec"1 reported by several researchers, as summarized by Reiss et al., for various
indoor environments20. The penetration factor for ozone was determined from the steady state
solution to Equation 6 and data when the outdoor ozone was relatively high and constant for
several hours and there was no ozone generator indoors. The best estimate of the penetration
factor is 0.15. This value is somewhat uncertain because the outdoor and indoor ozone
concentrations were near the quantification limits of the instruments.
IAQ MODELING OF RESEARCH HOUSE
The IAQ model RISK18 was used to model the research house experiments. The ozone
deposition rate experiments and experiments with the ozone generator located in the den and
operated continuously were modeled. The ozone emission rates for the modeling were based on
the pilot results and included the effects of dial setting and RH. Because the RH in the research
house was low during these experiments, the RH effect on ozone emission rates was important.
The research house experiments that were modeled, the house operating conditions for the
experiment, and the final predicted and measured ozone concentrations are shown in Table 3.
The measured and predicted final ozone concentrations are generally in good agreement. Note
that the sensor-on experiments were not modeled.
For most of the experiments, the predicted and measured concentration time histories are also in
good agreement. However, it should be noted that in some of the experiments, the predicted and
measured concentration/time histories do not agree as well as do the final concentrations. An
example of good agreement is shown in Figure 7 for test THVD01C. An example of poor
agreement is shown in Figure 8.
11.
-------
Table 3. List of research house experiments that were modeled.
Test#
THVD01C
THVD03
THVD04
THOZ1
THOZ2
THOZ3
THOZ4
THOZ5
THOZ6
AC
Location
Kitchen
Kitchen.
Kitchen
Den
Den
Den
Den
Den.
Den.
HAG
ON
OFF
OFF
ON
ON
ON
ON
ON
ON
Air cleaner
M1,A(0)
M1,A(0)
M1,A(0)
M1,A(1)
M1,A(1)
M1,A(1)
M1, B(2)
M1,A(1)
M2, B(1)
Setting
Full
Scale
Full
Scale
Full
Scale
1200f
1200?
1200?
Full
Scale
1200 I3
Full
Scale
Sensor
OFF
OFF
OFF
OFF
ON
ON
ON
OFF
NA
ACH
0.48
0.39
0.42
0.35
0.40
0.42
0.44
0.37
0.39
RH%
15
24
23
26
29
18
Ozone
Emission
rate mg/h
215
181
197
58
58
40
Predicted /
Measured
Den
172/172
305/305
318/310
113/140
Not
modeled
Not
modeled
Not
modeled
112/158
77/65
Predicted/
Measured
MBR
175/170
180/185
207/225
52/38
Not modeled
Not modeled
Not modeled
51/48
35/18
The results of the two tests THOZ1 and THOZ5 along with model predictions are shown in
Figure 8. The data show a long buildup to steady state that the model predictions do not show.
The buildup is not due to air exchange changes because the model predictions use the 1 hour
average air exchange rates measured. The model also predicts a higher concentration in the
master bedroom than the measured concentration. This may be due to mixing that is not
accounted for in the model. The model predictions assume that the only mixing is that provided
by the HAC.
The final research house experiment, THOZ6, was conducted with the M2, B(l) ozone generator.
The results of the experiment and IAQ model predictions are shown in Figure 9. The model
predictions are in fair agreement with the measurements, except that the data show a large dip in
concentration between 12 and 13 hours. This dip is not due to changes in air exchange, which
are accounted for in the model. A similar dip was observed in the large chamber emission rate
tests with this unit. Neither the large chamber nor the research house data provide an explanation
for the dip.
EFFECT OF SENSOR CONTROLS
Three 24 h tests were conducted in the research house with the Ml model A and model B
appliances where the ozone sensor-controller was placed in the automatic mode. The results
were similar for all three tests in that the ozone concentrations remained below 50 ppb in the
room where the appliance was placed. Results for one model A and the model B are shown in
Figures 10 and 11.
12
-------
Figure 7. Measured and predicted concentration/time history for experiment THVD01C.
15 20
Elapsed Time (h)
Figure 8. Measured and predicted concentration/time history for experiments THOZ1 and
THOZ5.
lei Den
idel MBR
Data THOZ1 Den
Data THO25 Den
Data THOZ1 MBR
Data THOZ5 MBR
Elapsed Time (h)
Quality Assurance (QA) and Quality Control (QC)
The tests were conducted under approved QA project plans that detailed test conduct, data
collection, data review, and reporting. QC measures included calibration with traceable
standards for ozone (EPA SRP01) and NOX (certified NO, NO2, Air Products Corp.)
measurement systems, and daily zero and span checks for ozone, NOX, and SF6 measurement
systems. Technical systems and performance audits were conducted by Research Triangle
Institute under EPA contract 68-C-98-173. Results of the technical performance audits
demonstrated that all measurements systems operated within QA goals. Maximum deviations
(%) at 80% of full scale were: 0,6,7.8 for ozone (Teco Model 49, API Model 400), 2.6, -2.3,4.0
for NO, NO2J NOX (Teco Model 42), and 1.3 for SF6 (HP5890 GC).
13
-------
Figure 9. Measured and predicted concentration/time history for experiment THOZ6.
Elapsed Tim e (h )
CONCLUSIONS
The IAQ mode! RISK provides reasonable predictions of ozone concentration in a research
house. The model predictions and the research house data show that the appliances tested have
the potential to generate steady state ozone concentrations that exceed existing standards. The
ozone generators equipped with sensors and control circuits designed to limit ozone
concentrations to 50 ppb did in fact limit ozone concentrations in the research house to less than
50 ppb—usually much less than 50 ppb.
Figure 10. m-roorn concentrations for a Ml, model A with sensor-controller activated
Elapsed Time (h)
14
-------
Figure 11. In-room and outdoor ozone concentrations for a Ml, model B operated with sensor-
controller activated.
^^^:^v^
&s^^&i&te«£&k+ +++ +*++ * «
** % $ *•* * *
Sttj^***4$44 * *
>!«**%**:*«> •%«»• _t <
Elapsed Time (h)
Ozone in Den D Ozone Outdoors
References
1. Weschler, C. J.; Shields, H. C; Naik, D. V. JAPCA. 1989, 39, 1562-1568.
2. Shair, F. H.; Heitner, K. L. Environ. Sci. Techno!. 1974, 8, 444-451. '
3. Sabersky, R. H.; Sinema, D. A.; Shair, F. H. Environ. Sci. Technol. 1973, 7, 347-353.
4. Leovic, K.; Sheldon, L.; Whitaker, D.; et al. Journal of the Air & Waste Manangement
Association, 1996, 46, 821-829.
5. Steiber, R, S. Ozone Generators in Indoor Air Sellings. EPA-600/R-95-154 (NTIS PB96-
100201), National Risk Management Research Laboratory, U. S. Environmental
Protection Agency, Research Triangle Park, NC 27711, October, 1995.
6. Phillips, T. J.; Bloudoff, D. P.; Jenkins, P. L.; Stroud, K. R. Journal of Exposure Analysis
and Environmental Epidemiology, 1999, 9, 594-601.
7. USFDA (U. S. Food and Drug Administration). Federal Register, 1974, 39^ 13773.
8, USOSHA (U.S. Occupational Safety and Health Administration). Code of Federal
Regulations, Title 29, Part 1910.1000, January 19, 1989.
9. USEPA (U. S. Environmental Protection Agency). 40 CFR Part 50, National Ambient
Air Quality Standards for Ozone; Final Rule,. Federal Register 62 (138), 1-37.
10. Foarde, K. K.; VanOsdell, D. W.; Steiber, R. S. Appl. Occup. Environ. Hyg. 1997, 12,
535-542.
11. Atkinson, R.; Carter, W. P. L. Chem. Rev. 1984, 84, 437-470.
12. Ahang, J.; Wilson, W. E.; Lioy, P. J. Environ. Sci: Technol, 1994, 28, 1975-1982.
13. Weschler, C. J.; Hodgson, A. T.; Wooley, J. D. Environ. Sci. Technol 1992, 26, 2371-
2377.
14. Wolkoff, P.; Clausen, P. A.; Wilkins, C. K.; Hougaard, K. S.; Nielsen, G. D. Atmospheric
Environment. 1999, 33, 693-698.
15
-------
15. Living Air Electronic Air Purification Systems, Owner's Manual, Living Air XL-15 and
XL-15S. Living Air, 310 T. Elmer Cox Drive, Greeneville, TN 37745.
16. ASHRAE, ASHRAE Handbook of Fundamentals, American Society of Heating,
Refrigerating and Air-Conditioning Engineers, Inc., Atlanta, GAj 1985; Chapter 22.8.
17. ASTM, "Standard Test Method for Continuous Measurement of Ozone in Ambient,
Workplace, and Indoor Atmospheres (UV Absorption)," (ASTM D 5156-95), Annual
Book of ASTM Standards, American Society for Testing and Materials, Philadelphia,
PA, Vol. 11.03, 1997, pp. 461-467.
18. Sparks, L. E. (1996); IAQ Model for Windows RISK Version 1.0: User Manual, EPA-
600/R-96-037 (NTIS PB96-501929), Air Pollution Prevention and Control Division,
Research Triangle Park, NC 27711.
19. Cano-Ruiz, J. A.; Kong, D.; Balas, R. B.; Nazaroff, W. W. Atmospheric Environment,
1993, Vol. 27A, No. 13, 2039-2050.
20. Reiss, R.; Ryan, B. P.; Koutrakis, P. Environ. Sci. Technol. 1994, 28, 504-513.
16
-------
4. TITLE AND SUBTITLE characterization of Ozone Emissions
from Air Cleaners Equipped with Ozone Generators
and Sensor and Feedback Control Circuitry
N RMHL- RTF- P- 532
TECHNICAL REPORT DATA
(Please read Itatouctiam on the reverse before completing/
1, REPORT NO,
EPA/600/A-00/059
2.
3, «EC
5. REPORT DATE
6. PERFORMING ORGANIZATION CODE
?. AUTHQRtSi
M. A. Mason, L. E. Sparks, S.Moore, I.Dolgov, and
R. Perry
8. PERFORMING ORGANIZATION REPORT NO.
9, PERFORMING ORSANIZATION NAME AND ADDRESS
10. PROGRAM ELEMENT NO.
See Block 12
11. CONTRACT/GRANT NO.
NA (Inhouse)
12, SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Air Pollution Prevention and Control Division
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Published paper; 3/98-3/00
14, SPONSORING AGENCY CODE
EPA/600/13
15, SUPPLEMENTARY NOTES APPCD project officer is Mark A. Mason," MaiTbrop~54, 919/541-
4835, For presentation at Engineering Solutions to IAQ Problems, Raleigh, NC,
7/17-19/00.
is. ABSTRACT
paper gives results of a characterization of ozone emissions from air
cleaners equipped with ozone generators and sensor and feedback control circuitry.
Ozone emission rates of several consumer appliances, marketed as indoor air treat-
ment or air purification systems, were determined in a well- characterized '30- cubic
meter environmental chamber. Additional tests were conducted in a research test
house to determine the effects of appliances on research house ozone concentrations
and to evaluate the performance of an ozone generator feedback control system.
Ozone emission rates determined in the chamber studies ranged from 0, 5 to >150
mg/hr for one appliance at 50% relative humidity (RH), Ozone emission rates inc-
reased as RH decreased for all models tested. Nitrogen dioxide generation rates
ranged from 0.06 to 0.16 times the ozone emission rate and varied among models of
different manufacturers. When chamber- derived ozone emission rates were input
into the RISK model, the model provided reasonable estimates of the time history
of ozone concentrations that were measured in the research house. The research
house experiments demonstrated that these types of devices can produce ozone con-
centrations above accepted health guidelines.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
tUOENTIPIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution
Ozone
Emission
Air Cleaners
Nitrogen Dioxide
Mathematical Models
Properties
Analyzing
Pollution Control
Stationary Sources
Characterization
13B
07B
14G
13 A, 131
12A
14B
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 222O-1 (9-73)
------- |