United States
Department of Housing
and Urban Development
Office of Policy Development and Research
Office of Community Planning and Development
Washington DC 20410
United States
Environmental Protection
Agency
Environmental Monitoring and Support
Laboratory
Research Triangle Park NC 27711
EPA-600/7-78-229a
December 1978
Indoor Air Pollution in
the Residential Environment
Volume I
Data Collection, Analysis
and Interpretation
Interagency
Energy/Environment
R&D Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide'range of energy-related environ-
mental issues.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/7-78-229a
December 1978
Indoor Air Pollution
In the Residential Environment
Volume I. Data Collection, Analysis,
and Interpretation
by
Demetrios J. Moschandreas, John W.C. Stark,
James E. McFadden, and Sallie S. Morse
GEOMET, Incorporated
Gaithersburg, Maryland 20760
EPA Contract No. 68-02-2294
EPA Project Officer: Steven M. Bromberg
Quality Assurance Branch
Environmental Monitoring and Support Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
and
U.S. Department of Housing
and Urban Development
Office of Policy Development and Research
Washington, DC 20410
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Environmental Monitoring and Support Laboratory
Research Triangle Park, North Carolina 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Monitoring and
Support Laboratory, U.S. Environmental Protection Agency, the U.S.
Department of Housing and Urban Development, and non-governmental personnel,
and approved for publication. Approval does not signify the contents
necessarily reflect the views and policies of the U.S. Environmental
Protection Agency, or the U.S. Department of Housing and Urban Development,
nor does the mention of trade names or commercial products constitute
endorsement or recommendation for use. The views, conclusions and
recommendations in this report are those of the contractor, who is solely
responsible for the accuracy and completeness of all information and data
presented herein.
-------
EXECUTIVE SUMMARY
INTRODUCTION
Between March 1976 and March 1978 GEOMET, Incorporated conducted a
study of residential indoor air pollution for EPA and HUD. The work has
been described in a series of GEOMET reports, the principal ones being:
The Status of Indoor Air Pollution Research—1976
EPA Report Number 600/4-77-029, May 1977
Survey of Indoor Air Quality Health Criteria and Standards
EPA Report Number 600/7-78-027, March 1978
Proceedings of the GEOMET Program Review Workshop--Air
Pollution, Energy Conservation and Health Effects in tFe
Indoor Residential Environment
GEOMET Report Number EF-646, September 1977
The GEOMET Indoor-Outdoor Air Pollution Model
GEOMET Report Number EF-628, February 1978
Indoor Air Pollution in the Residential Environment
Volume I - Data Collection, Analysis and Interpretation
Volume II - Field Monitoring Protocol, Indoor Episodic
Pollutant Release Experiments and Numerical
Analyses
GEOMET Report Number EF-688, August 1978.
This Executive Summary outlines the highlights of these reports.
OBJECTIVES OF THE STUDY
The indoor residential environment constitutes a major component of an
individual's total exposure to air pollution. Residential air pollution has
not been studied to a great extent in the past and has not been considered
in the setting of regulatory standards for air pollution.
The recent national effort to conserve energy has led to a justifiable
concern about the possible impact that energy conservation measures may have
on indoor residential air quality. The following questions arise concerning
the public health and welfare:
• What are the air pollution characteristics within residences?
• How do they relate to indoor-outdoor air exchange rates?
• How are they affected by energy conservation measures?
• What are the potential health effects of residential air
pollution?
-iii-
-------
Questions of this kind prompted the decision by EPA and HUD to sponsor
a comprehensive study of the air quality in residences. The experimental
design for monitoring in residences, the mathematical and statistical tech-
niques used to analyze the data, the significance of the findings, and the
areas that warrant future research are briefly discussed in the following
pf;ges.
TECHNICAL APPROACH
Phase I of the project was a literature search of the indoor air pol-
lution studies as described in published literature and unpublished ongoing
r
-------
Analysis of the large data base generated by the field program was a
multiphc'.sed procedure which ranged from interpreting isolated short-term
behavior of indoor pollutant concentrations to establishing trends of the
fluctuations of air pollutant concentrations in the residential environment.
A data set from a 14-day monitoring period was treated as an independent
entity; each pollutant was investigated separately. However, interelement
comparisons were carried out, seasonal variations were examined, and a classi-
fication of residences from the perspective of air quality was undertaken.
Analytical techniques used for data interpretation included simple observa-
tions, descriptive statistics and complex analytical expressions denoting
the dynamic relationships involved in the explanation of the observed
behavior of indoor pollutant concentrations.
In addition to data collection and analysis, two numerical models were
formulated to predict the indoor pollutant concentrations. The GEOMET
Indoor-Outdoor Air Pollution (GIOAP> model predicts indoor concentrations
of gaseous pollutants. The fundamental principle involved in the formu-
lation of the GIOAP model is the principle of mass balance. This model is
a first order differential equation which dynamically relates the rate of
change of an indoor pollutant concentration to the rate of introducing a pol-
lutant indoors through ventilation, infiltration, recirculation, and indoor
sources, and to the rate of removing the pollutant concentration from the
indoor environment through the mechanisms of exfiltration, exhaust, indoor
chemical sinks, and air cleaning devices. Validation studies performed
with the GIOAP model indicate that it satisfactorily predicts corresponding
observer1 values.
The TSP (total suspended particulate) empirical model was the second
model formulated for this study. It is a steady-state model which was based
on the available data. A portion of the TSP matter found indoors is of
outdoor origin, while the remaining portion is attributed to indoor activi-
ties. Studies on particulate matter have concentrated on quantifying source
strengths of individual indoor TSP generating mechanisms, such as vacuum
cleaning, operating a fan, smoking, frying, house cleaning, using sprays,
moving in and out of the house, ventilating devices, and others. This study
did not attempt to measure the strength of individual TSP sources; rather,
it provided a scale of the activity of each residence and quantified indoor
TSP levels as a function of the family activity index. This approach took
advantage of the data available to the project and utilized the question-
naire which was answered on a daily basis. The procedure used for the model
formulation utilized a portion of the available data to define the indoor TSP
source strength as a function of the activity scale, and the remaining data
to verify the numerical predictions. The TSP model realistically predicts
the observed indoor TSP concentrations.
Analysis of energy data collected in monitored residences demonstrated
that the energy consumed is a function, among other factors, of the air
exchange rate. It has also been demonstrated that changes in the air
-v-
-------
exchange rate affect the residential air quality. The relationship between
eneroy conservation measures and air quality in residences was studied by
two methods: 1) analysis of the observed energy and air quality data;
and 2) numerical simulations using the state-of-the-art models for esti-
mating residential energy consumed and the GIOAP model for calculating
the indoor pollutant concentrations.
RESULTS
The air quality in the indoor residential environment has been found
to be markedly different from the ambient outdoor air quality. Three
classes of air pollutants have been identified with respect to indoor-
outdoor pollutant relationships:
1. Concentrations of carbon monoxide (CO), nitric oxide (NO),
carbon dioxide (COp), hydrocarbons, and aldehydes in the
residential environment were often higher than outdoors.
2. Indoor concentrations of nitrogen dioxide (NOp), TSP, and
respirable suspended particulate matter (RSP) are some-
times higher and sometimes lower than outdoors.
3. Indoor concentrations of sulfur dioxide (S0?), ozone (0.J,
sulfates (SO,), nitrates (NO.,), and lead (Po) are often lower
than corresponding outdoor pollutant concentrations.
The observed indoor air pollutant concentrations were, on the average,
not very high, but persistent moderate concentrations and at times elevated
pollutart levels were observed in the monitored residences. Air pollution
standards for the residential indoor environment have not been promulgated.
However, field measurements from this program have established that concen-
trations of 03, nonmethane hydrocarbons, and TSP matter exceed the National
Ambient Air Quality Standards (NAAQS) in the indoor environment. The 8-how
NAAQS for CO was not exceeded by the registered indoor concentrations, but it
was equaled several times. The American Society of Heating, Refrigerating
and Air Conditioning Engineers (ASHRAE) has recommended standards for, among
others, C02 and aldehydes. These standards have been exceeded by the
observed concentrations of these contaminants and, on certain occasions, by
a factor of two or three. The observed high levels of CO2, nonmethane
hydrocarbons, aldehydes and TSP matter may be attributed to indoor sources,
while the elevated levels for the other pollutants may be caused by high
ambient concentrations.
Two major classes of indoor residential environments have been identi-
fied by this study: 1) residences with indoor pollutant sources; and 2)
residences without indoor pollutant sources. This classification is pollu-
tant specific; it is quite possible that the same residence may belong in
-------
one class for one pollutant and in another class for another. A few of
tie many possible indoor pollutant sources are gas appliances, smoking,
cleaning and cooking activities, aerosol sprays, wall paint, and chipboard.
Pollutants emitted by these sources include CO, NO, NOp, organics, TSP
and RSP matter, aldehydes, ammonia, and other contaminants. An attempt to
put indoor sources in order of importance is not appropriate because their
impact on the indoor environment depends on their frequency of use and mode
of operation. However, for three pollutants (CO, NO, and NO,,), residences
with qas appliances have stronger indoor sources than residences with elec-
tric appliances. The field program of this study has shown that the indoor
eir quality deteriorates when an indoor pollutant source is operating. It
has also been observed that the adverse impact of an indoor air pollution
source is accentuated in residences with low air exchange rates because the
contaminated air cannot exfiltrate.
The air exchange rate is the parameter that associates energy conserva-
tion measures with air quality in residences. The data base collected for
this project, combined with the results of numerical simulations carried
out with the GIOAP model, demonstrates that from the perspective of air
quality, reduction of the air exchange rate, an energy conserving measure,
may lead to deterioration of the residential air quality. It has been deter-
mined that retrofitting existing residences down to an air exchange rate
between 0.4 to 0.6 air changes per hour conserves energy without inducing
drastic deterioration of the indoor air quality.
This study has clarified several important aspects of exposure to air
pollutants in residential spaces. These results have shown that indoor air
duality, as a composite, may present quite different exposure conditions
than the surrounding ambient air quality. The study has confirmed three
particular characteristics of indoor air quality that could have significant
implications for the health and comfort of building occupants:
1. Emissions from gas-fueled appliances and a variety of other
household sources can add considerably to indoor pollutant
levels.
2. Buildings often protect individuals from high outdoor con-
centrations of the more reactive ambient gaseous pollutants,
e.g., 03, and probably from particulate matter with large
mean mass diameter.
3. Measures taken to reduce the building air exchange rate also
tend to increase the persistence over time of indoor pollutant
levels.
These factors work in combination to govern indoor pollutant concentrations
and should be viewed from both a short- and long-term perspective.
-vii-
-------
Investigation of the air quality control measures in the residential
environment has shown that the few commercially available control devices
are either inefficient, noisy, energy consuming, or expensive. Among the
most frequently used air quality control instruments in the residential
environment are the fan in the range hood and the panel filter in the
HVAC system. The efficiency of these instruments needs to be studied
and improved.
SIGNIFICANCE
This study has established that the character of the residential
environment is unique and that conclusions reached by examination of the
ambient environment do not necessarily apply to the indoor environment.
The present study on the residential air quality has significantly
enhanced the scientific knowledge of the subject. The state-of-the-art,
which was reviewed by a literature search in Phase I of this study, has
been advanced by the enlargement of the available data base, the formula-
tion and application of numerical models, the examination of the feasibil-
ity of an epidemiological program relating indoor air quality and health
effects, and the investigation of the relationship between residential
energy conservation measures and air quality.
The data base collected by the field program has greatly increased the
available information, especially if the real-life aspects of the monitoring
design are taken into consideration. The emphasis placed on the validation
and application of numerical procedures for predicting residential pollutant
concentrations is unique with this study and it points toward new avenues of
relevant research. Technical questions resolved and assumptions verified
appear throughout the text of the final report. However, one widespread
assumption must be mentioned: it is incorrect to assume that the indoor
environment shelters its inhabitants from all high ambient pollutant concen-
trations. This assumption is true only for certain pollutants. It is not
always true for CO. Sheltering factors are currently under consideration but
it would be misleading to establish a sheltering factor for CO concentra-
tions, because this pollutant may be generated indoors.
The recent emphasis in energy conservation has increased the public
awareness of the dynamics involved in the energy-environment complex. How-
ever, this awareness has been confined to the ambient outdoor environment.
The present work has indicated that the indoor environment may be another
area where the two apparently conflicting national goals manifest themselves.
Both objectives can be realized if appropriate steps are taken. This study
suggests a number of specific scenarios towards this goal. A most important
step is that research on conservation of energy consumption in residences
and on indoor air quality should be an integrated process. Protecting the
welfare of individuals should not be examined in isolation either from the
-viii-
-------
energy or the air quality perspectives, rather the issue should be dealt
with comprehensively. Coordinated effort will avoid unnecessary duplica-
tion and will speed up the resolution of several existing problems.
FUTURE NEEDS
In addition to problems resolved and questions answered, the indoor air
quality assessment study has uncovered a series of new research requirements.
Further research is needed on the indoor source emission rates, the chemical
nature of certain air contaminants, such as aldehydes, RSP, organics, and
others. New information on infrequently studied pollutants such as benzo-a-
pyrene, asbestos, nitrosamines and others will provide additional insights
on the health effects of the indoor air quality. Further research is war-
ranted on residential air pollution control techniques and energy management
procedures which conserve energy and do not affect the indoor air quality.
Possible emissions of oxides of nitrogen from residential gas furnaces
have been the subject of recent controversy. Results of past studies have
been inconclusive. The data obtained in the course of this study do not
resolve this controversy since the experimental design did not call for mon-
itoring in the furnace areas. Additional studies must be undertaken to
examine this potentially serious source of indoor MO generation.
The next step in evaluating the importance of residential air quality
is the assessment of health effects of indoor air pollution with an epi-
demiological study of community populations in a variety of Indoor air
environments. The research tools are available and the need for a compre-
hensive study is apparent. The eventual outcome of such an undertaking
will lead to improved protection of the public health and may help con-
serve additional energy in residences.
"Retrofitting" is a term used by specialists to denote efforts toward
reducing energy consumption in existing residences. Reduction of the
residential air exchange rate 1s one form of retrofitting that affects the
Indoor air quality. Research must focus on forms of retrofitting that do
not impact on the indoor air quality. Building codes for new residential
structures are under consideration. The driving concern is energy conser-
vation; however, protection of the occupants' health in the new residences
should be an additional input. Thus, while air quality specifications are
not integral parts of the proposed building codes, they must be considered
in the determination of the code itself.
-ix-
-------
CONTENTS
Executive Summary iii
Figures xii
Tables xv
Acknowledgments xviii
1. Introduction 1
2. The Air Quality of the Indoor Residential
Environment: Experimental Design, Analysis
Procedures, and Data Interpretation 3
Field Program Design: Data Collection 3
Characterization of the Observed Indoor
Air Quality 8
3. Numerical Models 75
The GEOMET Indoor-Outdoor Air Pollution
(GIOAP) Model 75
Conclusions 126
The Steady-State TSP Model 129
Technical Approach 130
Discussions and Conclusions 135
4. Energy Considerations 139
Energy Data Collection 139
Calculation of Energy Use 147
5. Residential Air Quality and Energy Conservation
Measures 156
Energy and Cost Savings through Air Exchange
Rate Reduction. 156
Air Quality Impacts of Energy Conservation 159
Air Quality Control Measures 164
Energy Management in Residences 175
References 181
-xi
-------
FIGURES
Number Page
1 Sampling system for continuous monitors 5
2 Equilibrium situation 11
3 Percentage of hourly average Og concentrations 17
4 Indoor and outdoor carbon monoxide (CO) variation
at the Baltimore conventional residence 22
5 Typical RSP indoor-outdoor levels 29
6 Typical pattern of observed nitrate levels 32
7 Pittsburgh low-rise #1 S0| 35
8 Observed indoor and outdoor aldehyde time (day)
variations 40
9 Time variation of Pb and Br concentrations and Br/Pb
ratio in Denver single-family dwelling 50
10 Time variation of potassium (K) concentrations in
Baltimore experimental residence 52
11 Time variation of iron (Fe) concentrations in
Baltimore experimental residence 53
12 Time variation of sulfur (S) concentrations in
Baltimore conventional residence 54
13 Episodic release of SF^ gas 56
14 Daily activity record 58
15 Revised version of daily activity record 59
16 Typical calibration curve for SFg 63
17 Instantaneous release 66
18 Continuous release 68
19 Effect of temperature on infiltration 69
(Continued)
-xii-
-------
FIGURES (Continued)
Number Page
20 Effect of HVAC operation on residential air
exchange rates 71
21 Effect from occupant activity 73
22 Graphical illustration for the three cases of
constraints on internal source rate 88
23 Estimated vs. observed pollutant concentrations
for 7 consecutive days 92
24 Estimated vs. observed pollutant concentrations
for 7 consecutive days 93
25 Scatter diagram with r = 0.96 98
26 Scatter diagram with r = 0.82 99
27 Scatter diagram with r = 0.62 100
28 Scatter diagram with r = 0.62 101
29 Nominal values 123
30 Comparison of nominal values obtained
by perturbing Cin 124
31 Comparison of nominal values with values obtained
by perturbing S 125
32 Comparison of nominal values with values obtained
by perturbing 125
33 Estimated values of indoor TSP using the steady-
state model for the Pittsburgh high-rise
apartment #3 137
34 Energy use profile for Baltimore conventional
residence on January 31, 1977 150
35 Energy use profile for Washington conventional
residence on July 10, 1977 151
36 Energy use profile for Pittsburgh low-rise
apartment on April 7, 1977. . 152
(Continued)
- x i 1 i -
-------
FIGURES (Continued)
Number
37 Energy use profile for Pittsburgh Mobile I
on February 17, 1977
38 Effect of reducing the air exchange rate in
residence with indoor CO sources
39 Effects of reducing the air exchange rate in
residence without indoor CO sources. . . .
40 Characteristics of particulate pollutants
41 Single-family detached—Baltimore, heating
42 Single-family detached—Baltimore, cooling
-xi v-
-------
TABLES (Continued)
Number Page
19 Ratio Values of Outdoor Pb Concentrations with
Corresponding Indoor Averages 47
20 Asbestos Count Results 48
21 Activity Point Scale 60
22 Characterization of Family Type 60
23 Physical and Meteorological Parameters 64
24 Experimentally Determined Air Exchange Rates 67
25 Seasonal Variation in Air Exchange Rates 70
26 Effect of HVAC Operation 70
27 Decay Factors (per hour) Used in the GEOMET Indoor
Air Pollution Study 80
28 Air Exchange Rates. 90
29 Statistical Data Summary 96
30 Statistical Data Summary for Carbon Monoxide (CO ppm) . . 103
31 Statistical Data Summary for Nitric Oxide (NO ppm) . . .105
32 Statistical Data Summary for Nitrogen Dioxide (N02 ppb) . 107
33 Statistical Data Summary for Sulfur Dioxide (S0£ ppb) . . 108
34 SOg Frequency Distribution. 109
35 Negative COg Interference on SOg Levels 110
36 Statistical Data Summary for Nonmethane Hydro-
carbons (THC-CH^ ppm) 112
37 Statistical Data Summary for Methane (CH^ ppm) 113
38 Statistical Data Summary for Carbon Dioxide (COg ppm) . .115
39 Nominal Conditions Used in the Sensitivity Study
Examples 123
(Continued)
-xvi -
-------
TABLES (Continued)
Number Page
40 Indoor TSP Source Strength 135
41 Comparison of Indoor Observed TSP Level, Against TSP
Levels Estimated with Steady State Model . . 136
42 Estimated Values for Cw, Cp, and Cwr 144
43 Equivalent Orifice Areas 145
44 Orifice Coefficients 145
45 Energy Savings through Reduced Infiltration for a
Single-Family Detached Residence 158
46 Estimated 24-Hour Indoor Carbon Monoxiode Exposure .... 162
47 Summary of Pollutant Emission of Gas Appliances for
Several Typical Operating Conditions in Hartford
Dwellings 166
48 Performance of Panel Filters 169
49 Performance of Electronic Air Cleaners 171
50 Specifications for a Typical and Well-Insulated
Residence 178
-xvii-
-------
ACKNOWLEDGMENTS
The contents of this report are the results of a research program
performed by GEOMET, Incorporated under the joint sponsorship of the U.S.
Environmental Protection Agency and the Office of Policy Development and
Research of the U.S. Department of Housing and Urban Development. The
following EPA research laboratories sponsored this program: Environmental
Monitoring and Support (EMSL), Health Effects Research (HERL) and Indus-
trial Environmental Research (IERL). A multidisciplinary project, such
as the present one, requires the talents and the efforts of many people
to whom the authors wish to express their appreciation. In particular,
we wish to thank Mr. Steven Bromberg, the EPA Project Officer, who has pro-
vided guidance, support, and assistance throughout this project. We would
also like to recognize the essential and valuable contributions made by
GEOMET's subcontractors: PEDCo Environmental Specialists, Mr. Lawrence A.
Elfers, Project Manager; Hittman Associates Incorporated, Mr. Daniel W.
Talbott, Project Manager; IIT Research Institute, Dr. Andrew Dravnieks,
Program Director. Special credit should be given to three of our consul-
tants for their valuable scientific assistance: Dr. Frederick H. Shair
of California Institute of Technology; Dr. John W. Winchester of Florida
State University; and Dr. Ian T.T. Higgins of the University of Michigan.
Many GEOMET scientists have contributed significantly to this project
and two must be mentioned by name: Mr. John L. Swift, Project Manager of
Phase I of this program, has contributed throughout the project by providing
scientific and managerial assistance; Mr. John R. Ward has served as staff
consulting advisor concerning statistical techniques and applications
required by this project. In addition, we express our appreciation to
Ms. Leonora L. Riley, Manager of Technical Support, and her staff, espe-
cially Mrs. Efegenia G. Maxwell, for the excellent effort involved in
preparing and typing the final report.
Finally, the authors would like to acknowledge the occupants of the
residences who participated in this program for their help and indulgence
while their homes were being monitored.
Demetrios J. Moschandreas
Gaithersburg, Maryland
August 31, 1978
-xviii-
-------
SECTION 1
INTRODUCTION
The recent emphasis on reducing energy consumption for heating and
cooling residences has precipitated scientific interest in residential air
quality and its relationship to energy conservation measures. It has been
recently recognized that indoor air quality may be as important as ambient
air quality as a health factor; that substantial amounts of energy may be
conserved in the residential environment by processes that may lead to
deterioration of the indoor air quality; and that the relationships
between outdoor pollution levels and indoor pollutant concentrations
are not well established and warrant further research.
In response to those broad requirements, the U.S. Environmental
Protect on Agency (EPA) and the U.S. Department of Housing and Urban
Developnent (HUD) have jointly sponsored a study of the residential air
environment. The project entitled "Indoor Air Pollution Assessment,
Control and Health Effects" was undertaken by a multidisciplinary team
directed by GEOMET, Incorporated. GEOMET was responsible for management
as well as the health effects, air quality analysis, and modeling aspects
of the technical program. GEOMET was assisted by the efforts of three sub-
contractors: Hittman Associates, Inc. for energy considerations; PEDCo
Environmental Specialists, Inc. for field monitoring; and IIT Research
Institute (IITRI) for analysis of organic contaminants released episodi-
cally in the indoor environment. Consultants from several major univer-
sities, in the fields of epidemiology, monitoring, and modeling of indoor
air pollution, provided advice to the project through GEOMET.
The 24-mo study involved a literature search to define the state-
of-the-art prior to this undertaking, and an 18-mo period of monitoring,
analysis, and interpretation. The results of the literature search have
been described by GEOMET and its subcontractors in an EPA publication
entitled "The Status of Indoor Air Pollution Research, 1976." The final
report of the project consists of two volumes and an executive summary.
The executive summary presents an overview of the project. Volume I of
the final report, the present document, is entitled "Indoor Air Pollution
in the Residential Environment: Data Collection, Analysis and Interpreta-
tion." Volume II entitled "Field Monitoring Protocol, Indoor Episodic
Pollutant Release Experiment and Numerical Analyses," presents all of the
supportive documents including further information on the experimental design
and the monitoring instruments housed in the mobile laboratory, the numeri-
cal methods for the GEOMET Indoor-Outdoor Air Pollution (GIOAP) model, and
the IITRI reports on the episodic release of pollutants.
The indoor environment was studied comprehensively; information on
the air quality, meteorology, energy, air exchange rates, and family
activities was collected. Over 250 pages of computer output and data
sheets were required to list the raw data obtained from each of the
-1-
-------
22 monitoring periods. While analysis, numerical models, and data inter-
pretation are the subject of this final report, it is apparent that the
wealth of the data collected can be the subject of additional research by
the authors of this document and by others who wish to study this larne
data base.
Section 1 of this volume is a brief outline of the final report;
Section 2 discusses the experimental design, the analysis procedures, and
the data interpretation. The GIOAP model and the steady-state TSP model
are two numerical models formulated during the course of this study in
order to predict indoor pollutant concentrations; the models are discussed
in Section 3 of this volume. Energy considerations are the subject of
Section 4. Section 5 discusses the relationships between energy conserva-
tion measures and indoor air quality in residences. Section 6 lists the
references cited in this report.
-2-
-------
SECTION 2
THE AIR QUALITY OF THE INDOOR RESIDENTIAL ENVIRONMENT:
EXPERIMENTAL DESIGN, ANALYSIS PROCEDURES, AND DATA INTERPRETATION
The air quality of the indoor residential environment will be charac-
terized by the information collected from the 18-mo monitoring program of
the indoor air quality assessment project. The monitoring design and
the initial format of the processed data are outlined in the opening sec-
tion. The question of air quality zones within a residence is addressed
next, followed by a discussion of the analytical procedure for investiga-
tion, which leads subsequently to consideration of each pollutant separately.
The data base is interpreted from the perspective of individual pollutant
behavior.
FIELD PROGRAM DESIGN: DATA COLLECTION
The objectives of this research program include the determination of
indoor pollutant concentrations, the identification of indoor pollutant
sources, and the estimation of indor pollutant concentrations. In order
to realize these objectives, an 18-mo field monitoring program was under-
taken. Indoor air quality was monitored for continuous periods of approxi-
mately 14 d in each of the five detached dwellings, two semidetached
dwellings (townhouses), six apartment units, two mobile homes, and one
school; in addition, a 5-d period of monitoring was performed in one hospital.
Three of the dwellings were referred to as experimental in the sense that
they were designed to conserve energy. The remaining dwellings are referred
to as conventional or by their structural type. These structures are located
in five metropolitan areas: Baltimore, Washington, D.C. , Chicago, Denver,
and Pittsburgh. The dwellings in Baltimore, Washington, D.C., and Chicago
were monitored twice to obtain seasonal variations.
Details of the monitoring design, the instrumentation, and the mobile
laboratory (van) used in the field study appear in Volume II. However,
brief descriptions of the above Items are provided In this section. The
mobile laboratory, fully equipped with the necessary monitoring equipment,
1s placed in close proximity to the structure being monitored. Gas con-
centrations are measured at one location outdoors (adjacent to the building)
and at three locations indoors (typically, the kitchen, a bedroom, and the
living room). Twenty-four-hour averages of particulate pollutants are
measured 1n the same four locations. These field observations have been
classified in seven generic categories; see Table 1. The energy data will
be discussed separately 1n Section 4 of this document.
Four-minute samples of gaseous pollutants are obtained three times
each hour from a continuous nonltorlng system which 1s used 1n conjunction
with a programmable solenoid switching mechanism at each of the four loca-
tions. Figure 1 presents a diagram of the sampling system for continuous
monitoring. Air samples from each location are carried in through Teflon
-3-
-------
tubing at a rate of 10 1/min. At the sampling trailer each sample line
is connected to a pump and a three-way solenoid valve. When the Teflon
solenoid valve is activated, the sample is introduced into a 91.4 cm (3 ft)
glass manifold. A pump purges a sampling line when the line is not acti-
vated. Each line is activated by a programmable timer system in a pre-
determined sequence. The monitoring sequence is as follows: outdoor,
kitchen, master bedroom, and activity room (often the living room).
TABLE 1. DATA CLASSIFICATION
1. Continuous monitoring:
2. Intermittent monitoring:
3. Physical data:
4. Energy data:
5. Tracer data:
6. Family daily logs:
7. Elemental analysis:
CO, NO, N02, S02, 03, CH4, THC, COz
TSP, RSP, SO4, NOj, ALD, Pb, and organic
compounds
Outdoor: wind speed and direction
Indoor and outdoor: temperature and
relative humidity
kWh for heating, air conditioning, and
total house consumption
Furnace efficiency
Number of door openings and closings
Structural specifications
House blueprints
Crack length investigation
Air exchange rate; indoor zone
identification
Daily occupant activity record
Proton-induced X-ray emission analysis
(PIXE)
Eight pollutants are monitored continuously by the above system: CO,
NO, NOp, S0«, 03, ChL, THC, and CO^. Specifications of the instruments
used in the project are provided in Table 2. Since meteorological condi-
tions affect the outdoor pollution levels, pollutant reaction rates, and
air exchange rates, and are therefore of importance in this study, they
are monitored continuously by Instruments listed in Table 2.
Twenty-four-hour averages are obtained for Total Suspended Particu-
late (TSP) matter and for Respirable Suspended Particulate (RSP) matter.
Indoor and outdoor levels for nitrates, sulfates, and lead are obtained
by chemical analysis of the TSP samples. Four-hour sampling periods are
required for determining indoor aldehyde levels (ambient aldehyde levels
are obtained from 24-h samples). Ammonia concentrations are obtained
-4-
-------
TO PUMP
tilt
® CONTROL VALVE
0 SOLENOID VALVE OPEN
(?) S0CEN0I0 VALVE CLOSED
I
(ft
I
SAMPLE
INLETS
i.
FLOWMETERS
¦
-------
TABLE 2. CONTINUOUS MONITORING EQUIPMENT SPECIFICATIONS EMPIOYED
FOR MONITORING INDOOR -OUTDOOR POLLUTANTS
Concentration (ppm)
Pollutant
Principle of
Detection
Manufacturer,
Model
Range(s)
Limit of Response Time to
Detection 90°/o or Greater
Precision
NO
NO (NO + NO )
X b
C°2
CO
°3
so2
ch4
THC
Wind/Speed Direction
Temperature, Relative
Humidity
Chemiluminescent Meloy NA-520-2
Chemiluminescent Meloy-NA-520-2
Nondispersive
Infrared
Nondispersive
Infrared with
flowing Ref. Cell
Beckman-865
Beckman-865
Chemiluminescent Meloy-OA-350-2
Flame Photometric Meloy-SA-185~2A
Flame Ionization
with Selective
THC Oxidizer
Flame Ionization
with Selective
THC Oxidizer
Direction-Synchro
Speed-d. c.
Magneto
Bimetallic Strip,
Human Hair
MSA-11-2
MSA-11-2
Bendix Arovane 141/120
Weather Measure H-311
0-0. 5
0-1. 0
0-2. 0
0-5. 0
0-0. 5
0-1. 0
0-2. 0
0-5.0
0-2,500
0-50,000
0-50
0-0. 5
0-0. 5
0-5
0-20
0-5
0-20
0-100 mph
0-540°
Adjustable
0-100°/o
0. 005
0. 005
25
500
0. 50
0. 005
0. 005
0. 05
0. 20
0. 05
0. 20
0. 5 mph
5°
1°F
1% RH
100 s
100 s
2. 5 s
2. 5 s
15 s
60 s
15 s
15 s
+ 1% Full Scale
+ 1% Full Scale
+ 1% Full Scale
+ 1% Full Scale
+ 2% Full Scale
+ 1% Full Scale
+ 1% Full Scale
+ 1 % Full Scale
+ 1% Full Scale
+ 1% Full Scale
-------
from hourly samples. Ammonia is introduced into the environment by clean-
ing the kitchen floor with ammonia cleansers. The instrument specifica-
tions for intermittently monitored pollutants are given in Table 3.
TABLE 3. INTERMITTENT SAMPLING AND ANALYTICAL METHODS
FOR INDOOR-OUTDOOR POLLUTANT MONITORING
Pollutant
Sampling Rate
(1 min)
Sampling Period
(hours)
Analytical
Method
Limit of Detection
(Working)
Total Suspended
Particulates
100
24
Filtration/ gravimetric
0.1 yg/m3
Respirable Suspended
Particulates (3.5yg)
SO
24
D ichotomous/ gravimetric
0. 1 yg/m3
Organic Vapors
0. 2
24
Charcoal absorption/
gas chromotography
ppb as CH4
Aliphatic
Aldehydes
0.5
4
Bubbler/MBTH
1. 5 yg/m3
Ammonia
0. 5
1
Bubbler/phenate
5 yg/m3
Sulfates from
TSP Samples
100
24
Filtration/methyl-
thymol blue
0. 5 yg/m3
Nitrates from
TSP Samples
100
24
F titration/ brucine
0.1 yg/m3
Lead from
TSP Samples
100
24
Filtration/ atomic
absorption
0. 005 yg/m3
Elemental Analysis
Atomic No. 16
through 35 Plus
No. 82
1
Continuous
Streaker sampler/
PIXE
ppb to ppt
Time sequence total filter samplers, designed at Florida State Uni-
versity (FSU), are used to obtain a continuous time record for elemental
analysis. A Nuclepore filter strip is placed over the intake to produce
an 85-mm long strip sample of one-week duration and 2-h time resolution.
Proton-induced X-ray emission (PIXE) analysis is performed on these
aerosol samples by sequentially bombarding 2-mm wide segments of the strip.
Spectral analysis, formulated by Dr. J.W. Nelson of FSU and his team, is
utilized in this project for the determination of elemental content of
aerosols.
-7-
-------
The various types of data (continuous pollutant data, physical data,
etc.) are collected via strip chart recorders; the data are then taken
from the strip charts and entered onto coding forms (each type of data
has its own specific coding form). Next, decks of punched cards are pre-
pared for each type of data from the coding forms. Then each card deck
(and in certain cases additional inputs) is used as an input to a program
that creates a disk file. The result is a set of three disk files: con-
tinuous pollutant data, physical data, and 24-h sampling data. These
files are used as inputs to a collection of programs that produce data
tables and reports. Finally, the three disk files are input to a program
that creates the data tape used by GEOMET and also produces a listing of
the entire tape. For more detailed information concerning the data pro-
cessing procedure described in this paragraph, refer to Volume II.
CHARACTERIZATION OF THE OBSERVED INDOOR AIR QUALITY
The large data base generated by the 18-mo monitoring program pro-
vides a unique opportunity for detailed analyses of isolated pollutant
events and statistical interpretation of identified patterns. The out-
door pollutant concentrations, the indoor pollutant sources, the air
exchange rates, the chemical nature of each pollutant, and the behavioral
patterns of the inhabitants are among the parameters that determine indoor
residential air quality. The observed variation of indoor pollutant con-
centrations is caused by a complex dynamic system which involves these
and other parameters. The dynamics of this system are examined in sub-
sequent discussions on the formulation and validation of the GEOMET Indoor-
Outdoor Air Pollution (6I0AP) model. The short-term history, time
periods of 24 h or less, of residential pollutant concentrations can be
numerically simulated by the GIOAP model. However, the establishment of
identifiable pollutant variation patterns and the characterization of the
indoor air quality require extensive interpretation of the observed data.
Four-minute average pollutant concentrations are obtained sequentially
from the four sampling sites. While the 4-min averages can be used for
a detailed examination of the observations, the fundamental concentra-
tion for data interpretation is the hourly average concentration. The
initial step 1n the analysis of the data relates to the uniformity of
pollutant levels indoors. Corresponding hourly average pollutant concen-
trations 1n the three indoor sampling sites are not always equal. The
problem Is best seen 1n terms of the following questions. Do Indoor zones
(Independent areas) with distinct pollutant patterns exist? Does the
hourly average of the three corresponding indoor concentrations adequately
characterize the residential air quality? Statistical tests easily answer
the questions; however, the problem is not simply a statistical one but a
many faceted one. Air pollution measurements are made in order to determine
-8-
-------
the potential impact of high pollutant concentrations on human health.
Therefore, conclusions on the existence of independent indoor zones must
consider the health significance of the existing indoor concentration
gradients in addition to the statistical results. Thus, it is the
magnitude of the differences that is of importance and not the existence
of the differences.
The null hypothesis, tested by a two-tailed, paired t-test, is that
the mean of the differences of the concentrations of corresponding hourly
pollutant averages from two indoor sites is equal to zero. For each resi-
dence investigated, a table 1s generated which summarizes this statistical
test for all pollutants. Table 4 illustrates the results for the Chicago
experimental residence. As the table indicates, the null hypothesis is
rejected in more cases than 1t 1s accepted. Comparison of the observed
range and the calculated mean of the differences for the illustrated
residence, and for all residences sampled in this project, leads to the
following resolution of the question posed earlier: corresponding hourly
indoor pollutant concentrations are not uniform throughout a residence,
but the existing differences between sampled indoor sites are small and
probably of minimal health significance. The data interpretation will
therefore proceed by assuming that one indoor hourly average concentration
per pollutant characterizes the residential air quality for that hour.
The one-zone representation of observed data critically reflects on
the data analysis. The following remarks will help to clarify this inter-
pretation of the data:
1. In this extensive analytical study of indoor air quality, Shair
and Heitner (1974) assumed that no pollutant gradients exist in the indoor
environment. The data base of this study verifies that the pollutant
gradient in the residential environment is negligible. A study of the
air quality in a scientific laboratory by West (1977) shows an almost
uniform distribution of an inert tracer continuously released in the
laboratory. Similar experiments in residential environments show that
an equilibrium situation is reached throughout the house within an hour
(see Figure 2).
2. The majority of the dwellings sampled 1n this project had an
air exchange rate (v) of 0.4 air changes per hour or more. Distinct
zones may exist in houses with very low air exchange rates (v < 0.2 air
changes per hour) because dispersion of the Indoor pollutants Tn these
dwellings 1s expected to be very slow.
3. The one-zone concept does not require instantaneous mixing
because it is based on the behavior of hourly average pollutant con-
centrations.
-9-
-------
TABIJ-: 4. STATISTICAL SUMMARY OF THE ZONE CONCEPT
Level of Significance 0. 01
Building: Chicago Experimental Residence, Visit #1
Zones
Pollutant Kite hen-Bedroom Kite hen-Livingroom Bedroom -Livingroom
Observed ~ ~ —~~ ¦
df Mp ff[) t df pD CTp t df ll£) op t
CO 311 - 0.09 0.37 -4.35 310 0.02 0.37 0.74* 311 0.11 0.45 4.19
SO2 344 - 1.40 3.42 -7.58 34S -0.61 3.15 -3.62 344 0.80 1.72 8.61
NO 348 - 0.88 4.42 -3.70 347 -0.68 4.31 -2.96 347 0.18 4.99 0.67*
NO 2 345 0.67 4.35 2.86 344 0.20 4.44 0.82* 345 -0.43 4.49 -1.79*
03 349 0.00 0.03 0.28* 349 0.00 0.03 -1.00* 349 0.00 0.06 -0.63*
CH4 329 0.02 0.18 1.56* 329 0.01 0.13 1.87* 32 9 0.00 0.15 -0.24*
C02 347 -27.62 105.14 -4.90 348 -0.01 67.37 0.00* 346 27.79 127.84 4.05
THC-CH4 316 - 0.11 0.56 -3.35 316 -0.37 0.90 -7.34 318 -0.26 0.90 -5.19
TSP 13 12.95 13.95 3.47 12 2.78 14.89 0.67 12 -9.24 5.05 -6.60
SO4 13 - 0.04 1.45 - 0.09* 11 -0.07 1.07 -0.22* 11 -0.20 1.33 -0.52*
NO- 11 - 0.02 0.15 -0.38* 10 -0.05 0.07 -2.65* 10 -0.07 0.13 -1.90*
3
df - degrees of freedom
]j£) - mean of the differences
j - standard deviation of the differences
t - paired t-statistic
* - accept the null hypothesis; i. e. , HqI = 0
-10-
-------
BALTIMORE EXPERIMENTAL
JANUARY 12, 1977
TIME 0935
i—
i
t
>
»
<
»
* «
»
V
A
~
jP
~
¦
*
it
~a
<
i
>
m
~
A
$ m
V
~
;
Ai
itairw!
Jving
ledroo
4 aster
11
Room
1\ #2
Bedro<
m
10
-10
10
.-11
15 30 45 60 75 90 105 120 135 150 165 180 195 210
ELAPSED TIME, MINUTES
Figure 2. Equilibrium situation.
-11-
-------
4. In certain cases the observed differences are statistically
significant, but the one-zone concept does not require uniform residen-
tial concentration, rather it states that the existing differences between
the average house concentration and the concentration of each separate
site are very small and of minor significance to the well-being of the
inhabitants.
The interpretation of the large data base generated by the monitoring
of the residential environment endeavors to identify significant relations
between pollution levels indoors and ambient pollution concentrations
and the occurrence of augmenting or sheltering effects caused by residen-
tial structures. The air quality information sampled from each of the
monitored dwellings is routinely summarized in a series of descriptive
tables, statistical illustrations, diagrams, and histograms. In agreement
with the one-zone concept, the average hourly indoor and outdoor concentra-
tions of each of the pollutants monitored are listed in Table 5. Long-term
averages are more suitable for identifying daily patterns than are the hourly
averages, because the latter are influenced by transient variations. Three-
hour, 8-h activity-period averages and 24-h averages are routinely generated
for every gaseous pollutant. Table 6 is an example of these averages for
both the indoor and outdoor pollutant (in this case CO) concentration. Pre-
vious studies have calculated night and day averages for corresponding outdoor
and indoor pollutant concentrations; Table 6 shows that in this project
"activity" averages have been estimated. The first period (hours 0-6)
represents the nighttime period with minimal indoor inhabitant activity,
the second period (hours 7-19) is the daytime average which summarizes
the typically active hours of each 24-h period; and the third period (hours
20-23) shows pollutant concentrations averaged over the transient period
between (the active daytime hours) and the inactive nighttime hours.
The data base generated for each dwelling is considered a separate
entity or unit of data in the analysis of observations. For each such
unit the observed concentration range for a pollutant is of interest.
Table 7 illustrates a summary of the concentration ranges for each day
of the monitoring period and for the entire monitoring period for all
sites. While these bounds show the extreme concentrations, they provide
no information on the concentration distribution of the sampled pollutants.
The percent distribution (for an example, see Table 8) depicts how the
observed values are distributed. The concentration range for a given pol-
lutant is divided into 10 intervals, and the frequency concentrations
occurrring within each interval is calculated. Information presented in
the percent frequency distribution table may be graphically illustrated
as 1n Figure 3. The range of the observed pollutant concentrations is
divided into 10 intervals denoted by index numbers from 1 to 10 for each
-12-
-------
TABLE 5. DAILY INDOOR-OUTDOOR CONCENTRATION SUMMARIES
T6"* :
p^fiL^rn vr • rn
Pf)ir,t!TANT UMTTP! PP«
nni'ns
t>Afr
L.UC
0
I
?
1
4
S
7
p
q
10
1 1
1 7
i 3
1 4
15
1 ft
17
19
19
20
21
22
23
7/
=>/ J
•jii r
-1 .
-i. o
-1.0
-1.0
-1.0
-1.0
-1.0
-1 .0
-1.0
-1 .0
0.0
1 .0
1 .0
1 . 0
1 .0
1.0
1.0
1.3
1.0
2.0
-1.0
0.0
0.0
0.3
1 /
S/ 0
l i
-1.0
-1
-1.0
-1 .0
-1 ."
-1.0
-1.0
-1 .0
-1.0
-1 .0
7.4
3.1
7.9
3.0
3.0
1.1
3.2
3. n
2.8
2.8
-l.o
1.3
l.o
1 .9
7/
5/1"
hi r
0.11
r>. <>
0.0
0.0
0.3
0,0
o.o
(1.0
1.1'
0.3
n.o
0.0
O.n
n.n
0.0
0.7
0.5
n.n
0.0
o.n
0 . 0
0.0
0.0
0.3
7/
s/m
i»
1
1 . 7
! . 1
I.«
1 . "
1.7
1.4
1 .7
7.7
2.1
1.»
) . 7
1.1
1 . 1
1 . 1
7.0
7.2
1.1
1.4
1 .7
2.2
2.1
1.7
1.9
7 /
5/1 1
HIT
0.7
1.3
0.7
o.n
o.o
0.3
1.7
3.0
1 .7
0.3
n.n
0.0
0.0
o.o
0.0
0.0
0.3
1 .n
0.3
0.7
1.0
0.5
7.7
2.7
7 f
S/1 1
1 4
? . 1
?.*<
1 .»
1 .7
' .3
1 . 3
2.2
3 .ft
« . 3
4.1
3.0
1 .«
7.1
7. 1
7.3
3.0
i.O
2.9
7.1
2.0
1.8
1 . 1
7.7
3.2
U
5/1 ;>
Til T
t.J
1 . ' >
• .0
0.7
o.o
n.3
1.3
3.3
3.0
3.0
0.3
0.0
0.0
0.0
0.0
0.0
0.0
n.o
0.0
0.0
0.0
0.0
0.0
0.0
7/
5/1 /
1 !
*. 1
>. 1
7.0
1.0
1 .7
1 .7
1.0
1.3
4.4
4 ~ ft
3.7
2.1
1 .3
1.3
1.3
0.9
O.n
0.9
0.9
0.3
0.1
0 . 0
1 .7
o. n
7 ~
S/t 3
VfT
".0
o.n
o.o
n.n
0.0
o.n
0.0
0.0
0.0
0.0
7.0
0.3
1.0
I ."
o. 3
0.0
0.0
0.7
0.3
0.3
2.0
1 . 3
0.0
0.0
7 /
S/t J
I *
1
o.o
1 .1
1.3
1 .«>
1 .°
1 .9
1 .7
1 .9
7.1
3.4
2 . P
7.R
7.4
1 .9
1 .6
7.0
7.2
1 .9
2.0
->.o
2.0
1 .ft
l. 1
"7/
S/T*
nsfT
~ nrr
" o.^?
0.1
-t;tt
- r.n
1.0
'1.0
.3
0.7
0 . 0
1 .n
-1 .o-
-1 .0
"-1.0
-o.o
-o.o
O.o
n.fl
- 0.0
0.0
0.0
o.n
1.3
1 .3
7/
S/l 4
J 'i
1 .0
1.9
1. >
?.i
2.0
1.7
1 .7
1 . P
1.4
1 .5
-1.0
-1 .0
-1 .0
1 .S
7.3
2.3
2.0
2.0
1.0
1.0
1 .4
1.8
2.0
7/
5/15
rinT
7.?
1.7
0.3
n.o
n.o
n.n
0.0
O.n
1.7
o.o
3.0
7.0
7.3
1 .7
1.0
1 ."
0.3
1 .0
2.0
2.3
2.3
1.7
0.3
n.3
If
5/15
i *
f . 1
3
1
1 . 3
1 .*
1 .»
O.rt
0.4
0.9
7.6
2.7
2.6
3.7
3.7
7.3
7.1
1 .0
1.7
7.7
3.0
7.9
2.8
1.9
1.1
7 /
5/1 ft
in r
1.0
n.n
n.n
n.n
n.n
n.n
1.3
1 .3
? . 3
3.0
3.3
1 .3
O.O
0.0
0 . 0
0.0
0.0
o. n
0.0
0.0
1.0
7.3
7.0
1 .7
7 /
•i/l*
!•*
1 .7
2.0
1 .6
1. t
0.7
0.7
1.3
7.7
7.7
3.9
5.4
4.3
7.2
1 .ft
7.1
2.0
7.0
2.3
7.6
2.3
1.6
7.3
7.9
3.0
r/
5/17
TltT
t .n
-1 ,n
1-o
-tr;*-
— o~; 0
0.0
0.7
1 .7
7.0
7; 7
1.0
0.0
0-.7
1 ;ft
0.0
0;0
0 ; 0
n.o
0;o
0 ; 3
2 ; 0
4.0
5 ; 0
5.0
7/
S/17
l'«
?. 3
>. i
2.1
1.4
1.1
0.9
1.1
t
2.9
¦5.0
5.7
4.8
3.2
3.0
7.R
2.0
1 • P
1.5
1.0
0.8
1.3
3.0
4.0
4.7
7 /
5/18
TUT
i.r>
3.<1
7.1
7.o
2.0
2.7
3.0
4.0
3.7
2.3
1.5
0.7
O.o
0.3
0.3
0.0
1.0
1.0
o.o
0.0
0.0
0.0
0.0
0.0
7/
s/i*
H
*. t
4. 3
3.»
3.3
3.0
3.2
3. 3
3.4
4 . 1
4.1
3.3
1 .«
1 .*>
1 .6
1.2
1.4
1.7
2.6
1.8
1.3
1.7
1.8
2.6
2.7
7/
5/ 1 u
TIT
o.3
n. 3
".0
0.0
0.0
n.n
0.0
1 . 3
0.0
o.o
0.0
1.0
0.7
1.3
1.0
1.0
0.3
0.3
0.0
0.7
1.0
2.0
1.0
0.7
7/
s/n
M
i. 3
7.1
1 .3
1 .3
t .3
1.?
1.7
1 .*
7.3
1 .<
1.0
2.4
2.*
3.0
7.1
1.0
0.9
0.4
1.1
1.2
t .7
I.ft
l.ft
1.2
1 /
•>/;*«'
M)T
•>. *
'J. 7
t .
0.0
-0;0
0.0
0; 0
1 .0
1.3
1.0
1 ."5
1 i7
1 . 3
1 .3
1.0
0.3
0,5
7*0-
0.7
1.3
3,7
6. 3
6.0
5.0
11
* /*<>
1 i
1 . t
/. 3
?.n
1.4
1.2
1.0
1.0
1.3
7.0
7.2
3.2
3.1
3.7
4.0
3.3
2 . ft
1 .9
4.0
7.7
2.9
'.7
6.1
7.4
7.0
7 /
5/21
O'fT
7.7
3.1
7. '
7.0
1 .0
1 .n
1.3
1 .0
1 .3
1 .0
1.3
3.0
4.0
4.0
3.7
2.3
0.5
1.0
1 .0
1.0
1.7
2.3
3.0
0.7
7/
r>/?1
f.:
¦>.4
*.*>
' .9
1.*
3.2
2. i
2.3
2. 3
7.6
7.7
7.9
4.9
s.1
5.P
4.3
2.7
1 .5
1.1
1 .0
1.0
2.1
3.1
3.3
2.ft
/~
S fit
VIT
'».7
0. ?
0.7
1 . i
1 .0
O.o
0.3
O.o
0.7
0.3
1.0
0.3
0.7
0.3
1.0
0.7
0.0
0.0
0.0
0.0
1.3
3.5
4.3
3.3
7/
1/22
f t
1 .*
1. V
1 .u
1 .7
1.4
1 . 1
0.4
1 .3
1 .«
2.ft
2.S
1.3
t .0
1.1
1.0
1 .0
1.7
1.6
1 .8
1.4
2.2
3.8
4.9
4.8
7/
1//3
1U+
».o
?i"7"
?. 3
?.<>
~Ti o
7.0
3 ; 3
3.3
3 ; 7
1.3
i.y
1.0
-1 id
-1 .0
-1 iO
-1 iO
-1,0
-no
-1,0
-1 .0
-1.0
-1,0
-1.0
-1 ,0
7 /
S/^3
I '
* . *
3.7
3.»
7.7
7.2
2.1
7.6
3.0
4.0
3."
3.1
2.9
-1 .0
-1.0
-1 .0
-1 .0
-1.0
-1.0
-1.0
-1.0
-1.0
-1 .0
-1.0
-1 .0
* The designation (-1.0) indicates missing data.
-------
TABLE 6. POLLUTANT CONCENTRATION SUMMARIES FOR 3-h, 8-h, ACTIVITY PERIOD, AND 24-h AVERAGES
qm! q
VTSfTH! 1
rnr.UJTRMT! CO
Pftr I.IITR'JT UNITS! PPM
ns i f;
I.OC
0-7
3-5
7 /
5/ 9
iin r
-1
-1.0
7 /
5/ a
r u
-1.0
-1.0
7/
5/ t 0
0(1 T
0 . ('
0.1
7 /
5/1 '<
1.1
1.3
1.7
7 /
s/tl
hut
n.i
0.1
7 /
5/1 1
in
2.7
1 . 3
7 /
5/17
mi T
1.1
0.3
7 /
5/1 7
J N
>. 1
1 .4
7/
5/13
IlliT
o.o
o.n
7 /
K / \ 1
? fJ
1 .'1
1 .6
7/
5/1 *
miT
0.4
1 ;¦(>
7 /
¦> /1 a
1 it
1 .7
7.0
7 /
5/1S
out
' .4
0.0
7 /
5/1 S
1 n
7.1
1.3
7 /
5/16
miT
0.0
0."
7 /
5/1 h
IN
1 . 7
O.R
7 /
5/17
OUT
1 .0
0 . 1
7/
5/17
Hi
2.2
1 . 1
7 /
5/1 «
mi r
3.7
7.7
7/
5/1 B
IN
4. 7
3.?
7 /
5/19
out
0.7
0.0
! /
5/1 °
I :•
1.9
1 . 3
7/
5 / i 71-73
r -Mnim
0-7
a VFPAGFS
H-15 16-71
ACTIVITY
0-fr 7-
avfri\gf:s 74-nn-jR avfragf
iq 70-23
i.
5 .
0 .
1.
0.
7 .
0.
1 .
" .
7 .
0.
1 .
1 .
7.
0 .
7 .
o.
3.
0.
1.
1.
7 .
1 .
3 .
3.
5.
0 .
1 .
-1 .
-1 .
1 . 1
3.1
0.1
1.7
0. <
1.0
0.0
O.q
0.?
7.CI
0.0
2.2
o. a
1 .5
o.n
2.3
0.0
1.«
0.7
1.9
o.*
o.n
0.7
2.6
1.4
t . "
0.7
1 . 4
-1.0
-1.0
1 .4
o.n
1 .H
n.7
7.0
n. o
f>.4
o.q
7.0
o.o
! .3
7.7.
7.0
0.1
7.1
0.8
1 .0
0.0
1 .6
O.fi
1 . 3
1.9
3.1
1 . 7
1 . 4
0.4
1 .A
-1 .<»
-1.0
0.1
1 . 4
0. 1
1.«
7.1
7.3
n.o
0.4
0.4
1 .t
0.9
1 .7
0 . R
1
2.0
7.7
4
3
0
7
1
1
5.9
6.9
?.0
3.0
3.7
4.5
-1 .o
-1.0
-1 .0
-1.0
0.0
1.3
t .0
2.0
1.1
7.1
o.o
1 . 3
0.fl
1.»
0.5
1.4
0.3
1 .4
0.7
1 .*
7.9
3.6
0.7
1 .5
0 . 4
1.4
1 .«
3.5
0.5
1 . 3
7.6
3.0
O.q
?.q
0.7
1 .5
0.3
?.«
0 . R
7.4
0.5
7.3
0.3
1 .0
1.1
2.4
1.?
3.0
o.q
3.6
1 . 1
2.3
0.6
7.0
1.7
3.0
2.6
?.»
o.«
1.1
1.9
3.5
0. R
7.3
0. 1
t.H
t . 7
2.3
0.0
O.S
0.6
1 . «
0.3
1 . 7
1 . 3
2.1
O.fl
7.4
7.0
7 . 3
0.3
1.9
0.7
1.7
3.4
4.6
1.4
2.0
1.5
2.R
-l.o
-1.0
-1.0
-1.0
0.0
1.3
0.7
1 .n
O.S
2.0
0.0
1 .0
7.9
0.2
1 .6
0.6
2.R
0.7
2 . 0
0.4
2.2
0.3
1 .7
1 . 3
2.2
o.q
2.9
0.7
2.«
1 . 1
2.3
0.6
1 .6
1 . 1
2.R
2.0
2.9
0.4
1.5
2.2
3.4
0. 1
1.4
0.1
2.0
1 .«
2.1
0.0
0.3
o.n
2.4
4.0
3.3
0 . 0
2.1
1.2
1.5
5.3
6.1
1.9
7 . R
3 . 1
3.9
-1.0
-1.0
0.3
0.5
1.6
1.6
3.0
1.9
3.1
0.9
1.9
2.4
3.2
* The designation (-1.0) indicates missing data.
-------
TABLE 7. MAXIMA AMD MINIMA FOR I-h, 3-h, 8-h, AND DAILY A VERACE POLLUTANT CONCENTRA TIONS
StTP»s 1
VISIT#: 1
POLLUTANT* CO
por^orini unitss ppm
PATE .-ZONE. . . 1-HO.lJP. AVERAGES 3-HOUR AVERAGES 8-HOUR AVERAGES DAILY AVERAGFS
MN/DD/.T Y MTN MAX *IN "AX MIN WAX KIN MAX
7/12/76 AM0IENT 0.00 2.33 O.U 1.89 0.75 1 .21 1.00 1.00
1/12/15 KITCHEN _ 0.00 3.00 0.11 2.33 0.37 l.M 1.11 1.11
7/12/76 BEDROOM 0.00 2.33 0.11 2.11 0.46 1.48 1.10 1.10
J/1.2/7.6 LIVI.NGROOM ..... 0.00 2.33 0.00 2.1 1 0.37 1.50 1.06 1.06
7/12/76 INDOOR AVERAGE 0.00 2.44 0.07 2.1 9 o_40 1.53 1.09 1.09
7/13/.76 AMBIENT 0.00 4.67 0.00 4.44 0.00 3.75 l.«5 1.85
J7/13/.76 KITCHEN . 0.00 5. 33 0.00 5.00 0.00 4.37 2.15 2. 15
7/13/76 BEDROOM 0.00 5.00 O.no 4.67 0.00 4.04 2.06 2.06
.7/13/76 . [il VINGROOM 0.00 5.00 0.00 4.78 0.00 4.17 2.OB 2.OS
7/13/76 INOOOR AVERAGE 0.00 5.U 0.00 4.01 0.00 4.19 2.10 2. 10
7/14/76 AMBIENT 0.00 4.33 0.33 3.89 0.83 3.42 2.03 2.03
J/.14/76 KITCHEN .1.00 4.67 1.44 4.22 1.50 4.17 , „2.*3 2.63
7/14/76 BEDROOM 0.33 5.33 1.22 4.44 1.54 4.21 2.61 2.61
.7/14/74 bJVINGRQQM 0.33 5.67 1 .11 4.89 1.4? 4.46 2.57 2.57
7/14/.76 INDOOR AVERAGE 0.56 5.22 1 .30 4.52 1 .49 4.28 2.60 2.60
7/15/76 AMBIENT 0.00 7.33 0.44 5.99 1 .00 5.OH 2.66 2.66
3/15/76 JCIICHEN- 1.33 4.00 - 1 .44 3.67 1 .92 2.88 2.44 2.44
7/15/76 BEDROOM 1.00 4. 33 1.22 3.89 l.«»l 2.79 2.4fi 2.46
7/15/76 blVIMGPOON 1.00 4.67 _ 1.33 4.00 1 .96 2.71 - 2.45 2.45
7/15/.76 INDOOR AVERAGE 1. 1 1 4.22 1.33 3.85 1 .93 2.79 2.45 2.45
7/16/76 AMBIENT 0.00 3.00 0.44 2.56 1.09 1 .3* 1.21 1.21
7/16/7*. KITCHEN _ 0.33 3.67 0.89 3.22 1.33 2.58 _ 1.99 1.99
7/16/76 BEDROOM 0.33 3.67 0.78 3.44 1.25 2.54 1.99 1.99
1/16/16 JjIVIKGRODI .... ... 0.67 4.00 0.89 3.67 _ 1.38 2.63 2.0L 2.07
7/16/76 INOOOR AVERAGE 0.44 3.5S 0.85 3.44 1.32 2.58 2.01 2.01
7/17/76 AMBIENT 0.00 5.00~ 0.00 " 3.67 " 0.17 " 2.79 1.46 1.46
.2/17/76 KITCHEN __ __ 0.00 5,00 0.00 4.1 1 0.21 3.38 2.07 2.07
7/17/76 BEDROOM 0.00 5.33 0.00 4.44 0.46 3.42 2.15 2.15
J/1J/7S LIVINGRODM 0.00 ^5.00 _ 0.00 4.33 0.50 3.33 2.17 2.17
7/17/76 INOOOR AVERAGE 0.00 5.11 0.00 4.30 0.39 3.38 2.13 2.13
7/18/76 AMBIENT 1.00 4.00 1.11 3.56 1.22 2.67 2.05 2.05
7/18/76 KITCHEN 2.00 3.67 2.44 3.11 2.79 3.O0 2.88 2.88
7/18/76 BEDROOM 1 .67 3.67 2.44 3;00 2.75 2.94 2.83 2.83
7/18/76 LIVINGROOM 2.00 3.33 2.44 3.17 2.63 3.06 2.81 2.81
7/18/76 INOOOR AVERAGE 2.00 3.44 2.44 3.05 2.72 3.00 2.84 2.84
OVERALL AMBIENT 0.00 7.33 0.00 5.89 0.00. 5.08 0.44 3.84
OVERALL KITCHEN 0.00 5.33 O.OO 5.00 0.00 4.37 0.48 2.88
OVERALL BEDROOM 0.00 5.67 0.00 4.67 0.00 4.21 0.48 3.97
OVERALL LIVTNGROOM 0.00 5.67 0.00 4.R9 0.00 4.46 0.49 3.89
OVERALL INOOOR AVERAGE 0.00 5.22 0.00 4.81 O.OO 4.28 0.49 3.93
-------
TABLE 8. PERCENTAGE FREQUENCY DISTRIBUTION
S T TP fJUMBKR: 1
VISIT NUMPEP: 1
P 0 L b U TS'iT: r n
IJ \< T T S t PPA!
ZD!" r: ni.'Tnnop
IF HO'JPS: 3*5
T'-JTRR V A Tj p tr R n [T M t A G p
< o . oT " ' o. n
t o.oo,
0.73 5
<13.4
[ 0.73,
1.471
17.0
t 1.47,
7.201
13.?
[ ? . ? o ,
2.931
5.?
1
[ 2.93,
3.6ft]
10.1
a*
C * . ft
4.10)
7. R
1
C 'i.4n,
5.13)
1.0
[ «?. n,
5 . H ft )
o.s
[ "i , R r> ,
ft.60)
o.s
[ fr.^n,
7.33)
0 . 0
II
a
c
A
7.3 3
n.T
n r^s.:
1. } (,) TfJOITATRS that THF TMTpPViVf, T S 2 f.-OSR •"> OKi TF R LRFT A*JP H P R M ON THE RIGHT.
?) -1. imn1c& vrs ft mi ^simr, v.* i,ue.
-------
RESIDENCE: WASHINGTON EXPERIMENTAL HOUSE
POLLUTANT: 03 PPB
INTERVAL RANGE: 14 PPB; EACH INTERVAL IS CLOSED OT THE LEFT AND OPEN ON THE RIGHT, [ )
100* 1 1 » >i 1 1 y 1 1 1 1 1 i 1 1 1 1 n 1 11 1 11 1 1 1 1 1 1 1 1 1 1 1 1 1 11 1 r 11 1 1 1 1 1 1 i 1 1 11
Index 12345678910 123456789 10 12345678910 12345678910 123456789 10
Number Kitchen Bedroom Li ring Room Indoor Average Outdoor
Figure 3. Percentage of hourly average Og concentrations.
-------
sampled location. The vertical axis is the percentage of hourly averages
that the pollutant concentrations are in the various intervals indicated
by the index numbers on the horizontal axis. This particular example
indicates that the observed ambient ozone concentrations in the Washington
experimental house are above the 1-h NAAQS for 10% of the monitored hours
(approximately 32 h). It also shows that the indoor ozone concentrations
are extremely low, well within the NAAQS. Thus, this particular dwelling,
with no apparent indoor ozone sources, provides a shelter from high ambient
ozone concentrations.
A widespread assumption made by the public is that the indoor environ-
ment provides a shelter from high outdoor pollution levels. The ratio
frequency distribution generated by calculating the ratio of hourly ambient
pollutant concentrations over the corresponding indoor average concentra-
tion investigates the validity of this general assumption. Interval values
from 0.0 to 5.0 in steps of 0.25 are considered for the ratio values; the
percentage of the hours for which the ratio falls in any given interval is
tabulated for each pollutant (see Table 9).
Detailed analysis of the data requires an examination of all observa-
tions; however, a summary of the sampled concentrations has been generated
as an indication of the data collected by this project. Table 10 is an
illustration of such a summary; it shows the observed total concentration
ranqe, and the observed ranges in 85% of the concentration values. It
also shows whether or not there has been a violation of outdoor and indoor
(if any) pollutant standards. In addition, it demonstrates the existing
relationship between the outdoor and indoor values.
The tables described above are among the techniques used to interpret
the data and identify typical patterns. In addition, occurrences of maxi-
mum concentrations are of interest. Table 11 identifies time periods with
high pollutant levels.
Each of the techniques outlined above is applied to a given house-
pollutant combination and has been utilized in the pollutant data inter-
pretation. Additionally, factors such as interelement relationships,
seasonal comparisons, family composition, and fuel used for cooking have
been examined and will be discussed in the balance of this section.
Carbon Monoxide (CO)
Indoor CO concentrations are generally higher than corresponding
outdoor levels in all residences monitored. High indoor concentrations
may be attributed to two factors: 1) indoor CO emission sources, such
as gas-fired cooking appliances, attached garages, faulty furnaces, and
cigarette smoking; and 2) the residence time of CO levels is apparently
longer in the indoor environment than the time required for ambient
peaks to dissipate to lower levels.
-18-
-------
TABLE 9. RATIO FREQUENCY DISTRIBUTION
s r CK WIIVHKw • 17
VTSTT Nil MRFP • 1
PQLMITUNT! CO
I I'M TS : PP"
mrin; nfniFNr /imooor avtenn11"
MIIMHIvH TF HOIIPP! 370
INTFKVM. PF.RrFMTPCT
< o.oo o.n
o.oo, n.?5) pi.?
o.?5, n.501 p.a
O.50, 0.7S) 4.Q
0.75. I.oo) 1.*
1.00, J.25) 0.5
1.25, 1.50) O.S
1.^O, 1.75) o.o
1.75, 2.00) 0.0
7.00, ?.25) 0.3
2.2*, 2.501 0.0
2.50, 2.75) 0.0
2.75, 3.00) 0.0
5.00, 3.25) 0.3
1.25, 3.50) 0.0
3.50. 3.75) O.O
1.75, 4.00) 0.0
4.00, 4.75) 0.0
4.25, 4.50) 0.0
4.50, 4.75) 0.0
4.75, 5.00) 0.0
> OR = 5.00 0.0
NI1TFS:
I) l.) fwr»TC*TES THAT THF T^TfRVHl, 1 «5 Cl.nRFI ON THE t.FFT *NP OP^N ON THF PTGHT.
?) -i. i»n>rr»Tes a Mtssmn vm.hf.
-------
h - Sbpf
TABLE 10. POLLUTANT SUMMARY FOR COppm
' *£ h - 2.b pp>n
%&?\VZUOb
TIM£
tZKSOV
£HJfXP£>CI^
TcTAU yth'U
IMPSOR
TOTAL-
AV6.
»6?4
KrfCMB-n
ToTAL- && %
fOTAU &•» y«
UVIM£ ftaow
TOTAL. &t>
fK£#ue-N6Y P^TK)B>aTl6>Ki
Of AH&imT/tHRpOK VAUUa
1 HAJ£
£X~ZO
az
0.-Z-O
OZ
o-zo
o.z
o-2-C>
<>Z
0.-2-&
£>Z
&Xflg«ll^&NTAL.
X.
i> HOJK6
A- I -1
m
m
O- 1-6
1
m
W
'//.<
m
'f//
O.- \-°i
o.~ .71
w
/
C>. - . £2>
ip
a-n*>
m
U k\OU&CL-.A&
m
%
o.-.(cC
y//
o.-.44
O.-.iob
o.'.
m
m .
JL
\ ttOJR
59
^-5.2-
3«?
<>.-«5.^
33
C.-S7
2>-5
©.-6.7
3.9
:0O»
90K
<£6N\>els3T\ONAu-
3.-
5 HctJfW>
«
£-4-6
1
O.-^-O
a.- 4.7
I
c> Houee>
o.-<=>.e>
0.-4*,
O. -4.1
0.-4.5
1A H0U&
4-3.e
I
.6- 51
• S-Z1
i
6-4/3
.6-3.1
22
bL
H£>£JK
£>-<£*>
C>.-A-Cs
\A
1-4
&.-A.O
1.4
£-5.7
\A
^o»4ve**T>o»^Au-.
3-.
?>W>lif&
e.~\.*\
1
m
e> woa^p
m
§
O.'S.Co
§
o.-ze>
I:
o.-4*>
m
O.-^O
o - S.Z.
0.-2.6
i
GrM
Z4*tott!£
.Ol ~1*>
I:
ife-Z3
-21-2.4
• I/-20
(1
• \4-2A
1 WUK
&.-40
0.-A.Z
ZO
ex-4.3
2.0
o.-<*n
zo
0.-4.0
2.0
VlA6H\M6T^H
EXP&K)M6MTAu
5..
3
O.-S. (p
W
///'
r-/rh
0--3.S
W
0-1.1
& t£M&>
o.-z-i
m
o--so
0.-4.2.
o. -ze
rrp-.
/.
¦ /'
rf/p
it
o. - 3.t>
. /
'/
W;.
o.- 5.S
»
O.-2-A
2A
07-|(o
t:i:
11- ZS
-21-2,6
/¦ v
SO-24
¦///•¦/
¥
TfTT;
m-z<*
-------
TABLE 11. IDENTIFICATION OF TIME INTERVALS WITH MAXIMUM POLLUTANT CONCENTRATIONS
Hou.uTANr c.07 --ppm Ba 11imore-Conventiona 1 House
MONITORING
PERIOD
SUMMARY
(* —KRKQ)
MAX
time or
MAX
MAX
TIKE
OF MAX
MAX
TIHK OF
MAX
MAX
DAT OF
Zone
1-HR *V.
1»MR
*¥.
3-HR AV.
3-
HR AV,
B-IIR AV.
8-IIK
AV.
24-HR AV.
24-HR ,
0*1
HR
oAr
PERI00
DAT PERIOO
Outdoor
1
lit,3
• 2976
16
461.8
82976
6
195.5
87976
1
376.4
82R7S
Kitchen
7
168S.3
62976
>6
1308.0
82976
5
1178,2
82976
2
8H2.1
B2976
living Room
3
1394.0
82976
16
1173.6
82976
6
963.8
B2976
2
R32.2
8297ft
Bedroom
4
$664.0
82976
is
1340.6
82976
S
1149.9
02476
2
862.6
82976
Indoor Average
S
1461.6
82976
IS
1236.5
82976
s
1097.3
82976
2
859.0
82976
-------
Indoor concentration peaks of CO tend to lag behind outdoor CO peaks.
Due to the CO emissions, this behavior may be shortened in houses with
indoor sources. The observed large fluctuations of the hourly CO concen-
trations display a local structure without a general pattern. However,
examination of the CO data base from several weekdays leads to identifica-
tion of a typical pattern with respect to 3-h averages. Typically, the
time periods 0800-1000 and 1900-2100 exhibit the highest observed CO levels.
These 3-h indoor peaks correspond to outdoor peaks caused by automobile
traffic during the typical urban rush hours {0600-0800 and 1700-1900).
The association of rush-hour traffic and typical indoor high level
periods reflect the time lag monitored earlier. Figure 4 illustrates
the indoor and outdoor variation of CO concentrations for a typical day,
in a dwelling with indoor CO sources. The indoor peak at hours 1400 to
1600 is not a typically observed elevation of the indoor concentrations.
TIME OF DAY, hours
Figure 4. Indoor and outdoor carbon monoxide (CO) variation at the Baltimore conventional residence.
-------
Seasonal variations in outdoor CO concentrations have not been
identified. A seasonal effect, however, has emerged from the investiga-
tion of indoor CO levels. The data base obtained from six dwellings,
monitored once during the summer and a second time during the winter,
show that CO indoor concentrations are generally higher in the winter
than in the summer. The observed higher winter CO levels may be attri-
buted to many factors such as tighter houses, regularly operating gas
furnaces, and increased indoor social activities.
Increases in outdoor CO concentrations do not lead to proportional
increases in the indoor CO concentrations. However, long-term averages
of CO in dwellings with no indoor CO sources are equivalent to the cor-
responding outdoor averages. This trend has been observed in the electric
houses with nonsmoking occupants which were sampled during the course of
this project. In houses of this type the indoor environment does not
shelter its occupants from high CO levels since the exposure to CO concen-
trations is equivalent to the corresponding indoor and outdoor exposure.
In houses with indoor CO sources, the observed indoor long-term averages
are higher than the ambient CO concentrations, leading to higher indoor
than outdoor exposures to CO levels and to increasing possibilities of
adverse health effects.
Observed measurements of CO concentrations both indoors and outdoors
are generally not considered high enough to cause a health hazard. The
highest hourly outdoor level observed during this program was 15.7 ppm;
whereas the maximum hourly Indoor level observed was 22.0 ppm. The
majority of the observed hourly indoor CO concentrations are between 2.3
and 6.0 ppm, while observed hourly ambient concentrations fall between
1.0 and 2.5 ppm. The hourly NAAQS of 35 ppm has not been violated by
either indoor or outdoor observed levels. The 8-h NAAOS is 9 ppm. This
standard has not been exceeded by the observed indoor or outdoor data;
however, 8-h average CO concentrations of 9.0 ppm have been sampled in
a few cases in both the outdoor and indoor environments. Finally, it
must be emphasized that observed indoor values are generally higher than
observed outdoor values, and that the ratio of corresponding ambient to
indoor values is less than one for at least 80% of the total hours
monitored.
Nitric Oxide (NO)
The complexity of the dynamics involved in the establishment of an
indoor-outdoor relationship is clearly illustrated in the interpretation
of the data base generated for NO. From the perspective of NO indoor
variation and under real-life conditions, three types of indoor environ-
ments have emerged: 1) houses with electric cooking and heating applian-
ces; 2) houses that are heated by gas furnaces, yet serviced by electric
cooking appliances; and 3) houses that are furnished with gas cooking
and heating equipment. In houses equipped with gas cooking appliances,
observed indoor NO levels are consistently higher than observed outdoor
levels. Houses with gas furnaces but electric cooking appliances
-23-
-------
d splay higher NO indoor levels than outdoor levels, most of the time.
However, there are time intervals interspersed throughout the monitoring
period during which the observed NO outdoor levels surpass corresponding
indoor levels. Indoor NO concentrations in totally electric homes are
almost always lower than corresponding outdoor concentrations.
Variation of the indoor concentrations of NO is associated with
emissions from gas stoves. Houses with a regularxcoolcing routine and
gas cooking appliances show a strong association between indoor peaks of
NO levels and cooking periods; on the other hand such association cannot
be made in houses with electric appliances. The observed indoor NO con-
centrations from totally gas houses are generally higher than the observed
indoor NO concentrations in other types of houses; this may be attributed
to emissions from gas appliances and possibly faulty gas furnaces. A
seasonal impact on the NO indoor levels has been clearly identified only
for totally gas houses. During the winter months the residential NO
concentrations are higher than the NO levels during the summer months.
This seasonal effect may be attributed to the same factors given earlier
for a similar behavior observed for the indoor CO concentrations.
The typical range of the observed hourly indoor NO concentrations is
0-300 ppb; the maximum observed indoor hourly NO concentration is 470 ppb.
The corresponding levels for the outdoor environment are 1-150 ppb and
3oO ppb, respectively. NAAQS for NO do not exist; an 8-h average resi-
dential NO standard of 2.5 ppm has been recommended by the American
Society of Heating, Refrigerating, and Air Conditioning Engineers (ASHRAE).
This indoor standard is not exceeded by the observed NO concentrations.
The maximum 8-h average registered by the field program is 285 ppb, well
below this standard. Generally, the observed indoor hourly NO concentra-
tions are higher than the corresponding ambient concentrations. The ratio
of outdoor to indoor values is less than one for about 50% of the total
monitored hours in fully electric houses; the same ratio is less than one
for at least 85% of the monitored hours in residences with gas appliances
and/or gas furnaces.
Nitrogen Dioxide (NOp)
The residential environment often provides a shelter from high out-
door N0p levels. The three classes of residences identified in the
interpretation of the NO data also manifest themselves in the study of
NOp. The data base collected for this project indicates that the hourly
average indoor concentrations of N02 are almost always lower than the
corresponding ambient levels in totally electric houses. Houses equipped
with gas furnaces and electric cooking appliances also shelter their
occupants, but to a lesser extent during peak ambient N02 levels.
Totally gas residences do not appear to provide such protection.
-24-
-------
In the indoor environment the N0„ half-life is estimated to be
approximately 30 min; this is a relatively short half-life when compared
w'th 2.0 h, the half-life estimated for CO which is considered inert.
Tl-us, the observed low indoor levels of N0? may be attributed to the
demical reactivity of this pollutant.
Two major elements seem to determine the indoor N0? levels:
V the variation of outdoor concentrations, and 2) the strength of the
irdoor sources. A typical pattern of the fluctuations in the ambient
N(L concentrations has been identified: a broad maximum appears daily
between the hours of 0700 and 1000. Due to the high indoor emission rates,
on 1y the total gas residences register higher indoor NOp levels than the
corresponding outdoor levels during these ambient peak periods.
The NAAQS for N02 is an annual arithmetic mean of 50 ppb. Typical
observed indoor levels fluctuate between 20 and 60 ppb. If the observed
irdoor concentrations were to maintain the same trend throughout the year,
it is likely that the NAAQS would not be exceeded. The highest indoor
hcurly average N02 concentration observed is 180 ppb, which does not
exceed an hourly NAAQS currently under consideration. However, 70% of
the hourly values obtained from the indoor environment of all gas houses
are higher than the corresponding outdoor values.
Sulfur Dioxide (SOg)
S02 concentrations sampled in the residential environment are very
lew. One or more of the following factors may be contributing to this:
• The observed ambient S02 levels, although higher than
the corresponding indoor concentrations, are low.
• SO2 is relatively reactive, and it is absorbed by
inaoor surfaces.
• The high C02 concentrations found in the indoor environ-
ment interfere negatively with S0? in the monitoring
instrument used.
The observed quenching effect of S0« by CO, in the field instrument is
discussed in Section 3. The levels registered Indoors are often below or are
very close to the detection limits of the instrument; it is therefore imprac-
tical to use the correction factor that has been formulated in order to adjust
for the experimental errors. Indoor SOp concentrations remain lower than
corresponding outdoor levels even after the correction factor is applied.
Thus, it is concluded that the indoor environment provides a shelter from
high outdoor S02 levels; this statement, however, must be constrained by
the facts that the majority of both ambient and indoor levels was low and
that the quenching of S02 concentrations was due to instrumentation diffi-
culties. No indoor or outdoor violations of the NAAQS SO2 standards (24 h:
0.14 ppm; 3-h: 0.5 ppm) have been observed. This can be attributed to the
observed low ambient S02 levels.
-25-
-------
Carbon Dioxide (CO^)
Observed hourly indoor concentrations of CCL are constantly higher
than corresponding levels in the ambient environment. The ambient/indoor
ratio is less than one for at least 90% of the total monitored hours.
The hourly indoor C02 concentrations vary between 150 and 2200 ppm; the
observed typical range for the ambient C02 levels is between 100 and
500 ppm. The ASHRAE-recommended indoor 8-h standard of 500 ppm is vio-
lated frequently, often by a factor of three.
Since CO2 concentrations indoors are a function of the number of
Inhabitants, the activity period with high indoor levels should correspond
to high activity patterns of the residents. Thus, houses with inhabitants
who work all day away from the residence should have low readings between
the hours of 0900 and 1800, whereas houses that are normally occupied
during this time period should display higher levels. Such behavior has
been observed. For the same reasons, the activity period with the highest
observed indoor COp concentrations is the time interval between 1800 and
2300 hours during which all members of a family are typically indoors.
0;one (0^)
Indoor 03 concentrations are lower than outdoor levels; the ratio
value of corresponding hourly ambient over indoor concentrations is
greater than one for 95% of the monitored hours. 0, is the surrogate
pollutant for photochemical smog. Ambient 03 levels are primarily of
automotive origin, but other sources include the combustion of fuels for
heat and electric power, the burning of refuse, the evaporation of petro-
leum products, and the handling and use of organic solvents. 03 is
highly reactive and decays rapidly by absorption on indoor surfaces.
The half-life of 03 is variable because it depends on the surface-
to- volume ratio and the material of the furnishings. A 2-min half-life
for 0, has been estimated from the existing data base. This value
indicates that 0, is the most reactive pollutant found in the indoor
environment. Owing to its high reactivity, indoor 03 levels are nor-
mally found at levels 50-70% of the corresponding outdoor concentrations.
Ambient hourly 03 concentrations have been observed at levels
higher than the NAAQS of 0.08 ppm. 03 is not generated indoors in
great quantities; however, a few indoor violations of the 03 hourly
NAAQS have been observed. The present data base does not conclusively
identify any indoor 03 sources; however, Hollowell et al. (1976) have
attributed small increases in 03 concentrations to the use of electric
stoves. Although electrostatic air precipitators have been identified
as possible ozone sources, none of the houses monitored in this program
was equipped with such a device. A general ambient daily pattern has
been identified: an early morning peak is followed by a second one
in the afternoon. Due to the low indoor 03 levels, there are no
identifiable patterns in the residential environment.
-26-
-------
Total Nonmethane Hydrocarbons (NMHC)
The ratio of ambient NMHC concentrations over corresponding indoor
concentrations is less than one for about 90% of the total monitored
hours; that is, the NMHC concentrations observed in the residential
environment are almost always higher than the ambient levels. Typical
indoor levels vary between 0 and 8.0 ppm while typical ambient concentra-
tions register a range between 0 and 3.5 ppm. The 3-h NAAQS for NMHC
(0600-0900) is 0.24 ppm and is often violated in the residential environ-
ment. ASHRAE suggests that an indoor residential 8-h standard be 0.10 ppm;
this recommended standard is exceeded indoors with the same high frequency
that the NAAQS 3-h standard is violated.
Fluctuations in the NMHC indoor concentrations may be associated
with cooking, cleaning, and other activities. A general pattern of
variation has not been identified; however, the 2-h periods between
hours 1200-1400 and 1900-2100 display consistently higher levels in
all monitored dwellings. These periods may be associated with cooking
and general indoor activity by the occupants.
Total Suspended Particulate (TSP) Matter
In the residential environment TSP is collected on 47-irni glass fiber
filter material for a period of 24 h at a sampling rate of 3-4 ft /min.
In the outdoor environment TSP is collected simultaneously by two different
methods: on 47-mm glass fiber filter material as in the indoor environment,
and on an 8 x 10 in. glass fiber filter using a conventional high volume
air sampler. The outdoor high volume TSP measurements are for comparison
with the daily 47-mn (low volume) samplers. Examination of the correspond-
ing daily TSP value from the low and high volume samplers shows that they
correlate well with the average correlation coefficient for all residences
being 0.86. The ratio of corresponding low volume over high volume daily
TSP concentrations varies in the neighborhood of 0.85, with extreme ratio
values of 0.68 and 0.96. The data base obtained from the low volume air
sampler 1s used in subsequent data interpretation.
The steady-state TSP model, developed on the data base generated by
the project and discussed in detail in Section 3 on Numerical Models,
will show the Importance of the family activity as an indoor source of
TSP matter. Moreover, the data base illustrates that there are no con-
stant ratio values that relate the observed outdoor levels to Indoor
concentrations. The ratio of observed daily Indoor TSP concentrations
over corresponding outdoor TSP concentrations varies from 0.3 to 3.6.
3
The TSP 24-h maximum NAAQS is 260 ug/m . This value must not be
exceeded more than once a year. The typical range of observed Indoor
residential TSP concentrations is between 30 and 100 yg/nr, with
-27-
-------
3
a maximum observed level of 500 ug/m . A number of residences, each
monitored for a 14-d period, show indoor TSP concentrations that exceed
the NAAQS. In addition, the indoor-outdoor data base shows that the
NAAQS for particulate matter is exceeded in the indoor residential environ-
ment when the corresponding ambient levels are below the standard.
The effect of indoor TSP sources needs further investigation; it is
apparent that residential control measures are required in order to protect
the indoor environment from levels that are considered hazardous to the
public health in the ambient environment. Further examination of the TSP
indoor-outdoor relationships appears in Section 3 of this document, while
the TSP control measures are discussed in Section 5.
Respiratle Suspended Particulate (RSP) Matter
Twc samples of RSP matter are obtained daily. One sample is from an
indoor location, often the living room, and the other from an outdoor loca-
tion. The two corresponding RSP concentrations are 24-h average samples
collected with the Environmental Research Corporation (ERC) Model #200
Dichotomous Sampler. Particulates with a diameter of 3.5 um or less are
classified as RSP matter.
An effort to formulate a physical model relating the indoor RSP matter
concentrations to outdoor levels and to indoor sources was not successful.
The available data do not display correlations among the various parameters;
the correlation coefficient between indoor and outdoor levels is small,
there is no association between the activity index and the indoor RSP levels,
and the impact of varying air exchange rates on the indoor RSP concentra-
tions is not significant. Figure 5 shows typical RSP indoor-outdoor levels
and illustrates the lack of relationships. Many difficulties are involved
in the development of a theoretical relationship between indoor and outdoor
RSP concentrations: the deposition rate of respirable suspended particulates
is very low and very difficult to quantify, the efficiency of residential
filtering devices to remove RSP has not been well established, and the
generation rate of RSP by indoor activities is not known. Finally, the
data set on RSP indicates that while high activity may lead to high indoor
RSP levels, low activity does not necessarily imply low indoor RSP concen-
trations.
Ambient respirable suspended particulate matter concentrations varied
between 1 and 91 pg/m3. In the indoor location the concentrations ranged
from 1 yg/m3 to 260 ug/m3. Table 12 summarizes the RSP data obtained from
a majority of the residences monitored in the indoor-outdoor project. Each
residence is monitored for approximately 2 weeks. The ratio of the arith-
metic mean of a 2-week period of indoor daily levels over the correspond-
ing arithmetic mean of outdoor levels separates the residences into two
-28-
-------
40
38
36
34
32
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
OUTDOOR
INDOOR
I \\
I \
/ \
\
\ ~
l/s 1/6 1/7 1/8 1^5 l/7o 1/11 1 ni 1/13 l/l4 i/l's l/ifi l/if r/l8
TIME, month and day
Figure 5. Typical RSP indoor-outdoor level*.
-29-
-------
well-defined groups. One group of residences displays higher indoor levels
than outdoor, and the second group includes residences with indoor RSP levels
lower than outdoor concentrations. The demarcation line is very clear; the
[RSP]. /[RSP] t ratio varies from 2.0 to 3.5 for the group with higher
indoor levels, and from 0.4 to 1.0 for the residences with lower indoor RSP
levels. The houses in each group have two factors in common, and these may
be the cause of the observed difference in the ratios. All families in the
group with higher indoor RSP levels have children who are of preschool age
or in the first or second grades; none of the families who are in the
second group have young children. This activity association indicates that
young children may keep the matter airborne, while houses without children
are likely to have long periods of low activity, thus allowing RSP matter
to settle. The smoking-no smoking factor clearly influences the indoor RSP
matter levels. All but one of the families in the first group smoke more
than 10 cigarettes a day on a regular basis. The families in the second
group inc.ude no heavy smokers: four families have no smoking members,
and the remaining three families have members who smoke less than 10 ciga-
rettes a day, often less than four.
TABLE 12. RESPIRABLE SUSPENDED PARTICULATE (RSP) MATTER
INDOOR-OUTDOOR DATA SUMMARY
Statistics - Values /ig/m3 Children
—— —¦—"—" 7 Years
Residence
R?nge
Out
In
In/
Old or
Identification
Out
In
[RSP
J s. d.
[RSP
] s. d.
Out
Less
Smoking
Baltimore Conventional I
9-54
5-84
23
12
47
17
2,0
Yes
Yes
Denver Conventional
1-40
30-94
23
10
64
20
2. 8
Yes
Yes
Pittsburgh Low Rise 1
7-34
26-260
18
9
50
65
2. 8
Yes
No
Pittsburgh Low Rise 2
8-34
27-83
18
9
50
17
2. 8
Yes
Yes
Pittsburgh Mobile Home 2
16-42
35-210
23
7
82
44
3. 5
Yes
Yes
Baltimore Conventional II
7-S4
10-86
10
14
46
25
2. 5
Yes
Yes
Chicago Experimental I
11-34
6-14
22
6
10
2
0. 5
No
<10
Baltimore Experimental II
11-35
5-40
21
8
14
10
0. 7
No
No
Pittsburgh Mobile Home 1
37-91
5-60
58
22
21
16
0.4
No
No
Chicago Experimental II
1-35
1-29
19
9
8
7
0.4
No
<10
Pittsburgh High Rise 1
18-62
12-41
36
16
23
9
0. 6
No
No
Baltimore Experimental I
11-66
5-56
30
18
18
17
0. 6
No
No
Pittsburgh High Rise 2
12-49
1-34
28
14
21
10
0. 8
No
Nc
Pittsburgh High Rise 3
12-76
21-80
36
18
35
16
1. 0
No
<10
-30-
-------
The above analysis indicates that the combination of young children
and regularly smoking adults constitutes an indoor source of RSP matter.
Data from houses without young children and regularly smoking members do
not display characteristics attributable to indoor RSP sources. This
association can be used to estimate indoor RSP levels when the outdoor
concentrations are available. Thus, if members of a house include regular
smokers and have young children, one can assume that the ratio of indoor
over outdoor RSP matter is approximately 2.5; if there are no regular
smokers or young children, the ratio may be approximated to 0.8. The
demarcation line, obtained from the present set of data, is very defini-
tive; however, more data will be very useful in verifying the relation-
ship outlined and validating the suggested predictive claims.
Few RSP concentrations measured in the indoor environment exceed the
NAAQS 24-h maximum particulate matter standard. A few other indoor daily
averages come close to the 24-h maximum particulate matter concentration
hut the corresponding outdoor levels are considerably lower. New studies
are required to determine if the observed indoor RSP levels precipitate
adverse health effects.
Mater Soluble Nitrates (NQ^)
Particulate matter collected on the 47 mm glass fiber filter is
analyzed by the filtratlon/brucine laboratory analytical method for water
soluble nitrates. Daily concentrations are obtained from the outdoor
location and the three Indoor sampling sites. The Indoor residential
levels of nitrates are quite low, and they are mainly driven by the outdoor
N0§ concentrations. Figure 6 illustrates a typical pattern of the observed
nitrate levels. Ninety percent of the observed daily indoor nitrate averages
are lower than the corresponding outdoor levels. For over 60% of the total
days sampled (over 250 d), the ratio value of corresponding daily indoor NOr
levels over the ambient nitrate concentrations is less than 0.3.
The observed daily indoor concentrations of nitrates do not vary sig-
nificantly and display a small range between 1.0 and 6.0 ug/m , with
typical values at the lower end of this concentration interval. The maximum
observed daily outdoor concentration is 16 ug/m , and typical values are
in the neighborhood of 6.0 ug/m . There are no NAAQS for nitrates; there-
fore, it 1s not possible to determine whether the monitored residences
register elevated levels of nitrate. It 1s apparent, however, that the
residential environment shelters its occupants from the nitrate levels
observed outdoors.
Water Soluble Sulfate - (SOa)
Indoor and outdoor sulfate concentrations are obtained from the daily
TSP samples. Filtration/methyl-thymol blue is the analytical laboratory
-------
9
8
7
6
5
4
3
2
I
0
mmI — rn Outdoor
O-"—o Kitchen
O—-D Bedroom
A—A Living Room
~
?
I!
A
- — - ''''I T I i
1 T/3 4/44/5 476 4/74/84/9 4/10 4/11 4/12 4/13 4/14 4/15 4/16
TIME, month and day
Figure 6. Typical patterns of observed nitrate levels.
-32-
-------
method used. Ambient daily sulfate concentrations vary between 1.5 and
48.3 ug/m3; the average indoor sulfate concentrations ranged between 1.0
and 41 ug/m3. Table 13, the SO4 data summary, illustrates the general
trends of the collected sulfate data base. Both the indoor and outdoor
concentrations vary considerably within each monitoring period. Variations
in air exchange rates, indoor temperature, indoor humidity, occupant activ-
ity and combinations of these factors do not explain the observed sulfate
fluctuations. The TSP removal and reentrainment processes, and the involved
chemical conversion mechanisms of sulfur dioxide to sulfates are among the
difficulties involved in the development of analytical expressions relating
indoor and outdoor sulfate concentrations. It has been concluded that the
formulation of a dynamic model for sulfates must await improvement of the
state-of-the-art of various closely related scientific questions. The
da:a base generated by the indoor-outdoor project substantially adds to
th¦? existing relevant literature on sulfate levels.
Statistical analyses indicate that daily sulfate concentrations are
ap )roximately equal in the three indoor sites. Figure 7 is a typical
eximple of the variation of sulfate levels during a 2-week monitoring
period; this graphic representation illustrates the similarity of the con-
centrations at the three indoor sites. Given the uniformity of the sulfate
concentrations in the indoor environment, the data analysis is undertaken
on the basis of ambient and average indoor concentrations.
The signature parameter used for sulfate data interpretation 1s the
ra<:io of the indoor daily average concentrations over the corresponding
outdoor concentrations. Two categories of houses have been identified with
respect to this parameter: the average value of the indoor/outdoor ratio
for houses in the first category is 0.46, with 7055 of the daily ratio
values below a demarcation ratio value of 0.6; the average indoor/outdoor
ratio value for residences in the second category is 0.81, with 82% of
the daily ratio values above the demarcation ratio value of 0.6. In
residences that comprise the first category, electricity or propane is
used for cooking and electricity or kerosene is used for heating. The
fuel used in residences in the second category is gas.
The indoor-outdoor data base for sulfates shows that the indoor 24-h
sulfate concentration is almost always lower than the corresponding outdoor
sulfate concentration. The type of fuel used for cooking and heating 1s an
Important factor In determining the Indoor-outdoor relationship; houses
with gas appliances have a slightly higher sulfate Indoor/outdoor ratio
than houses without gas appliances. Sulfur is added to residential gas
appliances for detecting leaks of the otherwise odorless fuel. Hollowell
et al. (1976) have determined that approximately 10% of the sulfur added
to the residential gas manifests itself as indoor sulfates.
-33-
-------
TABLE 13. SULFATE (S04) DATA SUMMARY, pg/m3
Statistics
Indoor
Outdoor
Correlation Coefficient
Between
Indoor -O utdoor
Daily Concentrations
rvange
Standard
Deviation
Standard
Deviation
Residence
Fuel Type
Indoor
Outdoor
Average
Average
Pitt. Mobile 1
Kerosene
2. 6-10.6
4. 2-32. 8
5. 55
2. 42
18. 71
7. 86
0. 937
Pitt. Mobile 2
Kerosene
2. 9-8. 3
5. 8-18. 3
4. 87
1. 50
8. 99
3. 49
0. 024
Bait. Exp. I
Electric
2.6-27.4
3. 2-37. 4
11. 35
8. 54
15. 42
12. 35
0. 981
Chic. Exp. I
Electric
1. 1 -4. 9
3. 3-10. 4
2. 53
1. 30
5. 43
1. 84
0. 689
Chic. Exp. 11
Electric
0. 3-5. 3
0. 6-14. 5
2. 12
1. 55
8. 15
4. 44
0. 767
Bait. Exp. II
Electric
1. 5-5. 5
2. 6-10. 1
3. 25
1. 21
4.94
2. 34
0. 333
Wash. Exp. I
Gas
2. 6-22. 0
0. 9-30. 3
9. 09
5. 49
11. 36
9. 23
0. 927
Wash. Conv. I
Gas
2. 0-5. 7
0.9-21. 5
3. 67
1. 09
12. 02
5. 44
0. 678
Chic. Conv. I
Gas
1. 5-22. 7
1. 8-21. 9
8. 18
6. 36
7. 78
6. 14
0. 998
Den. Conv.
Gas
0. 9-4. 4
0. 8-4. 4
2. 20
0.95
2. 20
1. 20
0. 289
Bait. Conv. I
Gas
2. 4-24. 8
2.3-21. 5
9. 50
6. 60
10. 46
6. 44
0. 899
Bait. Conv. II
Gas
1. 7-5. 2
1.8-8.8
3.16
1.06
4. 27
2. 17
0. 789
Pitt. Low-Rise #1
Electric/Gas
3. 1-9. 5
3.3-16.4
4. 47
1. 72
8. 22
3. 95
0. 668
Pitt. Low-Rise #2
Electric/Gas
2. 2-6. 6
3. 8-12. 5
4.04
1. 38
6. 83
2. 17
0. 749
Pitt. Low-Rise #3
Electric/Gas
8. 4-20.1
10. 8-24. 1
12. 99
3. 37
15. 14
3. 77
0. 494
Pitt- High-Rise #1
Electric/Gas
2. 8-18. 5
5. 1-19. 9
8. 05
4. 36
10.99
4. 67
0. 741
Pitt. High-Rise #2
Electric/Gas
3. 1-17. 3
4.1-25. 5
8. 74
4. 76
11. 78
6. 78
0. 967
Pitt. High-Rise #3
Electric/Gas
4. 2-48. 3
6. 0-41. 1
14. 89
11. 81
16. 28
9. 65
0. 983
-------
TIME, month and day
Figure 7. Pittsburgh low-rise #1 SO^ •
-------
Indoor air quality standards for sulfate concentrations are not avail-
ab e. However, adverse health effects have been associated with ambient
levels of sulfates in urban areas. A summary of the ambient sulfate levels
by Altshuller (1973), gives the 5-y average concentrations for eastern
and western urban centers: the 5-y average of SO4 in western urban centers
is 6.4 ug/m3. and the corresponding 5-y average for eastern urban areas
is 13.5 vg/m^/ The arithmetic annual average ambient concentration range
is 2.4-48.7 yg/m3. A daily threshold value of 8-10 yg/rrr is suggested
by Colucci (1976) who claims that higher indoor sulfate concentrations
may lead to adverse health effects in the elderly. If the above levels
were to be considered as reference points, the observed indoor sulfate
concentrations are often within, and a number of times higher than, this
range. Thus, while the indoor daily averages are almost always lower
than the corresponding outdoor levels, the observed indoor sulfate concen-
trations are at levels equivalent to those that have been linked with health
problems.
/ nmonia (NH^)
Ammonia (NH3) experiments were among the few tests staged during this
program. Ammonia studies were performed in each residence monitored. The
process used to introduce ammonia into the indoor environment was mopping
the kitchen floor with a solution of 50 ml of a commercially available
cleaning agent, 100 ml ammonia, and 2 gallons of water.* Three 3-h experi-
ments were conducted in each residence. In some residences all experiments
were conducted on the same day as one continuous test, while in other resi-
dences the experiments were staggered over a 2- or 3-d period. Measurements
were taken in the kitchen, bedroom, and living room as soon as the kitchen
cleaning activities began.
The collected samples were analyzed using a colorimetric method. Since
the ammonia data base is not included in the magnetic tapes, Table 14
provides all information gathered from the staged experiments. As expected,
the highest ammonia concentrations occur in the kitchen the first hour
following introduction of the pollutant. During the second and third hours,
the pollutant is dispersed throughout the residence, and, often by the
third hour, ammonia concentrations become low and approximately equal in
each location.
The threshold limit value of ammonia for an 8-h day, 5 d per week is
25 ppm (ACGIH, 1976); the ASHRAE-recommended indoor standard is 2.5 ppm.
None of the ammonia levels that were observed indoors during the routine
floor washing procedure exceeded this level; the highest observed concen-
tration was 1.2 ppm.
* Before July 1, 1976 only 50 ml were used.
-------
TABT.E 14. AMMONIA LEVELS, ppm
Ran 1
Run 2
Run 3
Range*
Date
Date
Date
and
Bed-
living
and
Bed-
living
and
Bed-
Living
Bed-
living
Location
Time
Kitchen
room
Room
Tim*
Kitchen
room
Room
Time
Kitchen
room
Room
Kitchen
Room
Room
2/17
2/17
2/18
Pittsburgh
1020-1130
0.033
<0.033
0.11
1340-1440
0,012
<0.003
<0,003
1025-1125
0,025
<0.003
0,011
0.003-
0.008-
0.008-
Mobile I
1131-1231
0.084
0,003
<0.003
1441-1541
<0,003
<0.003
<0.003
1125-1226
0.025
<0.003
0.014
0.084
0.008
0,014
- 1232-1332
<0.003
<0,003
<0.003
1542-1642
0,025
<0,003
<30.003
1227-1327
0,006
<0.003
0.009
3/8
3/10
3/10
Pltttborgh
1000-1100
0.056
0.027
<0.001
1200-1300
0.0S5
0.017
0.011
1550-1630
0,017
<0.001
<0.001
0.006-
0.001-
0,001-
Mobil, n
1101-1201
0.237
0.141
0,073
1308-1408
0.019
0,001
<0.001
1651-1751
0.006
0.006
0.003
0.237
0.141
0.073
1202-1302
0.019
<0.001
<0,001
1411-1411
0.024
<0.001
*
1752-1852
0.024
<0,001
<0.001
5/16
5/16
Pittsburgh
1640-1740
0. 776
0.344
0.821
1956-2056
0.274
0.230
0,383
High RIm 1
1741-1841
0.265
0.263
0,307
2057-2157
0.692
0,321
0,476
1842-1942
0,482
0.3S4
0,406
21S8-2258
0,570
0,377
0,342
S/31
5/31
5/31
Pittlburgh
1030-1130
*
0.767
0,332
1350-1450
0.403
0.153
0.204
1700-1800
0,148
0.195
0.165
High RJf* 2
1131-1231
*
0.351
0,333
1451-1551
0,136
*
0.125
1801*1901
0,212
0,143
0.119
1232-1332
0.198
0,166
0,163
1552-1652
0,252
0,433
0.287
1902-2002
0,342
0,458
*
Pttoburgh
High RIm 3
6/ IS
1055-115S 0.154
1156*1256 0.053
1257-1357 0.308
0.049 0.051
0.052 0.050
0.147 0.178
6/15
1405-1505
1506-1606
1607-1707
0.339
0.367
0.252
0.227 0.265
0.320 0.369
0.242 0.127
6/16
1030-1130
1131-1231
1232-1332
0.289
0.239
0,448
0.144
0.273
0.177
0.201
0,239
0.171
0,053-
0.448
0.049-
0.320
0,050-
0.369
Pttoburgh
Low RlM 1
3/24
0822-0922
0923-1023
1024-1124
0.207
0.136
0.092
0.065 0.268
0.052 0.080
0.027 0.023
3/24
1132-1232
1233-1333
1334-1434
1.230
0.205
0.141
0.274 0.067
0.067 0.007
0.046 0.040
3/24
1446-1546
1547-1647
1648-1748
0.742 0.082 0.175 0.091- 0.027- 0.023-
0.185 0.077 0*070 1.230 0.274 0.268
1.50 0.52 0.057
4/5
4/5
4/6
PlmUuigh
1100-1200
0.626
0.539
1.04
1445-1545
0,299
0.087
0.291
1200-1300
0,691
0,241
0,348
0.299-
0,087-
0.249-
taw RIm 2
1200-1300
0.324
0.280
0,663
1545-1645
0.464
0,271
0,649
1300-1400
0,772
0,945
1.00
1,10
0,945
1.041
1300-1400
1,04
0.663
0,249
1645-1745
0,659
0,405
0,463
1400-1500
1.10
0.467
0.404
7/7
7/8
7/8
Wajfclngton
1330-1430
0.184
0.025
0,050
1200-1300
0,118
0.023
0.039
1507-1607
0,123
0,027
0.070
0.007-
0,007-
0.007-
Conwdosal 1
1430-1300
0.056
0.008
0,036
1302-1402
0.029
0.011
0.016
1609-1709
0,016
<0,007
<0.007
0,184
0.027
0.070
1530-1630
0,029
<0.007
<0,007
1405-1505
<0,007
<0,007
<0.007
1707-1810
0,053
<0,011
<0.010
8/4
8/4
8/4
Wuhington
0945-1045
0.233
0.033
0.064
1818-1418
0,984
0.0
0.0
1636-173*6
1,18
0,012
0.012
0.020-
0,0-
0,0-
Experimental 1
1052-1152
0,021
0,0
0.0
1420-1418
0,609
0.072
0.020
1741-1841
0.504
0.012
0.053
0,070
0.070
0,14
1154-1254
0,020
0.012
0.0
1525-1625
0.230
0.0
0.012
1849-1949
0,233
0,012
0,064
2/2
2/2
2/2
Baltimore
1148-1248
0.978
0,099
0.377
1451-1551
0.602
0,102
0,186
1758-1858
0.300
0,086
0,180
0.110-
0,041-
0.113-
Conventional 2
1248-1348
1,207
0.149
0,123
1554-1654
0,176
0,041
0.128
1901-2001
0.179
0.079
0.155
0,041
0,149
0,377
1351-1451
0.159
0.067
0.118
16S6-1756
1.171
0.125
0.151
2004-2104
0.110
0,075
0,113
9/8
9/8
9/8
Bel* mote
0985-1055
0.316
0.229
0.153
1350-1450
0.074
0,155
<0.005
1702-1802
0.094
0,066
<0.006
0.034*
0.0S6-
<0.005-
Coawtttonal 1
1057-1157
0.176
0,239
<0,005
1452-1552
0.084
0.066
<0,005
1804-1909
0.135
a 1X7
0,043
0.316
0,239
0.153
1159-1259
0.174
0,155
<0.005
1554-1664
<0.005
0,127
0.014
1906-2006
0,081
0.086
A0S5
1/8
1/6
1/6
Baltlmofe
1020-1120
0.285
0,086
0.161
1331-1431
0.301
0.177
0.221
1645-1745
0.286
&2S3
0.210
0.188-
0.086-
0.151-
Experimental 2
1121-1221
0.223
0,211
0.165
1484-1534
0.205
0,161
0.163
1748-1848
0.262
a 170
0.194
0,301
0.233
0.221
1222*1328
0.201
0,174
0.151
1536-1636
0.205
0.165
0.202
1850-1950
0.188
a 155
0.188
* No wimple Uft la Ami to i
(QQBttaMd)
-37-
-------
TABLE 14. (continued)
Rtm 1
Run 2
Run 3
Range*
Date
Date
Date
and
Bed-
Living
and
Bed-
Living
and
Bed-
Living
Bed-
Living
location
Time
Kitchen
room
¦Room
Time
Kitchen
room
Room
Time
Kitchen
room
Room
Kitchen
room
Room
a/23
8/23
8/25
Baltimore
1155-1256
0.365
0.260
0.232
1525-1625
0.334
0.270
0.252
0950-1050
0.270
0.239
0.273
0,165-
0.175-
0.132
Experimental 1
1257-1357
0.270
0.378
0.273
1628-1728
0.352
0.388
0.351
1052-1152
0,410
0,400
0,416
0.410
0.400
0,416
1359-1459
0.365
0.282
0.306
1732-1832
0.313
0.313
0.306
1154-1254
0.165
0.175
0,132
7/25
7/26
7/26
Chicago
1440-1540
Q.216
0,070
0,580
0955-1055
0,001
0.162
0.001
1308-1408
0,475
0,001
0, 109
0,001-
0.001-
0,001-
Experimental I
1541-1641
0.251
0.288
0.745
1056-1156
0.874
1.028
0.666
1411-1511
0.330
0.217
0.584
0.874
1.028
0.745
1642-1742
0,267
0.152
0,284
1157-1257
0.3S7
0.233
0.172
1S13-1613
0.137
0.245
0.326
12/16
12/16
12/17
Chicago
0925-1025
0.005
0.0
0.0
1241-1341
0.009
0.0
0.0
0915-1015
0,080
0.0
0.005
Experimental 2
1027-1127
0.0
0.0
0,0
1346-1446
0.0
0.0
0.0
1025-1125
0.028
0.0
0.001
1132-1232
0,0
0.0
0.0
1450-1550
0.0
0.0
0.0
1129-1229
0,001
0.005
0.001
8/3
8/3
8/3
Chicago
1055-1155
0.674
0,613
0. 902
1420-1520
1,017
0. 593
0.466
1815-1915
0.854
0,545
0. 520
0,342-
0,387
0,436-
Conventional 2
1156-1256
0.825
0.677
1.303
1521-1621
0.522
0.395
0.436
1916-2020
0.342
0.427
0.472
1.017
0.677
1.303
1257-1357
0.458
0.703
0.561
1622-*1722
0.580
0,525
0.463
2021-2124
0,525
0.387
0,503
11/30
11/30
11/30
Chicago
1250-1350
0.293
0* 123
0.323
1557-1608
0,031
0.018
0.075
1914-2014
0.091
0,008
0.020
0.022-
0,0-
0.017-
Conventional 1
1351-1451
0.140
0.036
0.352
1659-1759
0.070
0.0
0.016
2016-2116
0.022
0.0
0.074
0.293
0.123
0.357
1453-1503
0.087
0.081
0.030
1800-1900
0.034
0,0
0.023
2117-2217
0.039
0.012
0,012
10/9
10/9
10/10
Denver
1005-1105
0.609
0.108
0,172
1315-1415
0.791
0,008
0.161
1245-1345
3.085
0.069
0.191
0.006-
0.006-
0.112-
Conventional
1107-1207
0.394
0.322
0.156
1.18-1518
0.374
0.008
0.173
1349-1449
0.059
0,008
0.21S
3.085
0.322
0. 246
1210-1310
0.368
0.015
0,240
1>22-1622
0.238
0.006
0,112
1452*1552
<0.006
0.042
0.198
-38-
-------
The USSR air quality criteria state that human odor perception of ammo-
nia occurs at 0.56 ppm. Levels that exceed this value have been observed in
residences in the course of this project. Russian scientists have also found
adverse responses of human reflexes when ammonia levels surpass 0.49 ppm
(Stern, 1968).
Aldehydes (ALP)
Aldehyde concentrations were measured intermittently. In the outdoor
location one 24-h average is obtained daily. In each of the three indoor
locations, three 4-h averages were measured each day: the first 4-h period
began at 0600 hours, the second at 1000 hours, and the last at 1600 hours.
This design takes under consideration the expected low ambient levels;
it generates a daily structure and allows for identification of any indoor
patterns.
The observed outdoor concentrations of aldehydes were always lower
than the indoor levels, typically by a factor of six and quite often by
one order of magnitude. Figure 8 is an illustration of the data collected.
The observed outdoor concentrations of aldehydes were negligible.
Early in the study of residential aldehyde levels, it was discovered
that each dwelling has its own very distinct character and that directly
studying the raw data provides the best means of interpreting the observa-
tions. Table 15 shows a portion of the aldehyde data collected; Table 16
is an example of the basic descriptive statistic obtained for this set of
data. Similar tables were generated for each of the monitored residences.
The data from this study conclusively shows that there are indoor
sources of aldehydes. Emanation of aldehydes from chipboard used in
the construction of buildings, from pressed board used in mobile homes
and furniture may account for the observed high indoor levels.
Daily patterns of aldehyde levels have not been identified; the indoor
levels are generally uniform within each day, but they display considerable
fluctuations over each monitoring period. Andersen et al. (1975) have
formulated a mathematical model that estimates the room air concentration
of formaldehyde; this model has been established with the help of chamber
experiments. The Indoor air pollution study cannot estimate board surface
area which has been shown by Andersen to be a crucial parameter Involved
1n estimating the aldehyde concentrations. The temperature and humidity
relationships of the Andersen model have been qualitatively verified.
Although the source strengths were not quantified by this study, a
series of conclusions were reached.
-39-
-------
225
200
175
150
125
100
75
50
25
0
CHICAGO EXPERIMENTAL RESIDENCE
ALD
• OUTDOORS
O O KITCHEN
~ ~ BEDROOM
A A LIVING ROOM
X
X
X
I
I
X
x
X
X
X
X
X
-I
/5 12/6 12/7 12/8 12/9
12/10 12/11 12/12 12/13 12/14 12/15 12/16
TIME, month and day
12/17 12/18 12/19 12/20
Figure 3- Observed Indoor and outdoor aldehyde time {day) variations-
-------
TABLE 15. ALDEHYDE CONCENTRATION IN A SAMPLE RESIDENCE
Site #: 6
Visit #: 2
Pollutant: Aldehydes^
Pollutant Units: yg/m
Pollutant concentration values for three 4-h periods beginning
at 0600, 1000, and 1600 plus the daily average for each indoor zone
Date
Kitchen
Bedroom
Livingroom
Daily
Average
0600-1000
1000-1400
1600-2000
0600-1000
1000-1400
1600-2000
0600-1000
1000-1400
1600-2000
77/ 7/15
306.40
298.00
313.80
277.30
219.70
224.60
290.40
255.60
252.40
270.91
77/ 7/14
-1.00*
303.10
-1.00
-1.00
243.90
-1.00
-1.00
215.60
-1.00
254.20
77/ 7/15
-1.00
207.80
555.40
-1.00
277.30
246.90
-1.00
221.70
277.00
297.68
77/ 7/16
302.80
350.70
310.30
262.50
285.00
229.40
224.70
295.20
216.00
275.18
77/ 7/17
377.80
336.10
328.10
252.40
279.70
244.80
226.40
225.10
288.00
284. 27
77/ 7/18
409.80
445.20
499.70
309.60
336.20
424.30
371.20
401.60
415.80
401.49
77/ 7/19
-1.00
-1.00
314.70
-1.00
-1.00
320.70
-1.00
-1.00
554.60
396.67
77/ 7/20
450.90
422.10
415.10
400.50
321.70
328.00
412.10
374.60
308.80
381.53
77/ 7/21
369.40
396.60
343.50
352.20
326.30
330.00
317.00
312.80
269.70
335.28
77/ 7/22
216.20
391.10
313.60
393.90
326.60
429.30
362.60
347.20
303.00
342.61
77/ 7/23
390.70
386.40
420.30
316.70
313.00
295.20
305.70
295.20
336.90
340.01
77/ 7/24
427.40
447.40
386.40
332.20
328.80
322.70
315.60
346.70
317.40
358.29
77/ 7/25
339.10
320.50
-1.00
286.80
287.80
-1.00
261.40
241.80
-1.00
289.57
77/ 7/26
-1.00
-1.00
353.00
-1.00
-1.00
257.00
-1.00
-1.00
242.20
284.07
77/ 7/27
346.90
395.20
-1.00
313.10
289.50
-1.00
302.50
277.90
-1.00
320.85
*
The designation (-1.0) indicates missing data.
-------
TABLE 16. DESCRIPTIVE STATISTICS CORRESPONDING TO THE
ALDEHYDE CONCENTRATION OF TABLE 15
Site #: 6
Visit #: 2
Po'lnf-int: AlUtliycle^
Pollutant Units: g/m
Zone-period statistics
Kitchen Bedroom Liyingroom
0600-1000
1000-1400
1600-2000
Total
0600-1000 1000-1400
1600-2000
Total
0600-1000
1000-1400
1600-2000
Total
Mean
Sigma
357.945
66.347
361.554
67.964
379.492
80.145
366.431
70.357
317.927 295.039
49.039 35.196
304.408
69.395
305.156
52. 103
308.145
58.365
293.154
60.970
315.150
91.195
305.067
70.371
Zone-day statistics
Kitchen
Bedroom
Liyingroom
Daily
Date
Mean
Sigma
Mean Sigma
Mean
Sigma
Mean
Sigma
77/ 7/13
306.067
7.905
240.533
31.935
266.133
21.076
270.911
34.637
77/ 7/14
303.100
-1.000*
243.900
-1.000
215.600
-1.000
254. 200
44. 650
77/ 7/15
381.600
245.790
262.100
21.496
249.350
39. 103
297.683
129.377
77/ 7/16
321.267
25.764
258.967
27.968
245.300
43.433
275. 178
45.420
77/ 7/17
347.333
26.687
258.967
18.353
246.500
35.946
284. 267
53.401
77/ 7/18
451. 567
45.287
356.700
60.035
396.200
22. 785
401.489
56.980
77/ 7/19
314.700
-1.000
320.700
-1.000
554.600
-1.000
396.667
136. 807
77/ 7/20
429.367
18.974
350.067
43.790
365.167
52. 292
381.533
50. 821
77/ 7/21
369.833
26.553
336.167
14.008
299.833
26.181
335. 278
36.275
77/ 7/22
306.967
87.638
383.267
52.169
337.600
30.938
342.611
62.813
77/ 7/23
399.133
18.457
308.300
11.495
312.600
21.689
340.011
46. 962
77/ 7/24
420.400
31.097
327.900
4.814
326.567
17.459
358.289
49. 940
77/ 7/25
329.800
13.151
287.300
0.718
251.600
13.859
289.567
36. 045
77/ 7/26
353.000
-1.000
257.000
-1.000
242.200
-1.000
284.067
60. 155
77/ 7/27
371.050
34.153
301.300
16.687
290.200
17. 394
320. 850
43. 430
Monitoring period statistics
Mean Sigma
325.551 70.502
*
The designation (-1.0) indicates missing data.
-------
Table 17 shows two distinct classes of monitored residences one with
high indoor levels and the second with lower levels.
TABLE 17. STATISTICAL SUMMARY OF OBSERVED ALDEHYDE LEVELS
(Outdoor levels are very low)
14-d Monitoring Period
Residence
Observed Range
4-h Concentrations
(Jjg/m3)
Mean Concentrations
(yg/m3)
Standard Deviation
(yg/m3)
Denver Conventional
87-615
250
118
Chicago Experimental I
140-300
200
38
Chicago Experimental II
242-555
32S
70
Pittsburgh Mobile Home 1
200-938
470
167
Pittsburgh Mobile Home 2
136-934
387
159
Washington Conventional I
21-153
52
31
Baltimore Conventional II
34-150
75
25
Washington Experimental I
10-286
90
78
Baltimore Experimental I
17-162
78
38
Baltimore Experimental II
6-122
48
20
Pittsburgh Low Rise 1
51-152
91
34
Pittsburgh High Rise I
22-120
56
18
Chicago Conventional I
20-190
54
29
Chicago Conventional Ii
10-159
47
23
Pittsburgh Low Rise 2
35-149
78
29
Baltimore Conventional I
10-300
144
75
Pittsburgh High Rise 2
76-240
125
27
Pittsburgh High Rise 3
65-234
149
40
Pittsburgh Low Rise 2
20-102
110
32
-43-
-------
Table 17 illustrates that the highest levels are observed in the mobile
homes; emanation from the pressed board appears to be the primary source.
The high aldehyde concentrations in the mobile homes were expected, con-
sidering the fact that both structures were almost new. The Denver house,
a pre-1940 structure, indicates that emanation from construction material
may continue for many years; however, independent research shows that it
should decrease with time (Andersen et ah, 1975). In view of the fact that
all dwellings in the first group register aldehyde levels that exceed the
ASHRAE recommended indoor standard of 300 yg/m , the conclusion reached
is that further research is warranted on this subject.
It has been suggested that material used to insulate houses emanates
formaldehydes. The data base available to this study cannot verify this
statement; however, this energy conserving measure and others, which are
often taken at the same time, result in a decrease of the infiltration
rate of the retrofitted dwelling. Under these conditions, low infiltra-
tion rates and possibly high formaldehyde emanation rates, the aldehyde
concentrations in the indoor-residential environment will reach high and
possibly hazardous levels.
Although high aldehyde levels have been observed in this study, the
majority of the dwellings monitored have registered levels below 200 yg/m3.
Lead (Pb)
One half of the 47-mm glass fiber filter, used in collecting total
particulate matter, is analyzed in the laboratory to determine indoor and
outdoor Fb levels. The realization that Pb concentrations do not vary
greatly vithin a residence, combined with the high cost of the analytical
method used (atomic absorption spectrophotometry), induced the following
change ir the experimental design: instead of determining the daily Pb
concentrctions from all four sites simultaneously, the Pb levels were
determined from one site, and the site of measurement was rotated con-
tinuously, thus requiring 4 d to estimate levels from all sampling sites.
One of the sampled residences, the Baltimore conventional house,
registered Pb ambient levels between 4.0 and 12.5 yg/m; the indoor Pb
levels were equally high, varying between 0.1 and 12.0 yg/m . The daily
Pb concentrations observed in this house do not surpass the ASHRAE recom-
mended indoor standard of yg/m ; however, they are unusually high levels
and are a cause of concern.
Table 18, the Pb data collected for this project, shows that the
observed Pb levels during the second monitoring period in the same Balti-
more Conventional residence varied between 0.1 and 1.8 yg/m3 in the out-
door environment, and between 0.1 and 0.5 yg/m3 in the indoor environment.
These levels are within the typical observed range of 0.1 to 2.8 yg/m3.
In addition, the Pb levels sampled 2 weeks earlier in an almost identical
residence, the Baltimore experimental house, a half mile from the Baltimore
conventional dwelling, were within the typical range. Owing to expressed
-44-
-------
TABLE 18. Pb CONCENTRATIONS SAMPLED IN 18 RESIDENCES FOR BOTH THE INIKX3R AND OUTDOOR ENVIRONMENTS
(P^Ug/m3)
*
*
N
Ik
«fc
N
<4
*
£
N
%
d*t
location
1
8 8
*S
I*
1
i
Ia
i
it
a
si
ii
sl
3U
i!
* 6
3U
a
i!
1
|
ft !
3 1
3a
i
8
&!
8 1
a1
*3
9
&i
8 I
ns
1*
!
n
fi
Pittsburgh
Low fUje 3
Pittsburgh
Mobile 1
Pittsburgh
Mobil* 2
ft
Is
IM
s
6 *
i -ft
|9
Ml
fa
1-
ix
1
Outdoor
0.9
1.0
o.s
2.6
12.5
0.3
1.7
0.2
0.6
Void
0.1
0.2
2.7
1.2
0.3
0.4
0.8
0.6
2
Kitchen
0.3
0.3
0.2
0,4
1.5
0.1
1.2
0.5
0.2
Void
0.3
0.2
0.9
0. 2
0.1
0.4
0.7
0.6
3
ledtaom
0.8
0.3
0.2
0.4
12.0
0.4
0.7
0.1
0.2
0.1
0.3
0.1
0.6
Void
0.1
0.6
0.8
0,1
4
tiring
Room
0.6
0.2
0.1
0.1
0.2/
7.3
0.2
Void
0.2
Void
0.1
0.2
0.1
0.8
Void
0.1
0.3
0.3
0.4
5
Outdoor
0.7
0.3
l.S
0.4
4.7
0.4
0.6
0.2
0.2
0.4
0.1
0.4
1.1
0.7
0.7
0.8
0.8
!,3
6
Kitchen
0.9
0.1
0.1
0.3
6.2
0.2
0.6
0.2
0.2
0.1
0.1
0.3
0.4
0.1
0.1
0.5
0.4
0.7
7
Bedtooro
0.4
0.3
0.1
1.0
6.7
2.0
0.8
0.1
0.1
Void
0.1
0.1
0.3
0.2
0.1
0.7
0.1
0.2
8
tiring
Room
0,4
0.4
0.2
0.2
2.2
0.1
0.6
0.2
0,2
0.1
0.1
0.7
0.3
0.1
0.1
0.6
0.1
0.1
9
Outdoor
1.0
1.2
0.3
1.0
10.8
0.3
1.2
Void
0.2
0.3
0.2
0.9
0.7
0.4
0.4
0.9
0.4
Void
10
Kitcfa«n
lf6
0,9
0,3
0,5
5.8
0.1
1.0
Void
0.2
0.1
0.2
0.3
2.1
0.3
0.1
0.4
0.8
0.4
11
Bttitewm
0.7
0.4
0.2
0.7
8.2
0.1
0.8
void
0,3
0.1
0.6
0.3
0.4
0.2
0.1
0.7
0.4
0,5
12
IMag
Room
0.6
0.4
0.5
0.45
4.8
0.2
1.3
Void
0.1
0.1
Void
0.3
0.8
0.1
0.2
0.5
0.3
0.3
0
Oofefoe*
0.8
0*7
0.2
2.8
4.0
1.8
2.0
Void
0,5
0.5
0.4
0.6
1.9
0.4
0.9
0.5
0.4
1.1
H
KKchcs
0.5
0.5
0.1
0.4
0.1
O.S
void
Void
O.S
Void
Void
0.3
0.6
0.2
0.2
Void
0.4
0.3
IS
BeAorim
Void
0.1
0.1
0.7
Void
Void
Void
Void
0.1
0.2
Void
0.6
Void
0.5
Void
Void
0.2
Void
16
Urine
Room
Void
Void
Void
Vo4d
Void
Void
Void
Void
Void
0.2
Void
Void
Void
Void
Void
Void
Void
Void
-------
concern and the consequences of the magnitude of Pb levels observed in
this residence, a special effort was undertaken to determine the cause.
Unfortunately, the effort was inconclusive, and specific sources were not
identified. Therefore, the high ambient Pb levels were attributed to
unknown local ambient Pb sources. The possibility of improperly operating
sampling instruments or faulty analysis procedures has not been excluded.
The^State of California has promulgated a 30-d average Pb standard of
1.5 ng/m . This level is exceeded in only a few instances by the observed
24-h Pb averages. And, it is only in the special case of the Baltimore
conventional house that the data base indicated a violation of this stan-
dard over the monitoring period.
The observed outdoor Pb concentrations are typically higher than the
corresponding (within the 4-d time interval) indoor concentrations; how-
ever, there are a few instances of indoor sources. The ratio value of
outdoor Pb concentrations over the corresponding indoor averages is shown
in Table 19. Only a small percentage of these values is less than one,
indicating that the impact of indoor Pb-generating sources is evident
only under uncommon conditions; such uncommon conditions are identified
in the elemental Pb analysis. (See the discussion on elemental analysis.)
The major surce of outdoor lead is the automobile. A common source
of lead in the indoor environment is lead-based wall paint. Such paints
an3 found only in older houses because their use has been recently banned.
Titanium-basedpaint is used in newer residences. It has been recently
suggested that Indoor residential environments of affluent communities
contain high Pb levels of automotive nature. The study of the indoor-
ou:door Pb data base verifies only half of the above suggestion: newer
hojses manifest Pb concentrations of automotive nature; however, the
majority of the observed concentrations are low and therefore not hazardous
to the public health.
Asjestos
During the performance of this program, 10 air samples were collected
on 47-mm Nuclepore filters and analyzed for asbestos fibers. The samples
we^e obtained in two Chicago residences during the summer monitoring
period in 1977. Five samples were collected at the experimental residence,
ani! five were collected at the conventional residence.
For analysis, the filters were divided into eight sections. Each of
six sections per filter was examined by dispersion staining microscopy
at 100X magnification for one of the six principal asbestos minerals.
The six mineral types are chrysotile, amosite, crocidolite, anthophyllite,
act nollte, and tremolite.
Dispersion staining uses a liquid of specific refractive index for
each asbestos type. As polarized light passes through asbestos fibers
surrounded by the liquid, they appear in color. The color observed
-46-
-------
TABLE 1<>. RATIO VAUJF.S Op OUTDOOR Pb CONCENTRATIONS WITH CORRESPONDING INDOOR AVERAGES
1st 4-d
2nd 4-d
3rd 4-d
4th 4-d
Cycle
Cycle
Cycle
Cycle
Indoor
Indoor
Indoor
Indoor
Outdoor
Average
Out/In
Outdoor
Average
Out/In
Outdoor
Average
Out/In
Outdoor
Average
Out/In
W aihington
Experimental I
0.9
0, 63
1.4
0.7
0. 57
1. 2
1.0
0.97
1.0
0.8
0.5
1. 6
Baltimore
Experimental I
1.0
0, 27
3.7
0.3
0.27
1.1
1.2
0.57
2.1
0.7
0.3
2.3
Baltimore
Experimental II
0. 5
0. 17
2.9
l.S
0, 13
0.3
0.3
0. 33
0. 91*
0.2
0. 10
2.0
W aihington
Conventional
2. 6
0. 30
8.7
0.4
0. 50
0.8*
1.0
0.48
2.1
2.8
0.37
2.4
Baltimore
Conventional I
12. 5
5.3
2.4
4.7
S.O
0. 94*
10.8
6.3
1.7
4.0
0. 1
40.0
Baltimore
Conventional II
0. 3
0.23
1.3
0.4
0.17
2.4
0.3
0.13
2.3
1.8
0.5
3.6
Denver
Conventional
1. 7
0.95
1.8
0.6
0.67
0. 89*
1.2
1.0
1.2
2.0
-
-
Chicago
Conventional II
0.2
0. 27
0.74*
0.2
0.17
1.2
-
-
-
-
-
-
Chicago
Experimental I
0.6
0. 2
3.0
0.2
0.17
1.2
0.2
0.2
1.0
0.5
0.3
1. 7
Chicago
Experimental II
-
0, 1
-
0.4
0.1
4.0
0.3
0.1
3.0
0.5
0.2
2. S
Pitaburgh Low Rlae 1
0.1
0.27
0.37
0.1
0.1
1.0
0.2
0.4
0.24*
0.4
-
-
Pltaburgh Low Rlae 2
0.2
0.13
l.S
0.4
0. 37
1. 1
0.9
0.3
3.0
0.6
0.45
1.3
Pita burgh Low Rlae 3
2.7
0.76
3.6
1.1
0.33
3.33
0.7
1. 1
0.64*
1.9
0.6
3.2
Pltaburgh Mobil* 1
1.2
0,2
6.1
0.7
0.13
6.2
0.4
0.2
2.0
0.4
0.35
1.1
Plttiburgh Mobil* 2
0. J
0.1
3.0
0.7
0.1
7.0
0.4
0.13
3.1
0.9
0.2
4.5
Pltaburgh High RU* 1
0.4
0.43
0.93*
0.8
0.6
1.3
0.9
0.53
1.7
0.5
-
-
Pltaburgh High Rli* 2
0. S
0.6
1.3
0.8
0.2
4.0
0.4
0.5
0. 8*
0.4
0.3
1.3
Pltaburgh High Rla* 3
0.S
0.37
1.6
1.3
0.33
3.9
-
0.4
-
1.1
0.3
3.7
* Indoor •oasrc*.
-47-
-------
depends on the type of asbestos and whether the fiber is parallel or per-
pendicular to the plane of the polarized light. For example, amosite in
liquid refractive index 1.67 appears magenta when aligned with the polar-
ized light and yellow when perpendicular to it.
The seventh section of each filter was examined by the Analytical
Method for Detection of Asbestos Fibers (Federal Register, 1972) to obtain
a quantitative analysis of asbestos fibers. This method uses phase
contrast microscopy at 400X magnification to make filters appear dis-
tinct fronutheir background. A portion reticle with a field area of
3.08 x 10" mm was used; 100 fields were examined on each filter
section.
The minimum detection limit of fibers in the air sampled was deter-
mined as follows:
Usable area of filter = 1320 mm2
Area of 1 field = 3.08 x 10-3 mm2
Area of 100 fields = 0.308 mm?
Detection limit, = , ... 1320 mm? .. 10~6
fibers/cm^ x 0.308 mmz air volume, m3
The eighth section (1/4 of the filter) was retained as a reserve.
No fibers of any of the six asbestos forms were found during the
dispersion staining examinations. No asbestos form fibers were found
by the Federal Register method count. Table 20 below lists the filters,
the air volumes samples, and the detection limits.
TABLE 20. ASBESTOS COUNT RESULTS
Sampling
Air Volume
Detection limit
Residence
Site
Loc ation
Date
Time (hours)
(m3 at STP )
(Fibers/cm3)
Chicago
1
Ambient.
7/22/77
8
23. 7
1.81 x 10-4
Experimental
4
Living Room
7/22/77
8
23. 5
1. 82 x 10-4
4
Living Room
7/23/77
8
17. 8
2. 41 x 10-4
4
Living Room
7/25/77
8
23.8
1.80 x 10-4
4
Living Room
7/26/77
8. 5
25. 8
1. 66 x 10-4
Chicago
1
Ambient
7/31/77
8
47.4
O. 90 x 10-4
Conventional
4
Living Room
7/31/77
8
28.7
1.49 x 10"4
4
Living Room
8/1/77
7
24. 7
1. 74 x 10"4
4
Living Room
8/2/77
7. 3
26.2
1. 64 x 10-4
4
Living Room
8/3/77
8
28. 1
1. S3 x 10-4
-48-
-------
All filters had fiber counts below the detection limit. The detec- 3
tion limits are well below the 8-h time-weighted average of two fibers/cm
of air allowed by the U.S. Department of Labor for occupational exposure
(Federal Register, 1972).
Elemental Analysis
Aerosol samplers used in this study are continuous Nuclepore filter
devices which provide a smoothly varying sample streak by a sliding,
sucking orifice (Nelson, 1977). Elemental analysis is carried out in a
stepwise fashion using proton-induced X-ray emission, PIXE (Johansson
e: al., 1975), providing a record with 2-h time resolution of the time
variability in elemental concentrations of the suspended particulate
matter. Samples were collected throughout the indoor-outdoor field monitor-
ing program, for a number of residences monitored early in the program;
a total of 84 2-hourly analyses of each sample were performed by PIXE for
S, CI, K, Ca, Ti, Mn, Fe, Cu, Zn, Br, and Pb. Five samples were collected
for elemental analysis; three indoors (in the kitchen, living room, and
bedroom) and two outdoors (one in front of the house and the other in the
back). For a variety of technical reasons, analysis for each residence
was undertaken only on representative sites. The results of time vari-
ability of aerosol composition from three residences are presented in this
section.
During the first visit at the Baltimore experimental residence, the
indoor concentration of Pb fluctuated synchronously with the corresponding
outdoor concentrations. The air exchange rate is essentially instantane-
ous within the time resolution of the experiment—2 h. A similar corre-
spondence in the optimal points of Pb levels is observed in the Baltimore
conventional residence. A small filtering effect observed in this resi-
dence may indicate that the ambient site is not representative. The loca-
tion of the ambient sites with respect to the house is of greater importance
in aerosol considerations than in gas pollutants. The correspondence of
the maximum concentrations indicates that the indoor sources of lead, if
any, are not substantial.
The Denver single family house illustrates that under certain condi-
tions Indoor Pb sources may exist. Indoor sources of lead are attributed
to reentralnment of particulate matter caused by the dally activity in the
house. Outdoor concentrations of most elements sampled 1n this residence
were high during the first 40 h of sampling, then low for a 4-d period,
and again high during the final 30 h, apparently reflecting weather
changes during the sampling period. Of special interest are Pb and
Br, both derived primarily from automotive exhaust, which during the
middle 4 d had higher indoor concentrations than outdoor (see Figure 9).
The concentration units used in Figure 9 and all subsequent figures
illustrating measurements by PIXE correspond to ng/m absolute con-
centration to a significance of a factor of two depending on the flow
rate calibration of the experimental arrangement. Indoor sources of Pb,
-49-
-------
)
- 100
5 M
S w
Iw
5 90
40
30
Pb Out
—
door
| ~~
Tk
1
1
r
J-
/
f
u
.-¦jy
V
0 29 40 SO
T1m St*p
E *
.2
I* 1
if
sr!
<5:1
.4
A
e
Tint
«r-
St*t>
r
W
T
Ir/H flutdatr
1
'—f
1
A
r»
t
I
a/1
I
K
,r
* 1
1
14M
1
F
r*
n
T1m Stay
40 <0
Tlaw Stap
Figure 9. Time variation of Pb and Br concentrations and Br/Pb ratio in DenveT single-family dwelling. Time steps represent
2-hour sampling intervals, starting and ending at 1600 hoursOctober 5-12, 1976. Mean values Pb, Br, and Br/Pb for each of
three time periods are indicated by horizontal bars, with their standard deviations as vertical bars.
-50-
-------
with Br/Pb = 0.6, predominated in the middle period. Infiltration from
outside sources having Br/Pb = 0.3 predominated in the initial and final
periods. A simple mixing model for indoor- and outdoor-generated Pb and
Br within this residence leads to an estimate of 0.3 as the factor by which
outdoor lead is reduced in concentration during its penetration indoors
(Moschandreas et al., 1977). Owing to the prevailing meteorological condi-
tions, the Denver concept does not apply in the two Baltimore residences;
therefore, it has not been tested. However, a general agreement is
manifested among the residences when similar weather conditions and
ambient lead concentrations are observed.
Potassium is associated with organic materials. Indoor sources include
cooking, smoking, emissions from wood, and general human activity. It is
not considered hazardous to residential occupants, but its behavior as an
indoor-generated element is of interest. Potassium fluctuations are
associated with indoor activity in various indoor sites. Potassium levels
increase at times of cooking and when large numbers of people are gathered
for such activities as card games and parties. Concentrations of potassium
fluctuate indoors independently from the corresponding outdoor variation;
the variation of the indoor potassium concentrations within each residence
is connected with the life style of the inhabitants. The unusual diurnal
variations observed in the Baltimore experimental residence (Figure 10)
show peaks early in the morning; this variation may be explained by the
fact that a member of the family is a nurse who does some routine activi-
ties such as cooking and eating shortly after midnight when she returns
from work. Similar but not as clearly identifiable associations between
household activity and potassium concentrations have been observed in all
residences.
Iron is associated with relatively large particulate size soil dust.
Figure 11, illustrating the iron fluctuations in the Baltimore experimental
residence, shows that Fe concentrations are lower indoors than outdoors; it
also shows that optimal levels do not correspond. The overall trends within
each residence are similar. This lack of indoor-outdoor correspondence is
attributed to the large size of the Fe particles. It is of interest to
note that the ambient Fe concentrations 1n the Denver residence display
a three-step fluctuation similar to the one observed for Pb concentrations,
but the corresponding indoor Fe concentrations do not display fluctuations
that may be explained by an indoor source. The difference 1n the indoor
fluctuations between Pb and Fe may be attributed to the size of the particu-
lates; the larger particulate matter 1s heavier and therefore does not remain
airborne as long as the finer Pb particles.
Finally, the indoor structure observed for elemental sulfur follows
the trends of the outdoor behavior. Figure 12, illustrating the Baltimore
conventional residence, but representative of all houses monitored, shows
that outdoor levels are higher than indoor in all residences. Thus, it can
be concluded that the sulfur found indoors is of outdoor origin. The Denver
conventional residence and the Baltimore conventional residence have gas
cooking and heating equipment: the indoor/outdoor sulfur ratio appears
-------
(a) Outdoor
(c) Bedroom
Figure 10. Time variation of potassium (K) concentrations in Baltimore
experimental residence. Time represents 2-h sampling interval.
-52-
-------
(a) Outdoor
2K
P
0 * 10 IS n » SO IS 40 45 SO SS <10 US to 7S 80 HI
Tim* Step
(b) Kitchen
(c) Bedroom
Figure 11. Time variation of iron (Fe) concentrations in Baltimore experimental
residence. Time step represents 2-h sampling intervals.
-53-
-------
(a) Outdoor
(b) Kitchen
(c) Bedroom
Figure 12. Time variation of sulfur (S) concentrations in Baltimore conventional
residence. Time step represents 2-h sampling intervals.
-54-
-------
to be higher in these two residences but it remains less than one. Similar
analyses can be carried out for the remaining elements—Al, Si, CI, Ca, Cu,
and Zn--but this has not yet been done.
Episodic Pollutant Release
The objective of this series of experiments is twofold; the first objec-
tive is to continue testing the methods developed by I IT Research Institute
under the direction of Dr. A. Dravnieks (1977) for collecting samples of
organic contaminants; the second objective is to determine a time history of
aerosol pollutants introduced into the indoor environment by an episodic
release. As this is the beginning of a scientific effort in this specialized
air pollution field (episodic release in indoor environments), we must not
expect to totally resolve all problems; however, we will seek to answer some
of the major questions.
The sources selected to simulate episodic releases were an oven spray
cleaner, an aerosol anti-perspirant, and a spray furniture polish. Locations
for release of these sources were the kitchen to simulate oven cleaning,
the bathroom to represent morning activities, and the living room as prime
area for house cleaning activities. Each of the three Pittsburgh high-rise
apartments was used for a different episodic release experiment.
From the simulated releases the major contaminants found to originate
in the apartments were terpenes, especially limonene; some chlorinated and
Freon-related organic compounds such as dichlorobenzene; and oxygenated
organic compounds Including acetone and ethanol. These contaminants are
associated with the episodic releases as components of the odorant, pro-
pellant, or solvent. As such they were found 1n measurable concentrations
1n the Indoor samples, while in the outdoor ambient samples little or no
concentrations were detected for these contaminants.
Pollutant distribution was investigated by collecting samples at
designated time intervals, generally 15 min, in the location of release
and one adjacent area, plus continuous 1-h samples in other locations away
from the source. In general, peaks occurred in the area of the source
and the adjacent locations during the second samples, 15 to 30 min after
release. In other locations within the residence, the contaminants were
detected but generally at concentrations slightly lower than those at or
near the source locations. This Indicates that pollutant distribution was
accomplished throughout the entire residence.
Similar results were obtained with an episodic release of the SF6
tracer gas. Figure 13 shows the measured SF6 concentrations plotted against
time. The source location is the living room, and the adjacent locations
are the kitchen and hall. Since SFg is an inert gasf a requirement for a
tracer, 1t does not react in the same manner as the components released
1n the other episodic experiments. For this reason peak concentrations 1n
source and adjacent locations occur within 2 min. From this experiment,
it is shown that SFg tracer gas reached all areas in the residence
-55-
-------
PITTSBURGH HIGH-RISE #3
17 JUNE 77
10-
TIME 1645
9-
8-
7-
6-
Source
Livini
~
~
Rootr
Kitchen
Hall
Bedroim
9-
8-
7-
6-
Shi
iL
1
1*
To"
20
TIME, minutes
str
60
7b
Figure 13. Episodic release of SFg gas.
-56-
-------
and accomplished complete mixing in about 30 min, at which time all loca-
tions obtained approximately identical concentrations.
The other objective of this study was to test a method, developed
by Dr. A. Dravnieks, for collecting samples of organic contaminants from
the air without the use of air-contaminating electrical devices. The
method Dr. Dravnieks developed allows for the transfer of samples to a
gas-chromatographic/mass-spectrometric system for identification of con-
taminants. The data collected in this study substantiates Dr. Dravnieks
findings that this technique is suitable for detecting organic contaminant
species at parts per billion (ppb) concentration levels.
Data reports prepared by the I IT Research Institute in Chicago,
Illinois, are shown in Chapter 2 (Volume II) of this final report.
Questionnaire-Activity Index
An activity scale was developed to rate household activity in each
of the residences monitored. Information for the development of this
scale was drawn from the family daily log. The questionnaire, answered
by the head of each household, provided a record of the daily occupant
activity. The original and the revised version of this questionnaire
are shown as Figures 14 and 15, respectively.
A point scale, Table 21, was developed to interpret the information
made available by the family daily logs. The summation of these point
values for all activities denotes the daily activity index (daily totals);
the mean of the daily totals is the average daily index. A typical day is
identified when the observed daily index falls within a range of 4-8 points.
Those days when the activity index falls within this range are considered
"typical" or "routine." Days when the activity index exceeds 8 points are
considered "active days," and days when the activity index falls below 4
points are considered "inactive." On a daily basis, the distribution of
the activity index shows that 24% of the days observed are rated inactive,
56% typical, and 20% active.
The average daily index also enables a characterization of each
family as either typical, active, or inactive with regard to their average
daily activity pattern. Table 22 shows that seven of the families ques-
tioned are inactive, seven are typical, and two are active. This distribu-
tion is constrained by the integrity of the responses by the residents to
the questionnaire.
This activity index is an important element in determining the mode
of operation of indoor sources. The activity index and its relationship
to indoor TSP levels are discussed in Section 3 with reference to the TSP
steady state model.
-57-
-------
Site I. D.
Date:
Location;
Day of Week:
Yes
No
Comments
1. Did you cook breakfast? What time?
2. Did you cook lunch? What time?
3. Did you cook dinner? What time?
4. Did you have guests? What time?
If yes, how many?
5. Did you cook anything special which
took more cooking time?
6. Did you do any cleaning?
If yes, what kind of cleaning
material did you use?
7. Did you use any air freshener?
If yes, what kind (brand)?
8. Did you use any aerosols?
If yes, what kind (brand)?
9. Did you open the windows?
If yes, between what hours?
10. Did you vacuum? If yes, what time?
11. Did you turn on the range hood fan
while cooking?
12. Was there any period of time that
nobody was home; what hours?
13. Did you use the bathroom exhaust fan?
If yes, how many times?
14. Did you do the laundry today?
15. Did you use the clothes dryer? If
yes, how long did you run the dryei?
16. Did you, family, or guest smoke ?
What was smoked and how many?
During what period?
17. Did you use a fireplace? What kind
and for what period?
18. Other activities that were unusual.
Figure 14. Daily activity record.
-58-
-------
DAILY ACTIVITY RECORD
Dally activity record of the residence for the day of
1.
Did you cook breakfast?
What did you cook?
Yes( :
) No(
)
What time did you start?
How long did
you
cook?
2.
Did you cook lunch?
What did you cook?
Yes(
) No{
)
What time did you start?
How long did
you
cook?
3.
Did you cook dinner?
What did you cook?
Yes(
) N0(
) .
What time did you start? How long did you cook?
4. Did you cook or bake anthlng other than regular meals? Yes( ) No( )
What did you cook or bake?
What time did you start? " How long did you cook?
5. Did you turn on the range hood fan while cooking? Yes( ) No( )
What time did you turn it on? off?
6. Did you wash dishes today? Yes{ ) No( )
At what times? _For how long?
7. Did you have any guests today? Yes( ) No( )
How many guests did you have?
What time did they arrive? Leave?
8. Was there any period of time that nobody was home? Yes( ) No( )
What time? From to
9. Did you, your family or any of your guests smoke? Yes( ) No( )
What was smoked? cigarettes ( ) cigar ( ) pipe ( )
How many or how often? What times?
In what rooms?
10. Did you do any cleaning today? Yes( ) No( )
What rooms did you clean?
What time did you start? Stop
Was cleaning continuous? ~1 5 Intermittent? I 5
List cleaning materials used ____________________________
11. Did you vacuum clean today? Yes( ) No( )
At what time did you start? Stop?
Was vacuuming continuous? ( Intermittent? 5
What rooms did you clean?
12. Did you do the laundry today? Yes( ) No( )
What time did you start? Stop?
How many loads did you do?
13. Did you use the clothes dryer? Yes( ) No( )
What time did you Start? Stop?
Did the dryer run continuous? Yes( ) Hoi J
14. Old you use any air freshener today? Yes( ) No{ )
At what time? What brand did you use?
In what room did you use 1t7
15. Old you use any aerosols today? Yes( ) No( )
At what time? What brand did you use?
In what room did you use It? ~~ '
16. 01d you open any windows today? Yes( ) No( )
At what time were they opened?_ Closed?
In what rooms were they opened/
17. Old you use a fireplace? Yes( ) No( )
What time did you use 1t? Start Stop
18. Please 11st any additional activities that may have taken place In the house that have not been
listed. This would Include activities related to hobbles, repairs, painting, etc. A short des-
cription of the activity, any materials used 1f any, room 1n which the activity had taken place,
and time of starting and ending should be Included. If there are any questions on what should be
Included please ask the PEDCo personnel for assistance.
Activity record completed by Date
Activity record checked by bate'
Figure 15. Revised version of dally activity record.
-59-
-------
TABLE 21. ACTIVITY POINT SCALE
Preparation of Meals
Use of Bathroom Exhaust
Use of Range Hood Fan
Additional People in Dwelling
Housecleaning Activity
Use of Cleaning Agents
Use of Aerosols/Air Fresheners
Open Windows
Vacuuming
Doing Laundry
Use of Clothes Dryer
Cigarette Smoking
Empty House
Special Activities
GEOMET Test
1 pt. per meal
-1 pt.
-1 pt.
1 -3 people = 1 pt.
4-6 people = 2 pts.
7-9 people = 3 pts.
10+ people = 4 pts.
1 pt,
1 pt. per agent
1 pt. per agent
1 pt.
1 pt/30 min
1 pt. per load
1 pt. per 30 min
1 pt. for 1st pack
2 pts. for over 1 pack
-1 pt. for time over 2 h
-11 pts. for days when house is
empty 72 h
1 pt. per activity
1 pt.
TABLE 22. CHARACTERIZATION OF FAMILY TYPE
Average Daily
Residence
Rating
Category
Denver Conventional
3.4
Inactive
Baltimore Experimental II
3.4
Pittsburgh High-Rise Apt. 1
3. 5
Baltimore Conventional I
3.5
Pittsburgh High-Rise Apt. 2
3. 6
Pittsburgh Low-Rise Apt. 1
3. 7
Pittsburgh Mobile 1
3.8
Pittsburgh Mobile 2
4.7
Typical
Baltimore Experimental I
5.0
Baltimore Conventional II
5. 1
Washington Conventional I
5.2
Chicago Experimental I
5.6
Pittsburgh High-Rise Apt. 3
6.6
Pittsburgh Low-Rise Apt. 3
6.8
Pittsburgh Low-Rise Apt. 2
8.9
Active
Washington Experimental I
9. 5
-60-
-------
SF^ Tracer Gas Experiments
Air infiltration in residences is an important factor in energy con-
sumption as related to the comfort control of the indoor environment.
Indoor air quality characteristics are related to the rates of air infiltra-
tion by the introduction of outdoor air pollutants and the distribution of
indoor-generated pollutants. The investigation of residential energy con-
sumption and Indoor-outdoor air quality relationships are discussed 1n
other sections of this report. An important element of these investigations
is the determination of a structure's representative air exchange rate. The
purpose of the tracer studies is to experimentally measure the rates of air
exchanges and their associated dependencies in other factors. The primary
objectives of these experiments are as follows:
• Experimental determination of each structure's air exchange
rate under normal real-life conditions
t Investigation of parameters affecting air exchange rates
such as meteorological conditions, human activity patterns,
and structural and mechanical parameters
t Investigation of pollutant distribution potential.
This section discusses the experimental design used in the tracer studies
and the results and conclusions derived from over 6t) such experiments.
To determine a representative air exchange rate for each residence,
sulfur hexafluoride (SFfi) was used as a tracer gas. The tracer was released
in the residence and monitored throughout the structure. The decrease rate
of the SFfi concentrations is a measure of the air exchange rate. To moni-
tor the distribution and flow of the SFg tracer gas, two sampling techniques
were employed.
One procedure collected samples through a network monitoring system.
Sample probes were located 1n three major Indoor activity areas. These
areas were generally selected as the living room, kitchen, and master bed-
room. Samples were drawn through teflon lines at recorded time Intervals
Into the mobile laboratory adjacent to the structure under Investigation.
Concentrations were determined by electron capture gas chromatography.
A second technique developed to estimate air exchange rates with
SFg tracer gas investigated all individual areas within each structure.
Researchers obtained 30 cm disposable plastic syringes with the needle
cap cut and sealed to make the instrument air tight. Samples were collected
by hand at recorded time Intervals with numbered syringes in designated
areas of the residence. In general these areas were rooms, halls, stair-
wells, and foyers. Syringe samples and coded forms were sent to the
California Institute of Technology (Caltech) for analysis of SFg
corcentrations.
-61-
-------
The Caltech facility consists of 16 electron capture gas chromatographs
(constructed in-house), the output signals of which are processed through two
digital integrators (Spectra Physics-System I Integration). A stainless
steel coaxial electron capture detector, electronically insulated with Teflon
and Nylon plugs, is pulsed periodically with a -25 V, 1- ys pulse every 200 ps.
The radioactive source used is a 200 mCi H3 source bonded to a titanium sub-
strate (U.S. Radium Corp., Parsippany, New Jersey). Analysis for SF5 was
achieved using a stainless steel column (39" x 0.25" o.d., 0.18" i.d.) packed
with 5A 80-100 mesh molecular sieve (Alltech Assoc., Arlington Heights,
Illinois). The columns were conditioned at 300 °C overnight with N2 flowing
continuously. Using prepurified N2 at 60 cm-fymin as the carrier gas, SFg was
eluted in 18 s and O2 in 45 s. A typical calibration curve is shown in
Figure 16. Note that the linear range is from 10~8 to 10"^; at 10~12 the
signal-to-noise ratio is still better than 3 to 1.
Three categories of physical data were monitored in the course of
the SF6 tracer experiments. The first data set consists of the physical
characteristics unique to each structure. This data set contains informa-
tion on the type and quality of each structure and its building components,
along with their physical dimensions. Table 23 summarizes the portion of
this data set relevant to the infiltration studies.
The second data set consists of the meteorological parameters monitored
during the tracer studies. The average indoor/outdoor temperature and
relative humidity differences, alonq with average wind speed and direction,
are also shown in Table 23.
A third category of data contains information obtained from daily
family activity logs. These logs were investigated to determine the impact
of occupant activity on the air exchange rate for each residence. Informa-
tion obtained from this data consisted of the number, duration, and approxi-
mate size of window and door openings.
To determine the total air exchange rate of each different structure
the tracer dilution technique was employed. An instantaneous release of
a known quantity of SFg tracer gas was injected into the forced air
ventilation system just ahead of the blower fan. The blower fan was
initially turned on for 20 min to insure a uniform distribution throughout
the structure. Dilution of the tracer concentration occurs with the intro-
duction of infiltrated air. The decay in the tracer concentrations was
found to be exponential as shown by the following equation:
C = C0e"vt
where: C = tracer concentration at time t
CQ = tracer concentration at time t = 0
t = time 1n hours
v = air changes per hour.
-62-
-------
icr6 e—nnymrj i i 1j 11111—i ri|iiiii—ri i ]i
10-7 =
I I I |IIbfc
~
~
a
~
iUi I I I mill ' I I Him t I t mill i I Hun
103 10" 105 iO6
INTEGRATOR RRER
107
12/5/77
Figure 16. Typical calibration curve for SFg.
-63-
-------
TABLE 23. PHYSICAL AND METEOROLOGICAL PARAMETERS
Physical Meteorological
Residence
Structural
Type
Volume
(ft3)
Date
Average AT
(P°)
Average WS
(mph)
Wind
Sectors
Average &RH
(%)Range
Washington Conventional
Detached
21,585
July, 1976
0-14
1-2
SW, NE
10 - 21
January, 1978
37 - 42
3-6
NW - E
8-66
Washington Experimental
Detached
19,880
August, 1976
4-5
2-5
N
3
December, 1977
22 - 35
3-10
E - SW
47 - 53
Baltimore Experimental
Semidetached
11,650
August, 1976
0-11
3-5
E - S
5-10
January, 1977
47 - 57
2-17
SSE, NW
15 - 37
Baltimore Conventional
Semidetached
13,575
September, 1976
2-3
3-5
NW - N
10 - 18
January, 1977
42 - 55
5-20
S - NW
24 - 31
Denver Conventional
Detached
20,100
October, 1976
0-16
1-5
NE, S
9
Chicago Conventional
Detached
18,170
November, 1976
52 - 62
N/'A
N/A
1 - 23
August, 1977
3-4
1 - 5
S - W
3-21
Chicago Experimental
Detached
26,330
December, 1976
33 - 39
0-8
SE - N
9-33
July, 1977
5-8
5-16
NW - NE
5-7
Pittsburgh Mobile 1
Mobile
6,380
February, 1977
35 - 55
5-12
S - N
5-15
Pittsburgh Mobile 2
Mobile
6,850
March, 1977
9-36
2-12
S - w
10 - 26
Pittsburgh Low-Rise 1
Low-Rise Apt.
6,460
March, 1977
1-30
6-22
SW - NW
3-31
Pittsburgh Low-Rise 2
Low-Rise Apt.
6,460
April, 1977
7-30
10 - 19
W - N
3-23
Pittsburgh Low-Rise 3
Low-Rise Apt.
6,460
May, 1977
10 - 20
8-11
W
20 - 40
Pittsburgh High-Rise 1
High-Rise Apt.
7,240
May, 1977
2-12
5-11
NW - NE
12 - 17
Pittsburgh High-Rise 2
High-Rise Apt.
7,240
June, 1977
5-21
7-13
W - NE
20 - 53
Pittsburgh High-Rise 3
High-Rise Apt.
7,240
June, 1977
1 - 9
5-19
S - NW
5-10
-------
The building air exchange rate is obtained from the slope of a semilog
plot of the natural logarithm of the tracer concentrations versus time;
the calculated air exchange rate is a comprehensive one, it includes
infiltration rate, mechanical ventilation rate and recirculation rate.
Figure 17 shows a typical tracer curve from which the air exchange rate
is estimated. The experimentally determined ranges of air exchange rates
are shown for each residence in Table 24.
Additionally, diluted SFg gas was released continuously for periods
o1 3 to 4 h. A known concentration of the tracer is emitted at a constant
rcte and monitored as it disperses throughout the residence. The objec-
tive of this release is to experimentally determine the time required for
a pollutant qenerated by a single source, often in the kitchen, to disperse
in the residence, and to reach a constant level as illustrated in Figure 18.
The determination of a structure's air exchange rate requires that
the indoor and outdoor tracer concentrations are zero. This was verified
by collecting samples both indoors and outdoors before the tracer was
released.
The findings reported in this discussion are based on data obtained
from tracer experiments performed in inhabited residences under real-
life conditions. During these conditions the measurable parameters are
in a constant state of flux and what appears to be a predominant factor
in one experiment has minimal effects during another. For this reason
the effect on the structure's air exchange rate directly attributable
to any one particular parameter is difficult to quantify. However,
certain relationships became apparent and are presented in the following
discussion.
Differences between indoor and outdoor temperatures have been
considered as one of the major causes for air infiltration in struc-
tures. From the investigation of average indoor/outdoor temperature
differentials for structures less than three stories in height, a definite
relationship with the rates of air exchange was observed. Figure 19 1s
a graph of average temperature differentials plotted against measured
air exchange rates. As differences between indoor and outdoor temperatures
increase, there is a corresponding increase in the air exchange rates.
This relationship is represented by the following equation:
v = 0.235 + 0.013 AT .
This relationship can also be seen in Table 25 where variations in
air exchange rates associated with seasonal changes are shown. One can
conclude that during the colder winter months, when indoor/outdoor tempera-
ture differences are relatively high, there is a corresponding increase
ir infiltration rates within a particular structure. In comparison, low
temperature differentials durinq summer seasons are associated with lower
air exchange rates.
-65-
-------
Washington Experimental
2 August 76
Time 1345
f—
~
¦
•
4
¦
Fa
Wil
c
n on
idows
pen
¦
•
A
¦
•
•
4
¦
•
4|
—1
1
1
\
1
1
1
J
1
—
•
41
—
"1—
r—
1
1
1
¦
-
1
1
1
1
1
"1
1
1
1
1
1
•
~
¦
Ki
Bei
Livin
! 11J
. ** »r? t>0 1
n
1
10"
lO"9-
10 20 30 40 50 60 70 80 90 100 110 120 130 140
TIME, mirt
Figure 17. Instantaneous release.
-66-
-------
TABLE 24. EXPERIMENTALLY DETERMINED AIR EXCHANGE RATES
Residence Air Changes Per Hour
Washington Experimental (s)*
0.6
Washington Experimental (w)
0. 5-1.0
W ashington Conventional (s)
0.6-0.8
Washington Conventional (w)
0. 2-0. 4
Baltimore Experimental (s)
0. 5-0. 7
Baltimore Experimental (w)
0.8-1.2
Baltimore Conventional (s)
0. 6-0. 8
Baltimore Conventional (w)
0. 9-2. 0
Denver Conventional
0.8-1.0
Chicago Conventional (s)
0. 6-0. 8
Chicago Conventional (w)
0. 8-1.0
Chicago Experimental (s)
0. 1
Chicago Experimental (w)
0. 2-0. 3
Pittsburgh Mobile 1
0.4-1.0
P ittsburgh Mobile 2
0. 3-1. 1
Pittsburgh Low-Rise 1
0. 3-0. 8
Pittsburgh Low-Rise 2
0.7-1.4
Pittsburgh Low-Rise 3
1.6-1. 7
Pittsburgh High-Rise 1
0.9-1.4
Pittsburgh High-Rise 2
0.9-1.4
Pittsburgh High-Rise 3
0.9-1.2
* (i) = summer
(w) = winter
-67-
-------
Washington Experimental August 2, 1976 Time 1110
10-
9
•
1 C
Kitchen
Bedroom
a—
~
7—-
¦
Living
Roon,
6—
5—
4—
3—
2—
io-8_
•
i
>
<
~
o—
—•—
w
t
~
T
•
7—
6—
5—
*
ll
A
¦
A
it
4—
A
¦
1
3—
Al
1
2-
m
10"9^
1
~
~
i
Sou
Shut
fee
-Off
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
TIME, min
Figure 13. Continuous release.
-68-
-------
HVAC - Normal Operation
(Excludes High-Rise Structure)
1.6
1.4 "
1.2 "
i
A
<©
I
bJ
CS
z
-------
TABLE 25. SEASONAL VARIATION IN AIR EXCHANGE RATES
Residence
Air Changes Per Hour
Summer
W inter
Washington Experimental
0. 6
0. 5-1. 0
Washington Conventional
0. 2-0. 4
0.6-0.8
Baltimore Experimental
0. 5-0. 7
0. 8-1. 2
Baltimore Conventional
0. 6-0. 8
0. 9-2. 0
Chicago Experimental
0. 1
0. 2-0. 3
Chicago Conventional
0. 6-0. 8
0. 8-1. 0
A parameter that exerts an effect on the air exchange rate of a struc-
ture is the forced infiltration associated with the operation of the heat-
ing, ventilating, and air conditioning (HVAC) system. To show the relative
effect of HVAC operation, several tracer experiments were performed with
the air blower recirculation fan in a continuous operating mode and in
a normal mode as determined by the thermostat setting. Figure 20 shows
a tracer experiment where the blower fan was turned on to continuous opera-
tion after several hours of normal HVAC operation. The forced circulation
of indoor air increases the infiltration of outdoor and a corresponding
increase in the number of air changes per hour is measured. Table 26
shows a similar relationship for the experiments performed in other resi-
dences.
TABLE 26. EFFECT OF HVAC OPERATION
Air Changes Per Hour
Normal HVAC
Continuous HVAC
Residence
Operation
Operation
Washington Conventional
0. 2-0. 3
0.4
Baltimore Experimental
0. 5-0. 6
0.7
Chicago Experimental
0. 2
0.3
Pittsburgh Mobile 2
0.3-0. 5
0.8-1.1
Pittsburgh High-Rise 1
0.8
1. 1-1.3
Pittsburgh High-Rise 2
0.9-1.1
1.1-1.4
Pittsburgh High-Rise 3
0. 9-1. 0
1. 2
-70-
-------
PITTSBURGH MOBILE 2
MARCH 10, 1977
TIME 1200
TIME, min
Figure 20. Effect of HVAC operation on residential air exchange rates.
-71-
-------
Notice that the HVAC operatinq mode has a areater effect on the air
exchange rates of the mobile home and apartments than on the sinqle family
structure. The volume of air in the mobile home and apartments is smaller
than sinqle family dwellings. Thus, under forced circulation the infiltra-
tion of outside air has less conditioned inside air to mix with. This
creates an increased dilution rate of the tracer concentration and higher
air exchange rates are evident.
During periods of SFg tracer experiments the occupants' activities,
such as door and window openings, were recorded. The effect of human
activity was difficult to quantify with respect to measured air exchange
rates. However, there is evidence, as shown in Figure 21, that the
occupants' activity has a noticeable effect and can be verified by data
from the tracer studies. This figure shows a continuous release tracer
experiment performed on a semidetached residence. During this experiment,
two door openings occurred within the residence at different locations.
Investigation of this illustration shows that the tracer concentrations
reached an equilibrium level in about 60 min. The first door openinq,
shortly after equilibrium was obtained, occurred in the basement and
remained open throughout the test. The opening disrupted the equilibrium
and concentrations began to fluctuate for approximately an hour before
equilibrium was again obtained. Next, a door was opened in the living
room, this time for only a short duration, and again the equilibrium was
upset. The experiment was ended before a new equilibrium was reached.
This example shows the effect that occupant activity can have on the air
flow patterns within a residence.
An attempt to investigate the effects of wind velocity and direction
on air exchanae rates proved to be a difficult task. In several instances
these parameters displayed suggested relationships, but in an equal number
of cases no relationships were shown. The short duration of the experi-
ment (3-5 h) may be one reason why significant relationships could not
be investigated. Additionally, the obstacles and barriers surrounding
each structure varied greatly from site to site. This variation made
representative comparisons impossible.
The results of these tracer studies show that the air exchange rates
of residential structures are affected by many different parameters. When
experimenting under real-life conditions it becomes a difficult task to
separate the effects amonq these parameters. This discussion has
attempted to quantify several of the most important factors.
In conclusion, it is observed that environmental factors play an
important role in determining average residential air exchange rates. Of
particular Importance are the effects from indoor/outdoor temperature
differences. This does not exclude other parameters such as wind direction
and relative humidity from having some effect on air exchange rates, but
these factors are not evident in this investigation.
-72-
-------
Baltimore Conventional
February 2, 1977
Time 0918
1
1
¦
1
•
1
1
•
w
1
1
•
•
1
1•
•
1
(l
|
(
1
1
! ¦
¦
¦
¦
|«
1
1
¦
l
I
1
1
•
¦
¦
¦
1
1
1
1
1
1
1
}
•
~
1
l
1
1
1
Y
~
~
w
1
1
1
1
1
i-
~
¦r
1
1
¦
V
*
1
i
1
I
~
1
|
1
I
1
1
•
—
1
1
1
1
1
1
*
1
1
*
I
1
aseme
Door
Open
at
F
<:
1
ront D
Oper
1 Minu
3or
i
tes)
•
¦
*
Kite
ivlng
Baser
hen
Room
nent
15
30 45
60
75
90 105 120 135 150 165 180 195 210
TIME, min
Figure 21. Effect from occupant activity.
-73-
-------
The high-rise apartment structures react differently than low-rise
or single family structures. This is particularly true for the effects
of HVAC operation where the volumes of the hiqh-rise units are low and
display high air exchange rates during HVAC operation. The operation
of the HVAC systems has significant effects on the air exchange rates.
Studies on the impact of temperature differences in high-rise apartment
units were inconclusive.
This study has shown that a broad range of air exchange rates exists
for various types of residential structures. The variation in measured
air exchange rates during this study ranged from 0.1 to 2.0 air changes
per hour. Much of the variation was found to be dependent upon several
variables, making the prediction of structure's exact air exchange rate
a difficult task. However, evidence indicates that residential structures
with similar characteristics have similar rates and that estimates can be
made to determine a structure's "representative" seasonal air exchange
rat e.
-74-
-------
SECTION 3
NUMERICAL MODELS
THE GEOMET INDOOR-OUTDOOR AIR POLLUTION (GIOAP) MODEL
Introduction
Numerical simulation is considered among the most practical tools in
estimating indoor air pollutant levels as a function of the outdoor pollutant
levels plus other parameters. While its potential use has been realized by
many scientists in the field, Its application has been rather restricted,
because the indoor-outdoor numerical models require inputs not readily avail-
able to the researcher. The extensive field program of the EPA-GEOMET project
on the "Indoor Air Pollution Assessment Control and Health Effects" has pro-
vided a large data base. Using this Information we have formulated the GEOMET
Indoor-Outdoor Air Pollution (GIOAP) model.
The objective of the GIOAP model is to predict the indoor air pollution
levels by simulating a series of complex interactions involving outdoor
pollutant levels, structural characteristics of the examined dwellings,
and behavioral patterns of the inhabitants.
The motivation for the formulation and application of the numerical model
arises from the scientific recognition that measures to conserve energy within
buildings, through the introduction of new energy transfer systems and the
reduction of building ventilation rates, will result 1n changes 1n the indoor
air quality characteristics. These changes may affect air quality either
adversely or beneficially. Simulation of a large variety of indoor conditions
will quantify these effects. In addition, the validated GIOAP model can be
coordinated with an epidemiological study to determine the health effects of
Indoor air pollution.
The approach followed 1n the generation of the GEOMET model is a two-
step procedure: 1) mathematical formulation, and 2) model validation.
Each of these steps will be discussed 1n detail 1n the balance of this
section.
-75-
-------
The impact of a validated indoor-outdoor air pollution model will be
substantial in identifying the optimum scenario that meets the national
policy toward energy conservation measures in residential buildings without
endangering the health of the segment of the population that spends a
large portion of its time in the indoor residential environment. The last
subsection discusses the conclusions and impact of the GIOAP model.
General Principles
The numerical simulation model formulated by GEOMET for assessing air
quality in the indoor residential environment follows the general principles
of a mass balance equation. In the sense that the GIOAP model specifically
addresses residential environments (detached dwellings, row buildings, mobile
homes, and apartments), it is different from the general models which include
terms that are not applicable 1n the nonworkplace environment.
Air pollution in an enclosure may be of either outdoor or Indoor origin,
or both. If of outdoor origin, it enters through infiltration and ventila-
tion. If of indoor origin, it 1s generated from pollutant sources within the
enclosure. Regardless of source, air pollutants diffuse in the enclosure.
They are removed over varying periods of time by exfiltration and ventilation
to the outdoors and/or through indoor decay processes.
A1r Infiltration is defined as the change of air within a structure
without the interference of the inhabitants. Thus, the ambient air entering
an enclosure through cracks 1n its walls 1s infiltrated air which, whether
clean or contaminated, influences the indoor pollution levels. Ambient pol-
lutants may also be Introduced indoors through the ventilation process, which
may be defined as air changes induced by the occupants of an enclosure; this
can be natural ventilation through closing and opening of doors and windows,
or can be forced ventilation through the operation of attic fans, air condi-
tioning or heating systems. A1r exflltration 1s the opposite of air infil-
tration, 1n which Indoor air leaves an enclosure through structural cracks.
-76-
-------
Exhaust ventilation moves air from indoors to outdoors through the vents of
a forced circulation system as well as through door and window openings.
Pollutants are introduced into the indoor environment by means of sources
such as fireplaces, stoves, smoking, and cleaning devices. Finally, pol-
lutants may be removed from the indoor air environment through indoor decay
processes such as chemical transformation, settling, and absorption and adsorp-
tion by walls and furnishings (collectively termed pollutant sinks), and by
filtering procedures in the makeup air or in the recirculated air.
To characterize time-dependent aspects of pollutant behavior, it is useful
to deal with rates of change of air pollution within enclosures, rather than
simply with absolute amounts of pollution. Mass balance principles require
that the rate of change of an air pollutant quantity within an enclosure
equals the sum of the rates of all pollutant Introduction and removal processes
that operate upon the enclosure.
Mathematical Formulation
The GIOAP model illustrates the above general principles by specifically
simulating procedures present in residential environments. The GIOAP
model is based upon the- following mass balance equation:
v -ar ¦ VvCout + s - VvCm - VDCm (1!
where
C^n * the indoor pollutant concentration, mass/volume
CQut * the outdoor pollutant concentration, mass/volume
V s the volume of the building, volume
v ¦ the air exchange rate of the building, air exchange/time
S ¦ the indoor source strength rate (rate of Indoor pollutant
emission), mass/time
D » the decay factor, t1me"\
-77-
-------
The term on the left-hand side of Equation (1) denotes the rate of change
of the indoor pollutant mass. The first two terms on the right-hand side of
the equation represent the rate by which the pollutant 1s introduced indoors, by
infiltration of air (VvCQut) from outdoors, and by indoor pollutant generation
due to indoor sources (S). The last two terms represent the pollutant removal
rate due to exfiltration of indoor air (VvC^n), and due to indoor sinks, such
as decay processes (VDCin). The factor v, the air exchange rate, appearing
in the first and third terms of the right-hand side in Equation (1) is a total
rate; it is the sum of the infiltration, exhaust, and ventilation rates.
A series of approximations are necessary before the GIOAP model can be
applied to estimate the indoor air pollution levels. Over short time inter-
vals simulated, the outdoor pollutant concentration is approximated by a
straight line (an approach used by Shair and Heitner (1974)). The line is
given by:
Cout = "out1 + bout ^
wh^re mQut is the slope, t is the time, and bQUt is the y-intercept. It is
further assumed that the parameters v, S, and D are constant during the time
interval that is being modeled.
Thus, the model equation becomes:
dC1n
V -ar " Vv • VvC1n + s - V0C1n tolt
-------
The estimations of the indoor pollutant concentrations are obtained
from the solution to the initial value problem described by Equation (3).
The closed-form solution to Equation (3) is given below:
C (CI - fr n, /'It /I Wh + S 'Wil
C1n(t' " cm0 _ "out (b+7) 0 " (ffhTj (ubout + V ' T+77| 6
+ (&) (ubout + f - Tpf) + "out (t&) 1 toititf • <4>
This is the equation used for numerical simulations of indoor air pollution
levels.
The subject of relating the outdoor and indoor levels is relatively
recent, and research emphasis has been placed on field measurements of con-
taminant levels rather than the development of numerical models. Several
scientists have attempted to formulate models employing relationships
similar to the one expressed by Equation (3); Milly (1953), Calder (1957),
Turk (1963), Hunt et al. (1971), Shair and Heitner (1974), and others have
used more or less complex versions of the equation. The GIOAP model 1s
specifically designed to simulate residential conditions. The assumptions
concerning elements of the right-hand side of Equation (3) (S, v, and D
are constant over the time period being modeled, and Cout can be approxi-
mated by a straight line over the time period being modeled) are con-
sistent with those made for other models. Previous studies have emphasized
and simulated steady-state conditions; however, the GIOAP approach Includes
the transient portion of the solution to Equation (3). As a result, 1t 1s
possible to model both short- and long-term Intervals.
When Equation (4) 1s used 1n this study to model indoor air pollutant
concentration levels, several additional assumptions are required. These
assumptions are as follows:
1. The air exchange rate v remains constant for at least
one hour.
-79-
-------
2. The internal pollutant source rate S remains constant for
at least one hour.
3. The indoor pollutant removal procedure is modeled as a first-
order decay term with the decay factor 0 = ln2/t^, where %
is the half-life of the pollutant considered. For stable
pollutants with long half-lives D is approximated by zero.
A list of the decay factors used in this study is given in
Table 27.
The GIOAP model is capable of simulating any time interval, because
the principles involved do not constrain this aspect. However, the time
unit of the generally available ambient pollution data has led us to
specify one hour as the time resolution of the model. Finally, one hour is
the smallest interval that is appropriate for comparison to NAAQS.
TABLE 27. DECAY FACTORS (PER HOUR) USED IN THE
GEO MET INDOOR AIR POLLUTION STUDY
Pollutant
Decay factor (per hour)
CO
0.00
so2
1.04
NO
0.00
no2
1.39
°3
34.66
CH*
0.00
THC
0.00
co2
0.00
THC-CH4
0.00
In order to model the indoor air pollutant behavior over the time inter
val [tg, tf], the interval must be decomposed into the set of subintervals
-80-
-------
(*0» • •• ^ wliere *n a *f» and ti - tj.'j a 1 h. As a result,
Equation (3) becomes
<, • h-. ¦ •, ft) v, • ft) 4-^)|
' ' t - * "i (t^) li (5)
where
C4 ¦ the Indoor pollutant concentration level at time t1t
i * 1 n
CoutU) * the outdoor pollutant concentration level at time t
m1 * ^out^V " ^out^i-l^^^i " *1-1^* * a n
bi 58 Cout(ti) - m1t1. 1 » 1, .... n
S4 s internal source rate over the interval (t. ,, t.],
1 1=1, n 1-1 1
Vj * air exchange rate over the interval (t^-j, t^]»
i * 1, ..., n
V » volume of the building
Di * decay factor over the Interval (t^-j. t^3.
1 ¦ 1, ..., n.
MODEL VALIDATION
Introduction
The problem of estimating indoor air pollution levels involves both
physical and behavioral parameters; the outdoor levels vary as a function
of the local meteorology and other factors, while the Indoor source strengths
(Indoor pollutant generation rates) and air exchange rates depend on the
meteorology and the activity of the occupants. The combination of all Inputs
results 1n complex conditions which are either rarely repeated or very
expensive to duplicate 1n the laboratory. Numerical models enable scien-
tists to simulate these complex conditions, to stage specific Incidents,
and most importantly to estimate values for the indoor pollutant concentra-
tions.
-81-
-------
Two essential stages determine the predictive capability of the GIOAP
model:
1. Initial model validity. Do the predicted concentrations
reflect observed data?
2. Model sensitivity. How do the predicted concentrations change
in relation to changes in input parameter values?
An intrinsic element of these two stages is the ability to demonstrate the
validity of the model using "best" estimates of the input model parameters.
We have formulated a parameter estimation procedure that enables us to esti-
mate the values of indoor source strengths and air exchange rates from the
raw outdoor and indoor pollutant data. Recurring modes of pollution behavior,
episodes, are of extreme Importance in the EPA-GEOMET indoor air pollution
project. Numerically, a new episode is defined each time a new initial con-
dition is introduced. It is our objective to associate each episode with
stratified levels of indoor activity so that in the future we can estimate
the indoor pollutant concentrations from outdoor pollution concentrations
and Indoor activity levels. The validity of these estimates will be studied
1n the balance of this section.
A three-step evaluation design will be followed in the assessment of
the GIOAP model: we begin by estimating "best" values for input parameters,
continue with statistical validation studies which include tables and input/
output graphs, and conclude with a section on parameter sensitivity. This
design takes advantage of the extensive data base available to this project.
Parameter Estimation Procedure
In order to estimate the indoor pollution levels using the GIOAP numeri-
cal model, all parameters associated with the model must be given numerical
values. The iron1 toring data from the field studies of the indoor air pollu-
tion project constitutes a unique source of information for assigning numerical
values to the relevant parameters. Some parameter values, such as initial
Indoor concentration and the volume of the structure, are easily determined;
others, such as the air exchange rate v of the building investigated and the
internal pollutant generation source strength S, are more difficult to obtain.
-82-
-------
The validity of the GIOAP model obviously depends on the values given to
these difficult-to-quantify parameters. Since the model is to be validated
under "best" conditions, we must obtain the best possible values for v and S;
the methodology used to obtain these values is the Parameter Estimation Pro-
cedure described in this section.
Theoretical Approach--
Recall Equation (4):
c1n(t) * [c1nQ - "out (c£r) *0 " (l!Sr) (ubout + 7 * "D5v")] e ' °
* (wr) (vbout+1 - -§&-) * mout (tpt) 1 'o - 1 - lf ' <4>
Also, consider the following function of S and v:
'(S.m) - £ CCf(s.v) - CM,]2 (6)
1-1 ' 1
where
n » number of points
CM, ¦ 1th measured value; 1 ¦ 1, .... n
th
C4(S,v) ¦ 1 computed value via Equation (4) corresponding
1 to Cfy .
Tht^ fundamental problem in parameter estimation is to find value(s) of
v and S that minimize Equation (6). Two points must be made: 1) 1n the
case of Indoor air pollution studies, the parameters v and S are constrained
to lie v/1th1n certain Intervals specified by the nature of the Investigated
dwelling and the particular pollutant and source examined; 2) Equation (4)
1s not linear 1n v. These two points combine to make the problem of esti-
mating values for v and S difficult.
-83-
-------
A parameter estimation technique appropriate for the present problem is
the grid search method. Although this process is easily constrained, it is
lengthy, and its accuracy strongly depends on the number of grid points.
Because f(S,v) is a function of two variables, it is not possible to obtain
the required minimum by the technique used for finding extreme values of
functions of one variable taught to all students of beginning calculus.
This technique is easily constrained over a given interval, but it cannot be
applied to f(S,v) because it minimizes only functions of one variable. How-
ever, the Parameter Estimation Procedure used in this study is a cross
between the grid search and the optimization technique for a function of one
variable. A sequence of five steps must be followed in the Parameter Esti-
mation Procedure.
1. Define the constraints on v and S; i.e., v-j < v < vy and
Sl < S < Su, where subscripts 1 and u define the lower and
upper values of the parameter intervals.
2. Determine iv » (vu - v-.)/k, where k is the number of
Intervals.
3. For a given = v-j + (i-1) av, i s 1, ..., k+1, find
the points for which
dfiss> -)- * 0 •
(S,v) * (Sq^ »vi)
4. Determine as the value of S which gives
nrfn {f(Sl,v1)# f(SQ ,vf), fp(Su,vi)}
5. The required estimates of v and S are the values
that give
m1n tf(Sm »v^):i * 1t ..., k}.
The theoretical approach used in deriving this Parameter Estimation
Procedure is provided below. Owing to step 3 of the above sequence, all
-84-
-------
equations below refer to a given fixed value of the parameter v. Equation
(4) can be rewritten as follows:
-(v+D)(t-t0)
1 - e 11 s
V(D+v)J 3
+ "o-™(i&r) *0 - (wb) - ^r)[
(est) (vb - 15&) + m (557) 1
-(v+D)(t-to)
wlrjre
c0 =
m =
Cin(t)
cinrk
m,
b = b
out
out
or as
C ¦ aS + 8
where
1 - e
-(v+D)(t-t0)
V(D+v}
and
9 * [°o ¦ m (osr) *o • (sir) (vb - c£r)]
-(\>+0)(t-tg)
(b+c) fb ¦ b*t) + ¦ (bk) 4 •
-85-
-------
From Equation (6) we have
1=1
where
n ?
f(S,v) » £ (c^S + 6i - CMi) (9)
a,j 1s computed ? t s t| and tg 5 t^j
3i 1s computed @ t 3 t^ and tQ s t^ and CQ a CM^ -j .
Equation (9) becomes
f(S,v) a S2 53 ®4 + 2S 0,(84 - CM,.) +¦ 51 (si - CM.)2
1-1 1 i»1 1 1 1 i»l 1 1
f(S,v) = AS2 + 2BS + C (10)
where
n 9
A - L a
1 = 1 1
B » E o1 (64 - CM.)
1»1 1 1 1
C • £ (B, - CM.)2 .
i»l 1 1
Equation (10) Indicates that f(S,v) is parabolic in S. The critical point
is found by taking the derivative of f with respect to S:
$ - 2AS + 2B . (11)
-86-
-------
Setting df/dS = 0, it is seen that - (B/A) is a critical point; in order to
determine whether - (B/A) is a maximum or a minimum, the second derivative
of f with respect to S is taken:
= 2A
dS
(12)
Since A is a sum of squares, A > 0, which means that
thus S 3 - (B/A) minimizes f(S,v).
Completion of the square in Equation (10) leads to the same conclusion:
Again, since A > 0, the vertex S » - (B/A) is the desired minimum.
The parameter S, representing the indoor pollutant generation rate, must
be constrained to He 1n a physically meaningful Interval [S], 5U]. Thus,
while S ¦ - (B/A) 1s an absolute minimum, - (B/A) may not be within the
interval [S-j, Su]. Hence, 1n addition to computing - (B/A), a test must be
made to determine 1f - (B/A) lies in the Interval (i.e.. if - (B/A) e [S-j,
¦Su])» not the end P°ints of the interval must be examined. The three cases
to be considered are Illustrated 1n Figure 22. They are:
Case 1: < S,
Case 2: St < « Su
Case 3: Su < Sm3.
(13)
-87-
-------
Figure 22. Graphical illustration for the three cases of constraints on interval source rate.
-------
In Case 1, the constrained minimum occurs at S-j; in Case 2, the constrained
minimum occurs at $m ; and in Case 3, the constrained minimum occurs at Su.
Application of the Parameter Estimation Procedure to the Indoor Air
Pollution Data Base-
In order to apply the Parameter Estimation Procedure to the indoor
air pollution data base, the problem of degrees of freedom must be considered;
i.e., the fact that the number of observations used to estimate a set of
parameters must be greater than the number of parameters being estimated.
Thus, as a means of increasing the amount of data available for use in the
Parameter Estimation Procedure, the instantaneous outdoor values and
instantaneous indoor averages for each 20-min segment of each hour were used
because at this time the entire house is being modeled instead of individual
zones.
Hourly values of v and S are desired; however, there is a problem in
applying the Parameter Estimation Procedure to obtain hourly estimates of
v and S. The problem involves degrees of freedom. Even though three
values are available for each hour to be used in the estimation, one of
these is the initial value, which means that only two values can actually
be used in the estimation procedure.' As a result, the number of parameters
to be estimated equals the number of points to be used, which means that no
degrees of freedom are left for the estimation procedure.
The problem mentioned in the previous paragraph 1s resolved as follows:
2-h estimates of v are calculated from the data of a "nonreactive pol-
lutant;" then, using these estimates of v, hourly estimates of S are found
for all pollutants. The "nonreactive" pollutant chosen was CO. If any of
the 12 CO values (6 indoor and 6 outdoor over a 2-h period) are missing,
v is estimated using NO data; however, if both CO and NO data are missing
for a given 2-h period, neither v nor any of the pollutant source rates
for that period are computed. Finally, if, for a given pollutant for a
given hour, any values are missing, the corresponding source rate is not
computed.
Essentially, we are using CO as a tracer to estimate theoretical
values of air exchange rates for each of the Investigated dwellings. As
part of the field monitoring program of the indoor air pollution project,
-89-
-------
the air exchange rate of each residence is determined experimentally. The
tracer used for the experimental determination of the air exchange rate is
SF6; the monitoring protocol calls for three or four different 4-h experi-
ments per residence. The theoretical and experimental values for the air
exchange rate agree in most of the investigated periods, and, in those
cases of disagreement, the difference between the two values is not
appreciable. Table 28 shows a comparison between the estimated and experi-
mental values for the air exchange rates.
The estimated indoor source strength value S in mg/h is an "effective"
pollutant production rate; i.e., the internal source is treated as though
it operates for the entire 1-h period. The estimated indoor source is
also comprehensive; that is, if two indoor sources are generating a pol-
lutant simultaneously, the calculated theoretical value will be the sum
of the individual source strengths. Using the daily logs kept by the
occupants of the residences we monitored, we are often able to isolate a
single source; the estimated theoretical values due to isolated indoor
sources compare favorably with the available literature values.
TABLE 28. AIR EXCHANGE RATES
Air exchange ratea
AmMmm
1 Experimental
Chicago experimental I
0. 40
0 JO
0. 20
0.23
0. 22
0. 20
Pittsburgh low-riie apt I
0. 64
0. 58
0. SO
0.40
0. 34
0.63
Pitt*burgh mobile home I
0. ?S
o. u
I . OS
0. S3
Denver conventional
1. 10
1.02
Washington experimental I
0. 10
0.60
Washington conventional I
0.6
at
0.4
0.24
0.2
a 41
Baltimore conventional I
1.2
0. 7%
Baltimore experimental I
0. M
0.72
-90-
-------
Statistical Studies
The objective of the statistical studies performed on the 6I0AP model
is to define its ability to estimate indoor air pollution levels. In this
document we evaluate the results from a sample of eight residences; this is
a comprehensive review of the predictive power of the model.
In the first section we describe the procedure followed and outline
the motivation and the objectives of each step. The section on the statis-
tical assessment includes response graphs, statistical tables, and scatter
diagrams. In addition, it contains comments and conclusions on the model
for each gaseous pollutant monitored. The final section provides theoreti-
cal details on sensitivity coefficients and includes a series of simulations
that illustrate the errors introduced by not using the "best" estimated
value for any given Input parameter.
Statistical Procedure and Methodology--
The strength of a theoretical or numerical method to predict air pollu-
tion levels has often been demonstrated with graphical illustrations. This
has been the case In many studies with a small data base. Figure 23 pre-
sents a sample of Indoor values estimated by the GIOAP model against the
observed indoor values for a number of pollutants during a 2-week monitoring
period. While this randomly chosen set of illustrations Indicates the
predictive power of the model, it does not allow for general conclusions.
In order to validate the GIOAP model, a statistical approach is required.
A statistical analysis 1s performed on data sets consisting of pairs
of hourly estimated and observed indoor air pollutant concentrations. A
statistical approach is preferred to judgments made on the basis of compar-
ing corresponding estimated and observed indoor air pollutant concentrations
for two reasons. First, the GIOAP model simulates a variety of conditions for
a number of pollutants and generates a large number of data sets, each of
which contains many points (over 300). Second, direct comparison of estimated
and observed values Involves difficult judgments 1n deciding when the estimated
-91-
-------
* Wniuii Hourly Moa
« Cotaci*
Figure 23. Estimated vs. observed pollutant concentrations for 7 consecutive days.
-92-
-------
Figure 24. Estimated v». observed pollutant concentrations lor 7 consecutive days.
-------
value falls within an acceptable range of the observed value. Under such con-
ditions statistical techniques provide fast, efficient, reliable methods for
assessing the data, and make the judgements much less subjective.
The statistical analysis used to validate the GIOAP model is based on
the principle that If the model simulates realistically the involved complex
conditions, a plot of the estimated versus observed values would fall on or
near a line with a slope of one and intercept of zero.
The problem of measuring the association between the observed and esti-
mated values is divided in the following three sequential steps:
1. The degree of linearity in the relationship between the
pairs of estimated and observed values is established.
Linearity is required to proceed to step 2.
2. The slope and Intercept of the line expressing the linear
relationship between the observed and estimated values
are determined. Criteria on the proximity of the slope
and intercept to one and zero, respectively, are estab-
lished and must be met before proceeding to step 3.
3. The degree of dispersion about the line defined by the
slope and intercept in step 2 is calculated and must meet
certain acceptability criteria.
The statistical methods used 1n each of the three steps will be discussed in
the remainder of this subsection.
The Pearson product-moment correlation coefficient is calculated in
order to determine whether or not a linear relationship exists between the
pairs of estimated and observed values. The calculated correlation coeffi-
cient must be close to +1; if that is not the case, it should be concluded
that the numerical model does not realistically simulate the processes
involved.
The second step requires that the relationship be linear. A line is
characterized by calculating the regression parameters (slope and intercept)
of the plot of observed versus estimated values. In addition, the calculated
slope and intercept values are tested for statistical significance. Each of
-94-
-------
the following hypotheses is statistically tested by a two-tailed t-test
1) the slope is +1, and 2) the intercept is 0. In the following section
it will become apparent that this statistical procedure will reject
numerical estimations well within the accuracy limits of the input values.
Thus, it is necessary to ease the limits that strict adherence requires
by establishing a set of less restrictive criteria.
The final step is necessary only if a linear relationship exists
between the estimated and observed values, and if the regression line
meets the established criteria. This step determines how well the line
fits the data points. In order to estimate the dispersion of the data
points from the regression line, the Standard Error of Estimate (SEE) is
calculated. If the SEE is small, the model data set is acceptable; if
the SEE is large, it indicates that the points are widely scattered about
the regression line and that the modeled set should be rejected.
If a data set meets all these criteria, then it is concluded that
the GIOAP model adequately represents the simulated event.
Statistical Assessment of the GIOAP Model--
The statistical procedure outlined 1n the last section will be applied
to eight sets of data corresponding to continuous monitoring from eight
dwellings, each set consisting of seven gaseous pollutants. The model pre-
dicts the average Indoor pollutant concentration for three different time
periods (episodes), 3 h, 8 h, and 24 h. The model performance is evalu-
ated for all days of the monitoring period.
The statistical information generated for each pollutant, each episode,
and each residence 1s presented in tabular form (for example, see Table 29).
The first column Identifies the residence Investigated, the column labeled
tepis Indicates the duration of each episode. The column labeled r contains
the correlation coefficient; b 1s the Intercept of the regression line cor-
responding to the plot observed versus estimated values, and m 1s the slope
of this line. The null hypothesis that the Intercept Is zero 1s tested
against the statistic t^, while the null hypothesis that the slope equals
-95-
-------
TABLE 29. STATISTICAL DATA SUMMARY
Residence
'epli
r
b.
m
'b
t
in
Rang*
of Indoor
Observ. Value
SEE
No.
Of
Obterv.
Comment!
3-hr.
8-hr.
24-far.
3 far.
8-hr.
24-hr.
3-hr.
*-hr.
24-hr.
3-hr.
8 hr.
24-hr.
3-hr.
8-hr.
24-hr.
3-hr.
8-hr.
24-hr.
3-hr.
8-hr.
24-hr.
3-hr.
8-hr.
24-hr.
-------
one is tested against the statistic tm. The column to the right of tm con-
3 m * m
tains the range of the average pollutant concentration observed Indoors;
this range provides a value against which the calculated standard error of
estimate can be judged. The next to the last column presents the number of
observations which are used for estimating the various statistics; it must
be noted that this number is not always the total number of possible pairs
because of either missing observed values and/or missing calculated values
due to the lack of the initial value, the air exchange rate, and/or the
effective source strength.
Based on the Information Included in these columns, a conclusion is
reached on how well the model predicts the indoor observed values. Three
classes of acceptance or rejection comments are generated: Class I describes
the simulations that satisfy all predetermined tests; Class II refers to
numerical estimation of the Indoor pollutant concentrations which, although
not statistically acceptable, are judged to meet predetermined criteria which
will be described below; and Class III refers to model estimations that do
not meet any of the above requirements and must be rejected because they do
not realistically simulate the observed Indoor values.
The investigation of the model performance for each pollutant will be
presented in the balance of this section. However, 1t is essential to begin
by stating a set of rules for each of the classes outlined 1n the previous
paragraph. Since there are varying degrees of linearity, a decision must be
made on the cutoff level of the correlation coefficient. Figures 25 through 28
are scatter diagrams with four different correlation coefficients; their
values are r = 0.96, r = 0.82, r = 0.72, and r = 0.66, respectively. The
cutoff value chosen for this study for the correlation coefficient is r -
0.7; thus, if r 1s below 0.7, the relationship between the observed and esti-
mated value 1s considered nonlinear for the purposes of this study.
For Class I acceptance the criteria on b and m are set by the two-tailed
statistical t-test. For a significance level of 0.01, the t value tQ QQ5*00 a
2.576 (the number of degrees of freedom 1s considered to be Infinite because
1n this project it is almost always larger than the maximum finite degrees of
-97-
-------
toon VALIDATION -
8-HR CO
OB/I 2/77
PACE
FILE P613S (CREATION OATE • 08/17/77) I
scArreftGR** or toomn oas oasExvco pqli.
1.00 3.00 5.00
nou tip POLLUTION MOLiEL VALID*
AMT CnhCENTHATION (ACPOSS) ES
.00 9.00 11.00 11
20.00
19.00
lfc.00
14.00
12.00
10.00
It. 00
6.00
4.00
2.00
0.0
«
• * • *
« 2*«
• 2 2 *
0 *3
» *235
* *« )»M
~ • ••)•
• )»(«« • ••
• * 4265 *»
• *2449444 M
•232*942
* 2695342 •
•9955
•394*
*4 ~
*
on
0
• •
EST MATCH POLLUTANT "ONCCNTH
is.no ll.oo 19.00
>.0
2.oo
4.00
6.00
# .00
10.00
12.00
14.00
16.00
18.00
~ .
20.00
Figure 25. Scatter diagram with r = 0.96.
-------
NOOM* - tUSHINATIi* Ca«VCNTlrttiAI. t O-HU.
(1/ / IIH / IK
fkUh /
rii,«c puis
SCtrTMiGllftM
40.AO
J*.00
32.00
20.00
24.00
20.10
16.00
12.00
0.00
4.00
A.O
(ZRF.ftflDN l>»TF. * 02/00/18) 1000011 MP PDM.UTION MOnF.I. VAI.in*
F (DO**) OHS OBSFRVEO POI.L»T*MT CONCEUTRATIO* fACKOSS) KS
2.00 (.00 10.00 14.00 10.00 22.00. 16
1
I
I
I
I
I
I
I
(
I
I
I
I
I
1
I*
• 2
4 It
» 4 ~
••• 41* * •
4* 44 4 * * •
• 4 4 |4 2 4
444 * 4 *
4)444 4 4
344* 4 *4
*4 2 24 4 42 ~
4 41 4 2
4 42 I 4
*44 14
*23 • 14 ~
4 ~ 2*1
I
41
«!
I
*
I *
**
I ~
r
i
i
i
i
. —->4 —"t
0.0 4.<10 11.00 1?.«0 16.*0 70.0» V 4.<>A
ni
FSMH*TFn POLLUTANT CONCENTP AT Ifl
30.00 34.00 38.00
i ~ *¦.
9p.no i^.on 16.00 *o.on
Figure 26. Scatter diagram with r = 0.82.
-------
HflOeb VAiilOATION -
24-Hft TMC-CH4
Ott/16/77
PACK 2
PILE P6I IS tCfiF.A TION 0«f£ » OB/16/77) INftOOh AIR POLLUTION *Ol>H. VALIUA
SCArTEHGRAM OF
16.00
14.40
12.10
11.20
9.tO
o.oo
6.40
4.00
3.20
1.60
0.0
(OQHM) OftS
O.MO 2.40
OBSRBVfO P(I(
-------
hodec v»t.|i»*T|fHi - dkkkr riN»F»»rfn«i»L p-h« soj
07/0H/TI
PACK 7
rii.c p»i3^ ic»<5»Tin»t ofctc « np/a»/i»t jHoonn kin poj.wjticim «>orfi> m.trnt
SC*TTf»G«*"
25.!n
72.511
10.00
11.U1
15. 90
o
-¦ 12.50
10.10
7.S#
5.0»
2.50
rt.O
rnnwri) n«s orservfr pollh
t.?5 1.75 6.25
--4—--~•—4
? ~ ~
4JJ* » ~ »
»»«.« <1)41 1J»> M« »
ANT CONCFNTRATIOh (ACROSS* EST
.73 II.25 11.75 16.
2 ~
ON
FS1"I»"»TKIt POI-LIITAUT CnHCKNTBATTO
H.75 21.25 21.75
....t----+----4--~-4 -4
0.0 2.51 1.00 7.Sit in.Of, l>.50 IS. no 17.50 ?1.»0 ?2.5 0
Figure 28. Scatter diagram with r = 0.62.
75.00
22.50
70.00
17.50
15.00
1 2.50
10.00
7.50
5.00
7.50
0.0
-------
freedom specified in the statistical tables); thus if either t^ or t is
outside the interval -2.576 _< t <_ 2.576, the estimated indoor pollutant
concentrations do not fall into Class I.
For Class II, predetermined acceptability limits can be placed on the
slope m for all pollutants. However, limits on the intercept must be
determined on a case basis since the proximity to zero of the intercept
crucially depends on the range of the observed values which varies from pol-
lutant to pollutant. In Class II, slope values within the closed interval
[0.7, 1.3] are acceptable. In addition, values of the intercept that have
magnitude greater than 15% of the maximum observed pollutant concentration
are rejected. The last criterion for acceptance in Class II requires that
the Standard Error of Estimate (SEE) is less than or equal to 10% of the
maximum observed value.
When examining the tables containing the statistical validation data for
each pollutant (i.e., Tables 30 through 33 and 36 through 38), it will be seen
that the following phenomenon occurs several times: for a given set of con-
ditions the 3- and/or 8-h episodes will be Class II, but the corresponding
24-h episode will be Class I. Further examination will reveal that the
24-h episode statistical data is based on fewer observations the either
the 3- or 8-h episode data. The reason for this is that, when missing data
are encountered, the model calculations cease whether or not the end of the
episode has been reached. Thus, some episodes span fewer hours than indi-
cated by the headings, which results in less overall variation and gives
better statistical data.
In the balance of this section each pollutant will be examined indi-
vidually.
Carbon Monoxide (CO)--Table 30 provides all the statistical informa-
tion obtained by the previously outlined steps for CO. In all cases investi-
gated an acceptable degree of linearity (r >_ 0.821) exists, the intercept is
uniformly close to zero, and the slope is outside the prescribed interval
only once (m = 1.336 for the 24-h episode in the Pittsburgh low-rise apt. I);
-102-
-------
TABLE 30. STATISTICAL DATA SUMMARY FOR CARBON MONOXIDE (CO ppm)
Residence
lepi*
r
b
m
'b
t
m
Range
ofIudoor
Obcerv. Value
SEE
No.
of
Obcerv.
Comments
Pittsburgh
Mobile
Home 1
3-hr.
0.943
0.174
0.911
4. 779
-5.039
0.0-6.2
0. 310
337
II
8-hr.
a 900
0.242
0.874
4.810
-5.096
0.389
293
II
24-hr.
a 912
a 173
a 919
3.119
-2. 553
0.268
174
II
Denver
Conventional
3 l>.
a 970
0.041
0.980
0.744
-1.471
0.89-15.9
0. S70
320
I
8-far.
a 959
0.018
1.006
0.259
0.333
0.683
291
I
24-hr.
a 949
-0.024
0.993
-0.286
-a 285
0.549
168
I
Chicago
Experimental I
3-hr.
a 947
a 289
0.943
5.768
-3.004
0. 33-6. 33
0.352
282
11
8-hr.
a 888
0.683
0.862
9.350
-4.616
0.520
224
II
24-lw.
a 920
a 70
0.796
9.293
-6.890
0. 462
133
n
Pittsburgh
Low-Rite
Apt. 1
3-hr.
a 884
-0.029
1.128
-5.90
3.40
0.0-7. 0
0. 519
253
II
8 hr.
a 821
0.023
1.218
0.348
3.778
0.652
217
II
24-hr.
a 878
-0.105
1.336
-1. 378
5.143
0.571
126
III
Ba ttlmore
Experimental I
3-hr.
a 967
0.001
1.005
0.264
0.308
0.0-2.0
0.077
235
I
B-far.
0.969
0.004
1.096
0.709
5.062
0.075
213
II
24-hr.
a 960
-a 003
1.034
0.525
1.419
0.067
162
I
Washington
Conventional I
3-far.
a 961
0.040
0.972
1.222
-1.913
0.0-5. 2
0.396
360
I
8-hr.
<1938
-0.003
0.949
-0.068
-2.543
0.487
305
I
24-hr.
a 912
a 019
1.004
0.390
a 130
0.475
200
I
Baltimore
Conventional 1
3-hr.
a 851
0.107
O 778
3.683
-8.038
0.0-4.56
0.389
30S
11
8-hr.
a 876
0.036
0.882
1.178
-3.734
0.349
240
II
24-hr.
a 916
0. 034
a 894
1.235
-3. 363
0.271
154
II
Washington
3-hr.
a 930
0.080
0.942
3.224
-2. 740
a 0-4. 22
0.314
313
II
Experimental I
8-hr.
a 889
0.117
0.883
3. -594
-4.175
0.391
265
II
24-hr.
a 895
0.087
0.948
2.50
-1.427
0.358
169
n
-------
this is the only Class III case simulated. Let us investigate this Class III
case in more detail. The following points can be made:
1. Frequency distribution tables show that 91% of the
sampled values fall in the half-open interval [0, 2.8).
2. Straightforward calculations using the following rela-
tionship,
Observed Value = Intercept + (Slope) (Estimated Value),
indicate that the approximate maximum difference between
observed and estimated values is 22% of the observed
value.
A similar analysis applied to a randomly chosen Class II case gives an approx-
imate maximum difference between observed and estimated values of 2% of the
observed CO value. Analyses of this nature for a Class I case provide sim-
ilar or better results. It is concluded that the model predicts indoor values
acceptably.
Nitric Oxide (NO)--Table 31 illustrates a strong linear correlation
between observed and estimated values, r >_ 0.875. Similarly, the slopes of
the calculated regression lines lie within the predetermined interval. As a
first observation, the magnitude estimated for the intercept and the standard
estimates of error may seem large; however, after comparisons with the indi-
cated ranges of the monitored values, they are put in proper perspective and
are judged acceptable. Thus, 21 cases are accepted as Class II, while the
remaining 3 cases are accepted as Class I.
In order to provide a perspective on the model performance, an approxi-
mate estimate of the percent difference between observed and estimated values
will be calculated. The Denver conventional residence is chosen because it
was one of the extreme cases considered. Let us investigate the 8-h episode
simulations. The relevant statistics are r = 0.875, b = 10.135, and m =
0.793. Frequency distributions of observed indoor averages generated for
the data interpretation task of this study show that 94% of the hourly
values fall within the half-open interval [0, 120). Following the thinking
-104-
-------
TABLE 31. STATISTICAL DATA SUMMARY FOR NITRIC OXIDE (NO ppb)
Residence
*epls
*
b
m
'b
t
m
Range
of Indoor
Obterv. Value
SEE
No.
of
Obcerv.
Comments
Pittsburgh
3-1*.
0.970
6.873
0.962
2.377
-2.912
77-467
22.295
335
II
Mobile
8-hr.
0.941
S. 192
0.990
1.159
-0.480
29.690
274
I
flame 1
24-far.
0.917
12.541
0.931
2.213
-2.366
29.452
193
I
Denver
3 far.
0.909
8.307
0.798
4.304
-9.899
2.1-409
25.288
323
II
Conventional
8-hr.
0.875
10.135
0.793
4.072
-8.203
31.605
30S
II
24-far.
0.911
1.236
0.912
0.543
-2.550
18.175
147
I
Chicago
3-hr.
0.987
1.907
0.941
3.531
-6.753
0. 0-256
6.513
320
II
Experimental I
8-hr.
0.970
3.479
0.913
3.720
-6.003
10.281
251
II
24-hr.
0.976
4.733
0.836
4.642
-10.686
8.947
151
II
Pittsburgh
3-hr.
0.982
-1.135
1.034
-2.03
3.014
4.11-300.1
7.233
316
II
Low -R be
8 far.
0.982
-2.355
1.066
-3.525
5.101
7.805
258
II
Apt. I
24-far.
0.993
-2.235
1.048
-3.603
4.77S
5.717
155
II
Baltimore
3-far.
0.939
1.231
1.026
3.656
1.202
ft 0-87. 2
5.334
300
II
Experimental I
8-hr.
0.948
1.740
1.042
5.270
1.958
4,983
221
II
24-far.
0.946
2.402
1.039
6.239
1.596
5.327
221
II
Washington
3-far.
0.986
5.542
0.975
5.325
-2.619
18.6-279
9.101
303
II
Conventional I
8-hr.
0.966
15.105
0.919
7.784
-4.711
14.303
210
II
24-hr.
0.9S3
25.878
0.809
8.988
-7.957
15.482
117
II
Baltimore
3-hr.
0.906
2.314
0.841
2.328
-7.159
1.0-224.1
14.139
316
II
Conventional I
8-hr.
0.954
0.706
0.927
0.960
-3.969
9.235
252
II
24-hr.
0.961
0.207
0.931
0.256
-3.421
8.6S7
177
n
Washington
3-hr.
0.935
1.120
1.058
1.085
2.527
0. 0-283. 8
13.320
312
II
Experimental I
8-hr.
0.918
2.052
1.052
1.560
1.829
15.818
263
II
24-hr.
0.906
1.841
1.066
0.924
1.721
19.272
169
II
-------
expressed for CO, we conclude that for this case, within the specified
interval, the approximate maximum difference between observed and esti-
mated values is 15$ of the observed value; or within this interval the
statistical model value is at most 1.15 times the observed value.
This specific analysis is of course an example; however, we feel
that the model realistically simulates indoor average concentration for
NO.
Nitrogen Dioxide (NO2)—Numerical simulations of indoor NO2 concen-
trations require the use of a first-order decay term. The half-life used
:or these simulations is 30 min; this value is suggested by observations
of the indoor instantaneous NO2 values of this project as well as by C.
Hollowell (1977) (private communication) and Wade et al. (1975). The use
of first-order chemical decay terms, instead of the zero-order rate,
which was used originally, has substantially improved the predictive
power of the model. Six cases are rejected, 3 cases are accepted as
Class I, and 15 cases are accepted as Class II. The cases that have been
rejected are in residences without indoor NOg sources; the indoor concen-
trations are persistently low with very little variation. In cases like
this the model may overestimate the indoor NO2 concentrations by as much
as 50%. However, the model performs well for a total of 18 cases (out of
25) (see Table 32). The maximum difference between the statistical model
value and the observed value is 16%. This conclusion is reached by the
process described for CO, and it refers to a specific example; however,
the general assessment is that the GIOAP is realistically simulating
indoor average concentrations for NO2.
Sulfur Dioxide (SO2)—The nature of the SO2 data is a source of the
model's apparent Inability to estimate the observed Indoor values (see
Table 33). Table 34 shows portions of the Indoor average concentration
frequency distribution for SO2. Note that the Instrument used, a Meloy
A0-185-2A commercial detector, has a limit of detection of 0.005 ppm. It
1s apparent that almost all SO2 values observed Indoors are at or below
the instrument's lower limit of detection.
-106-
-------
TABLE 32. STATISTICAL DATA SUMMARY FOR NITROGEN DIOXIDE (N02 ppb)
Residence
*epb
r
b
m
-------
TABLE 33. STATISTICAL DATA SUMMARY FOR SULFUR DIOXIDE (S02 ppb)
Residence
*epi»
r
b
m
'b
t
m
Range
of Indoor
Ohserv. Value
SEE
No.
of
Obcerv.
Comments
Pittsburgh
3-hr.
0.491
4.848
0. 142
16.172
-6a Oil
3. 0-29. 1
4. 028
314
III
Mobile
8-hr.
0.488
4.346
0.130
12.044
-S9. 732
4. 262
258
III
Home 1
24-hr.
0. S17
3.963
0. 152
7.916
-44. 596
4. 682
176
III
Denver
3 hr.
0.692
0. 725
0. 361
8. 575
-30. 321
1. 0-16
1. 262
321
III
Conventional
8-hr.
a 619
0.778
0.277
9.021
-35. 481
1.253
300
III
24-hr.
0.713
a 696
0.259
7.091
-37. 442
1.055
168
III
Chicago
3-hr.
0.603
1.342
0.489
9.946
-14.083
2-19. 8
1.089
320
III
Experimental I
8-hr.
a 464
1.789
0.341
11.292
-15. 973
1. 214
252
ni
24-hr.
a 574
1.350
0.449
7. 23S
-10. 652
1.012
155
III
Pittsburgh
3-hr.
0.511
0.414
0.226
4.002
-35. 463
0.0-110
1.463
304
III
Low-Rise
8 hr.
a 505
0.398
0.166
3.417
-45. 249
1.455
240
III
Apt. I-
24-hr.
a 514
0.488
0.159
3.246
-40.140
1. 548
164
III
Baltimore
3-hr.
0.0
2.0
0.0
0.0
0.0
2. 0-4. 2
0.0
299
III
Experimental I
8-hr.
a 948
1.740
1.042
5.270
1.958
4.983
269
III
24-hr.
0.946
2.40
1.039
6.239
1. 596
S. 327
221
III
W asliing ton
3-hr.
0.945
-0.169
1.113
-1.705
5. 506
3-9.7
0.744
357
II
Conventional 1
8-lu-.
0.919
-0.474
1.231
-3. 487
7. 582
0. 874
301
II
24-hr.
0.906
-0.800
1.326
-4.366
7.413
0.896
200
III
Baltimore
3-hr.
0.704
1.127
0.448
15.164
-20. 740
1.8-114
0.668
290
III
Conventional I
8-hr.
0.627
1.384
0. 329
15. 933
-24. 627
0.809
228
III
24-hr.
0.825
1. 006
0.472
11. 709
-20. 109
0. 630
154
III
W ashington
3-hr.
0.140
1.920
0.041
57. 273
-58. 279
1. 4-2. 5
0.063
313
III
Experimental i
8-iur.
0.051
1.985
0.008
92.926
-96. 717
0. 060
264
III
24-hr.
O.OSO
1.993
0.007
89. 566
-95.012
0. 061
168
III
-------
TABLE 34„ SOj FREQUENCY DISTRIBUTION
Residence
Percentage
SO2 Range Is ppb
Pittsburgh Mobile Horn# I
44.2%
3. 00 - 5. 60
35. 2%
5.60 - 8.20
Denver Conventional
90. 5*
1. 00 - 2. 50
Chicago Experimental I
30. OK
2. 00 - 3. 80
15. 6%
3. 80 - 5 60
Pittsburgh Low-Rise Apt. I
60.9%
0. 00 - 1. 00
20.7%
1. X - 2.00
Baltimore Experimental I
99.7%
2. 00 - 2. 20
W ashington Conventional I
60.5*
3. 70 - 4. 40
28. 3*
7. 90-8. 60
Baltimore Conventional I
94.9*
2. 00 - 2. 90
Washington Experimental I
98. 8*
2. 00 - 2.12
Three factors influence these low levels of SO2: 1) the observed SO2
outdoor levels are generally low; 2) the pollutant is a moderately reactive
gas, so the indoor levels are lower than the outdoor levels; and 3) the
flame photometric principle of detection, employed in the commercial instru
ment used 1n the field operations, is subject to negative interference
from CO2. The instrument, although "approved" by EPA for sampling in the
outdoor ambient environment, is subject to quenching by C02» which is pres-
ent in high levels in the indoor environment. The extent of the negative
CO2 interference on the SO2 levels is illustrated in Table 35, which shows
the results of four tests performed by the field team of this project. In
each case the same correction factor is calculated; however, we have not
undertaken such corrections because the observed levels are almost always
close to very low, unreliable levels.
-109-
-------
TABLE 35. NEGATIVE COj INTERFERENCE ON S02 LEVELS*
[co2]
[*>2]
[S°2]
Introduced inco the
Output
Introduced
Test No.
SO2 Monitor
by the Instrument
into the Instrument
1
300
0 33
0.33
813
0. 265
0. 33
1460
0.205
0. 33
1975
0.16S
0. 33
2
300
0.23
0. 23
833
0. 18
0. 23
1450
0.14
0. 23
1975
0.115
0. 23
3
308
0.095
0. 095
850
0. 07S
0. 095
1500
0.055
0. 095
2037
0.045
0.095
4
312
0.050
0. 050
372
0.040
0 050
1525
0.030
0. 050
2088
0.023
0 050
* All concentration levels in ppm.
The point here is that while the model does not simulate the observed
SO2 concentrations, both the estimated values and the observed values are
too low and too close to zero to justify employment of correction factors.
It is concluded that the model's ability to correctly estimate SO2 values
has not been tested by the available data. SOg levels have been found to
be low in the indoor environment not only by the present study but in all
similar studies. SO2 concentrations decay at rates similar to NO2; thus,
it is expected that the 6I0AP model would realistically simulate higher
and more variable indoor levels.
Ozone (03)--An ozone table similar to the statistical summary tables for the
other pollutants would indicate that the model does not satisfy the predeter-
mined criteria. However, it is misleading to consider the model performance as
-110-
-------
unsatisfactory since the majority of the estimated values are within 2 ppb
of the observed values; this difference is smaller than the monitor's pre-
cision. The observed ozone levels in the indoor environment are low and
often constant for long time periods. The small variations in the predicted
values weigh heavily in the estimation of correlation coefficients and other
statistics that assess the power of the model to predict. The performance
of the GIOAP model to predict indoor ozone levels is judged adequate and not
unsatisfactory because it gives a realistic picture of the ozone variation
indoors. Finally, the linear dynamic model (Shair and Heitner, 1974) has
been utilized to predict higher indoor levels, and its use in conjunction
with the GIOAP model is recommended.
Nonmethane Hydrocarbons(NMHC)--The model simulates the majority of the
investigated cases well. The model requires knowledge of the molecular
weight of the pollutant examined; in the case of hydrocarbons we had to use
an average molecular weight representing the hydrocarbons most often sampled.
Thus, the uncertainty introduced may have caused some of the Class III
judgments. In spite of this uncertainty, Table 36 indicates that the GIOAP
model estimates the indoor nonmethane hydrocarbon levels satisfactorily in
the majority of the cases examined.
Methane(CH^)—Table 37 illustrates that the model estimates the indoor
methane values realistically. As always, 24 cases are run. For this
pollutant, 2 are judged Class III, 17 are accepted as Class II, and 5 are
classified as Class I. Following the techniques used in previous pollutant
analysis, we estimate a maximum difference of approximately 30% between the
statistically estimated CH4 concentration and its corresponding observed
value. This is one of the largest percent differences found in Table 37.
-Ill-
-------
TABLE 36. STATISTICAL DATA SUMMARY FOR NONMETHANE HYDROCARBONS (THC-CH4 ppm)
Residence
'epls
r
b
in
'b
t
in
Range
of Indoor
Obscrv. Value
SEE
No.
of
Observ.
Comments
Pittsburgh
Mobile
Home I
3-hr.
a 819
1.036
0.585
10.996
-18.596
a 0-13
1. 189
338
in
8-hr.
0.74S
0.884
0.665
6.752
- 9.451
1.365
285
in
24-hr.
0.721
a 765
a 718
4.511
- 5.690
1.368
196
11
Denver
Conventional
3 hr.
a 817
0.030
0.851
a 707
- 3.831
0. 0-5. 33
0.402
239
II
8-hr.
0. 792
0.120
0.714
2.408
- 7.448
0.393
209
II
24-hr.
Q. 662
0.186
0. 542
2.944
- 7.818
0.386
112
III
Chicago
Experimental (
3-hr.
0.951
a 305
0.899
3.532
- 5.829
1.3-16
0.769
292
II
8-hr.
a 873
0.533
0.788
4.059
- 7. 320
0.900
233
n
24-hr.
a 949
a 286
0.845
2.809
- 6.592
0.602
144
h
Pittsburgh
Low-Rise
Apt. (
3-hr.
0.85S
0.117
1.046
0.46S
1.295
0. 67-58. 3
3. 573
323
ii
8 hr.
a 853
-a 630
1.29
- 1.99
5.995
3. 85
262
ii
24-hr.
a 929
0.376
1.167
-1.795
4.399
2. 053
1S1
n
Baltimore
Experimental I
3-hr.
a 809
0.119 .
0.840
1. 333
- 4.469
0. 0-16
1. 184
290
ii
8-hr.
0.821
0.246
0.621
3. 358
-14.007
0. 921
256
ni
24-hr.
a 832
0.196
0.603
2.251
-13.909
0.964
201
hi
W ashington
3-hr.
a 879
0.043
0.915
0.818
- 3.045
0.0-12.78
0.642
320
n
Conventional 1
8-hr.
0.863
0.096
0.798
1.778
- 6.830
0.S18
252
ii
24-hr.
0.747
0.299
0.602
5.48
- 9.277
0.314
159
in
Baltimore
3-hr.
0.727
0.227
0.590
6.571
-12.421
0. 0-3. 78
0. 417
287
hi
Conventional 1
8-hr.
0.677
0.198
0.533
4.SOO
-11.805
0.441
216
in
24-hr.
0.771
0.121
0.633
2.785
- 8.270
0. 376
141
in
Washington
3-hr.
0.961
-0.006
0.936
- 1.07
-3. 856
0.0-12
0.664
267
u
Experimental f
8-hr.
0.942
0.006
0.886
- 0.095
.5.062
0.721
201
ii
24-hr.
a 984
-0. 051
0.921
- 1,101
-5.008
0. 395
110
ii
-------
TABLE 37. STATISTICAL DATA SUMMARY FOR METHANE (CH4 ppm)
Residence
•pit
r
b
m
'b
t
ui
Range
of Indoor
Observ. Value
SEE
No.
of
Observ.
Comments
Pittsburgh
3-hr.
a 914
a 352
a 769
5.246
-12.438
a 0-10.2
0.878
341
II
Mobile
8-hr.
0.895
a 251
0.861
3.197
- 5.50
0.917
291
II
Ilome I
24-far.
0.880
0.307
a 823
3.315
- 5.656
a 887
203
n
Denver
3 hr.
a 812
0. S79
0.716
7.227
- 8. 521
0.11-9.17
0.392
239
II
Conventional
8-hr.
a 689
0.930
0.556
9.199
-10.951
0.446
209
III
24-hr.
a 708
a 813
0.576
6.25
- 7.741
0.359
112
III
Chicago
3-hr.
a 921
a 086
a 931
2.629
- 3.884
0.0-3
0.198
301
n
Experimental I
8-hr.
a 847
0.068
a 917
1.226
- 2.220
0.268
242
I
24-hr.
0. 844
0.120
0.840
1.734
- 3.555
0.251
144
II
Pittsburgh
3-hr.
a 832
0.226
0.779
5.628
- 7.630
0.0-6
0. 442
324
n
low-Rise
8 hr.
0,816
a 268
0,753
5.618
- 7.469
0.460
262
ii
Apt. I
24-hr.
a 870
0.215
a 799
4.11
- 5.438
0.362
152
ii
Baltimore
3-hr.
a 871
0.441
0.763
7.582
- 9.364
0. 78-4.9
0.380
290
ii
Experimental 1
8-hr.
a 859
0.483
0.731
7.601
- 9.833
0.460
256
ii
24-hr.
a 850
a 474
0.723
6.080
- 8.704
0.444
201
n
Washington
3-hr.
a 943
a 127
0.907
4.281
- 5.316
a 67-3.0
0.181
338
ii
Conventional I
8-hr.
a 886
a 210
0.836
4.669
- 6.182
0.253
275
ii
24-hr.
a 918
0.108
0.913
2.335
- 2.949
0.206
183
ii
Baltimore
3-hr.
a 940
a 212
1.008
0.352
1.164
1.0-18.0
1.744
289
i
Conventional I
8-hr.
a 940
-0.332
1.141
-1.534
5.039
1.7S7
224
ii
24-hr.
a 951
-0.554
1.200
-2.222
6.027
1.664
150
ii
W ashington
3-hr.
a 928
0.147
0.995
1.794
-0.0233
0.0-8.1
0.685
297
i
Experimental I
8-hr.
a 919
a 179
a 986
1.898
-0.490
0.722
238
i
24-hr.
a 904
a 197
0.994
1.534
-0.164
0.820
157
i
-------
Carbon Dioxide(C02)--The last pollutant investigated in this section
is CO2; Table 38 illustrates that the GIOAP model estimates the indoor CO2
concentrations very well. Nine cases are judged Class II and 15 are judged
Class I. Following the procedure used in the other pollutant analysis,
the difference between the estimated CO2 concentrations and the corresponding
observed indoor values is at no time greater than 8% of the observed value.
Model Sensitivity--
An integral part of the model validation is a theoretical analysis of
the model sensitivity.- A study of this nature shows how errors in the esti-
mation of a model parameter affect the model output(s). One of the unique
features of the GIOAP model is the transient term. Previous numerical stud-
ies simulating the relative balance between the indoor and the outdoor
environments have included steady-state conditions only. An assessment
of the transient term indicates that this term is most important for stable
pollutants but that its contribution is minimal for reactive pollutants.
Exclusion of the transient term reduces the correlation between observed
and estimated values by about 50% of its value with the transient term
included. Exclusion of the transient term has no effect on the estimations
of indoor concentrations of the chemically reactive ozone. This behavior
is expected from theoretical considerations, since, for ozone, the decay
term in the exponent of the transient term is very large; therefore, the
transient term approaches zero, which is not the case for stable pollutants.
Thus, the GIOAP model becomes a steady-state model for reactive pollutants; the
model, however, is very sensitive to the transient term for the stable pollu-
tants. A sensitivity analysis will further indicate which parameters are
highly sensitive to errors (i.e., a small error in such a parameter would
result in a significant error in the output(s)), and which parameters are
relatively insensitive to errors. Knowledge of the sensitivity of the var-
ious parameters will assist us in determining priorities in the utility of the
model and in estimating parameter values when the model becomes an application
-114-
-------
TABLE 38. STATISTICAL DATA SUMMARY FOR CAR RON DIOXIDE (CO, ppm)
Residence
'epls
t
b
m
-------
tool. Model validation studies are often undertaken under the best possible
conditions; thus, it is necessary to study the model sensitivity in order to
define the model's limitations and capabilities.
For a given general model y 3 f()T,F), model sensitivity is defined as:
af(X.F)
3p1
(X,F) - (X0,P0) (14)
where
f ¦ the function defining the mathematical model
X* 3 U-|, X£, ..., xn> » the vector of independent variables
x0 - (x1q, x2q Xn0} s f1xed value of *
F * (P-|» P2» •••» P|<> s the vector of parameters
Fq 3 {PiQ» P20» •••» Pk0} 3 f1xed value of ?'
For some insight as to why this formula is used to measure model sensi-
tivity, one must refer to the following equation:
3 £ dpi * (15>
1-1 3p1 1
Equation (15) indicates that the approximate error (df) of f 1s a linear com-
bination of the errors in the individual parameters (dpj, 1*1, k) where
the coefficient of each dp-j, i=l, k 1s the corresponding sensitivity
coefficient. This approach is used for error or sensitivity analysis when
Af, the actual change 1n the function, can be approximated by df.
The GIOAP model is a first-order initial value problem given by
Equation (3). Thus, for this case, the function f referred to in the
definition of model sensitivity is replaced by Cin, Equation (4). The
GIOAP model sensitivity coefficients are as follows:
3C1n -(D+v)(t-t0)
*e t0
-------
3Cin
3mout
(*)[&-on
aC1n
sbout
/v\ / -(D+v)(t-t0)\
(b+V V " e ) 'o - * - V
3Ci
1n
3D
-[C1n.- {ok) ("out^O * '"out + 7 " ¥)]lt-V»'(°hlt,'V
'0
2m... .v
(18)
3C -(D+v)(t-t0)
IT1 * 1 " eV(D+v) toi'iV <19>
¦ {isk) fbout + I * ~T5+7~ + moutut) 'o - 1 - 'f <20>
. Sir i
(2,)
" - (l5*r) [moutDtO + bout° - I - "out (b^)]8 ' K 0>
¦[C1n0 - (lifcr) (™outvtO + boutv * 7 - T¥^)J ' 0>
+ (lJ*r) [boutD ¦ T - "out (8^) + "out"] *0 i i v (22)
-117-
-------
where
tg * initial time
tf 3 final time.
In the balance of this document the subscripts referring to indoors and
outdoors will be eliminated; thus, we will denote C-}nQ by Cg, C-jn by C, mout
by m, and bout by b.
The study documented in this report requires several additional, though
nonrestrictive, assumptions to be made (see the subsection The GEOMET Indoor-
Outdoor Air Pollution Model) in order to implement the model, Equation (4).
Those assumptions resulted in Equation (5). The sensitivity coefficients
for Equation (5) are as follows (see Volume II) for the derivations of the
sensitivity coefficients:
m
(23)
J.'"'"'1 -.,)]}
-(vj+Dj)"
(26)
-118-
-------
3C_ (i * v 2 r Sj 2m>v>
" "1 wv [Vi'i-i + uibi* r - b-^
-(v^+0^)
" [c1-l " (5-^7) ("1 "1*1-1 + "lbl + T * 5^7)] e
S1 Vl\J -K+O,)
1 ' T
" fe) (v'bi+ t ' 5^7* Vi'i)}
sC. , • 5
"W" al
.WZ(D1+v1>] ('
• "1 {" (b^) ['Wi-i + bl°i - T - mf (s^)]e'1 1
S1 miv1\] *(v1+0i)
C1-l • (b^") (Vll1-1 + b1v1 + T • B^)]
(b?^")2 h°t" t - mi + mi°iti]^
where
3C,
m _ 3C
| mro' V Sm' V Vf V Cm-1*
( ) » any one of the GIOAP model parameters
*0 - V - *n * *f1* 1 - 1 * m - n
(1, 1 * m
m
- ,.t+1 W
e } 11 , 1 < 1 < m .
(27)
(28)
(29)
-119-
-------
Two major advantages become apparent when the closed form partial deri-
vatives are available. The first is the ease with which the sensitivity
coefficients can be computed. The second is that the sensitivity coeffi-
cients can be given a thorough analytical treatment, which is difficult, if
not impossible, when the partial derivatives are not available in closed
form. Each of the seven sensitivity coefficients (Equations (23) through (29))
will be discussed below.
The first sensitivity coefficient to be considered is 3Cm/3CQ. Referring
back to Equation (23), it is seen that 0 < 3Cm/3Co < 1. Moreover, as time
increases, aCm/aCo decreases, which means that the effect of an error in Cq on
C diminishes with time. Finally, since aCm/sCg is positive, an increase in
Cq will cause an increase in Cm, and a decrease in CQ will result in a decrease
in Cm.
Next, the sensitivity coefficients involving m, the slope of the line used
to approximate the outdoor pollutant concentrations, will be dealt with. By
rearranging and deleting terms from Equation (24), it is seen that
Equation (30) shows that the effects on C of an error in m at t 3 ti will
dissipate with time. On the other hand, it should be noted that if there is
a recurrent error in m (e.g., every value of m-j, i = 1, ..., n is in error
by 20%), the effects will be additive, i.e.,
with the elements having the lowest indices contributing less and less to
the error 1n C.
* ai D,j+v.j ti' 1 - i' m - n
(30)
»• • 9 n
(31)
-120-
-------
The sensitivity coefficients involving b, the intercept of the line used
to approximate the outdoor pollutant concentrations, will be examined in this
paragraph. From Equation (25), it is clear that
0 *557<°i (5^7)• t32)
Thus, the effects of an error in b at t^ on C diminish with time. Also,
sCjjj/sb^ < 1, which means that errors in b are not magnified when they are
transmitted to C. In addition, the fact that aC^/ab-f is positive means that
an increase in b^ causes an Increase in Cm, and, similarly, a decrease in b..
results in a decrease in Cm. As in the case of m, recurrent errors in b are
additive.
The sensitivity coefficients involving S, the internal pollutant source
rate, will be discussed next. It is seen from Equation (26) that
3C otj
0 < aSp (33)
As before, due to the action of a^, the effects of an error 1n S at ti on C
will decrease with time. In addition, considering the fact that V 1s a large
number and that the product V times v+D is large (even though v is usually in
the interval (0.1, 2.0)), 1t 1s seen that 3Cjn/3S1 wil1 be small. Thus, C is
relatively Insensitive to errors in S. Moreover, since 3Cm/3S-j 1s positive,
it 1s seen that an Increase (decrease) 1n will result 1n an increase
(decrease) 1n Cn,. Finally, as 1n the case of m, recurrent errors in S are
additive.
The next sensitivity coefficients to be studied are those that deal
with"chemical decay rate, D. From Equation (27) 1t 1s seen that the expres-
sion for 3(^/30^ 1s a very complicated one. As a result, 1t is hard to make
any meaningful analysis or calculate any useful bounds. As before, the
effects of an error 1n v-f at t » tj will diminish with time, and the effects
of recurrent errors are additive.
-121-
-------
The next sensitivity coefficient to be examined is 3Cm/3V. Since V
does not change from hour to hour, any error associated with V will occur
initially and will dissipate with time as in the case of Cq. Also, it is
easily seen from Equation (28) that 3Cm/3V is less than zero. Thus, an
increase (decrease) 1n V would cause a decrease (increase) in Cm. Values
of V are usually obtainable from plans or blueprints, thus minimizing any
error connected with V; thus, V is, for all practical purposes, a known con-
stant, and aCm/aV was presented here as a point of interest.
Finally, the sensitivity coefficients associated with the air exchange
rate v will be discussed. As can be seen from Equation (29), 9Cm/av.. is
a complex expression, and, due to the interactions of the various elements
of the equation, 1t is difficult to make any meaningful analysis or to com-
pute any useful bounds. As in previous cases, the effects of an error in
vj at t a tj on C will diminish with time, and the effects of recurrent
errors are additive.
The balance of this section will consist of graphical illustrations
representing the effects induced on the indoor pollutant concentrations by
errors imposed on different input parameters. Figure 29 represents the
nominal conditions, i.e., a CO 8-h episode calculated by the GIOAP model;
this episode is extracted from the data set of the Baltimore conventional
residence, Visit Number 1. The baseline conditions are also shown in Table 39.
Figure 30 shows how a 50% error on the initial condition affects Cin
over the 8-h episode. The figure illustrates that for this example the
effects of the initial condition error on the indoor concentration are essen-
tially eliminated after 2 h. This case 1s an example of a nonrecurrent
error; i.e., a single parameter is perturbed only once in the episode, and
the effects of the Introduced perturbation are traced for the duration of
the episode. The same type of error behavior in Cin seen in this example
will occur for any other parameter under similar conditions.
-122-
-------
BALTIMORE CONVENTIONAL RESIDENCE
COppm, 9/7/76
2.7
2.4
2. 1
a
&
- 1.8
w 1 2
U •
o
0.6
0.3
0
TIME OF DAY
Figure 29. Nominal values.
TABLE 39. NOMINAL CONDITIONS USED IN THE SENSITIVITY STUDY EXAMPLES
Houiei Baltimore Conventional (Vliit #1)
Pollutant: CO
Yolumei 13,575
Hour
Cta
(PP»)
C
s
(mg/h)
if
(ak exchaagtt/h)
8
1.33
1.33
m
•
9
2.23
1.33
«77.77
1.20
10
1.04
0.00
0.00
1.20
11
0.31
0.00
0.00
1.20
12
0.68
0.00
440.14
1.20
13
1.02
0.00
613.13
1.20
14
1.01
0.00
528.17
1.20
15
0.30
0.00
0.00
1.20
-123-
-------
3.0
2.7
2.4
B2'1
a
a
2*1. 8
O
1.5
Uj 7
§
U
.9
.6
.3
BALTIMORE CONVENTIONAL RESIDENCE
CO, ppm
J I I I 1 1 I I 1 1 i-
9/7/76
Nominal
Q o Perturbed
J I I I I 1 1 1 L
88 8 8 8 8 8 8
1/5 VO 00
O O O O
h pj ff)
o o o
s
88888888888
O N ro LO >0 N 00 ^
O r-4 *H *H t-H
TIME OF DAY, hours
8 8 8 8
- - ^
CM
*h (N fO
N N N
Figure 30. Comparison of nominal values obtained by perturbing C
in*
Figure 31 Illustrates the errors in the indoor pollutant levels caused
by recurrent time-dependent errors in the internal source (S) rate term. The
magnitude of the error in S at a given time is 30% of the corresponding
internal source rate.
Figure 32 Illustrates the errors in the indoor pollutant level caused by
a recurrent constant error 1n the air exchange rate (v). The nominal input
value for v 1s 1.2 air exchanges per hour, the error used is 0.2 air exchanges
per hour.
Numerical investigations of the above illustrated cases appear in
Appendix C (Vol. II) where the sensitivity coefficients, the parameter
errors (a$, av, aCq), the actual output errors (aC^n), and approximate
output errors (dC^) are tabulated.
-124-
-------
IIAI.TIMOIU' I'ONVKN I IONAI IU-:NII >i :N I i.
CO,
'VV//o
Key:
o~—o
Nominal
Perturbed
J I I ( I L
J I I I I I I I I i
J 1 I L
8 8'8
H N fO
o o o
.88
in v© n. oo
8
o o o
o o
8 8 8 8?
h rj ro 2 L
TIME OF DAYt hours'
8
8 8 8 8 8
O fM CO
CM CM £M CM Csl
Figure 31. Comparison of nominal values with values obtained by perturbing S.
9/7/76
BALTIMORE CONVENTIONAL RESIDENCE
CO, ppm
7 _
4 -
1
8
Key:
— Nominal
O O Perturbed
9
6
3 U
J L
J I 1 I I—L
X
' » ' L
8 8 8 8 8
s s a
o o o
8
888888888888888888
5 oo K o th t\i ^ A to k 9>
-------
CONCLUSIONS
This study indicates that the GIOAP model performs well. The funda-
mental principles and assumptions used in the formulation of the model are
similar to the concepts used for other indoor numerical models; however, the
model is different from previous numerical efforts not only because it is
applicable to, and has been applied to, a large variety of pollutants, but
also because it simulates short periods of time since the transient term is
included. Most importantly, the model stands alone because it is the only
indoor-outdoor numerical model that has been validated against a large set
of observed data.
The GIOAP model has been tested under a wide variety of meteorological
and behavioral conditions. Weather conditions encountered ranged from late
autumn in Denver, to summer in Baltimore, to winter in Pittsburgh. Behav-
ioral patterns varied widely; e.g., families with children versus families
without children, and families with smoking members versus nonsmoking
families. In addition the model has simulated conditions in residences of
different structural characteristics; e.g., detached dwellings, row houses,
apartments, and mobile homes. Thus, the model not only provides good esti-
mates of indoor air pollutant concentration levels for different pollutants,
but also does so for various types of residential structures under diverse
meteorological and behavioral conditions.
The transient term Included 1n the GIOAP model does not appear 1n
other numerical models that estimate indoor air pollution levels. Examina-
tion of the GIOAP model and simulations with and without the transient term
have led to the following conclusions: 1) the transient term contributes
substantially when the variation of indoor concentrations for stable pollu-
tants 1s simulated, 2) the transient term becomes less significant for
moderately reactive pollutants, and 3) the transient term is unimportant
for ozone which 1s a highly reactive pollutant.
The model validation phase was undertaken under "best" conditions; i.e.,
the parameter values used for simulation were the best estimates available.
-126-
-------
FCK Cq, m, b, and V the actual values were obtained from the monitoring data.
The large data base available for this study was utilized to compute con-
strained best least-squares estimates for S and v, the two parameters most
difficult to obtain. The estimates for these two parameters were computed
with the Parameter Estimation Procedure. Finally, values for D for
various pollutants were obtained from the available literature.
The motivation for obtaining the "best" estimates for all the parameters
underscores our desire to validate the GIOAP model under ideal conditions, to
characterize its performance without considering the problems of obtaining
realistic values for the input parameter^ and, subsequently, to determine its
sensitivity resulting from errors in the values of the input parameters.
The GIOAP model was statistically tested using two sets of rules:
1) strict statistical tests, and 2) predetermined empirical criteria. Under
"best" conditions, the model performance is divided into three categories:
Not Validated—Due to low outdoor levels and a negative inter-
ference of CO? on the SO2 monitor, a large number of hourly
SO2 concentrations are measured close to the threshold value
of the monitor. In the indoor environment, SO2 and NO2 have
an approximately equal half-Hfe; it is thus judged that the
GIOAP model validate satisfactorily against higher Indoor SOg
concentrations.
2. Adequate—EiQhty-f1 ve percent of the observed indoor ozone
values are in the low range of 0-6 ppb. The model estimated
values are mostly within 2 ppb (less than the Instrument pre-
cision) of the observed values. The numerical output does not
satisfy the predetermined model validation criteria, but it
provides realistic estimations of the Indoor ozone concentrations.
3. Satisfactory—Model estimated values are within 25% of the
observed Indoor values for the following pollutants: CO, NO,
NMHC, CH4, CO2, and NO2.
-127-
-------
Under "best" conditions the GIOAP model provides realistic estimates
of the observed indoor pollutant concentrations. However, due to the
nature of model usage, it is necessary to investigate the performance of
the model under less than ideal conditions. As a result, a sensitivity
study of the model is necessary. The set of parameters associated with
the model has been divided into two groups: 1) those parameters that
remain constant throughout an episode (Cq, V), and 2) those that must be
estimated for every hour of the episode (m, b, S, v, D).
The model is comparatively insensitive with respect to parameters
in the first group; i.e., errors in Cq and V have relatively little effect
on the model output. In addition, the effects of errors in Cq and V
dissipate with time. Errors in the second group have more impact on the
model output. This is due to two factors: 1) estimates of parameter
values in this group are more susceptible to error than the parameters
in the first group, and 2) errors can be introduced at each hour of the
episode because these parameters must be estimated every hour. In an
effort to stratify the parameters of this group on a relative basis, they
are ranked as follows from the least to the most sensitive: 1) S, 2) b,
and 3) D, m, and v. S is the least sensitive because the magnitude of
the numerator is less than one, and the denominator, which contains a
factor of V, is relatively large, b is considered to be somewhat more
sensitive than S because aC/aS = (1/vV) (a C/a b) and because, within the
range of values being used, 1/vV is less than 1. Finally, D, m, and v
are considered to be the most sensitive, even though the complexity of
their respective sensitivity coefficients does not allow any general con-
clusions to be drawn. Intuition and the examples studied suggest that
these are the three most sensitive parameters.
-128-
-------
THE STEADY-STATE TSP MODEL
Introduction
A recent EPA indoor-outdoor study of Total Suspended Particulate (TSP)
matter has led to the general conclusion that indoor TSP levels are approxi-
mately equal to outdoor concentrations (Henderson et al. 1973). Examination
of the data base available to the GEOMET Indoor-Outdoor Pollution project
clearly shows that the ratio of TSP indoor concentrations to outdoor concen-
trations varies from 0.3 to 3.6. An effort has been undertaken to identify
the factors that influence this ratio and to formulate a model that would
predict the indoor TSP matter levels.
The need for a separate procedure to relate the indoor TSP levels with
corresponding outdoor TSP concentrations became evident early in the project.
The nature of the pollutant and the type of TSP data collected are not suit-
able for simulations with the GIOAP model, because the model requires fine
time resolution for the pollutant concentrations and other input parameters.
The experimental design for TSP matter requires that from each of the four
sites, one 24-h average be obtained for each day of the monitoring period.
The TSP empirical model formulated for this study is a steady-state
model built on the available data. A portion of the TSP matter found
indoors is of outdoor origin, while the remaining portion Is attributed to
indoor activities. Studies on particulate matter have concentrated on quan-
tifying source strengths of Individual Indoor TSP generating mechanisms,
such as vacuum cleaning, operation of a fan, smoking, frying, housecleaning,
use of sprays, moving in and out of the house, ventilation devices, and
others. The steady-state TSP model does not require knowledge of individual
TSP source strengths; rather 1t quantifies Indoor TSP levels as a function
of the family activity Index. This approach takes advantage of the data
available to the project and utilizes the questionnaire which was answered
on a dally basis by a responsible member of the household. The procedure
followed employs a portion of the available data to define the Indoor TSP
strength as a function of the activity scale, and the remaining data to
verify the numerical predictions.
-129-
-------
TSP matter is generated indoors by either thermal or mechanical sources.
Thermal sources are smoking and cooking. The term mechanical sources refers
to the processes of resuspension and dispersion of existing indoor particu-
late matter. In this section we will describe a phenomenological model which
is a combination of theoretical principles and empirical parameters. The
model quantifies indoor sources, both mechanical and thermal, as a function
of the indoor activity index and predicts the total indoor TSP levels as a
function of outdoor levels and indoor source strengths.
TECHNICAL APPROACH
Theoretical Considerations
The fundamental principle involved in relating the indoor and outdoor
TSP levels is expressed by a mass balance equation which requires that the
rate of change of the indoor levels equals the sum of four processes: 1) the
rate by which TSP enters the house; 2) the rate by which TSP is generated
indoors; 3) the rate by which TSP escapes from the house; and 4) the rate by
which TSP matter is "removed" from indoor air. Each term will be discussed
in detail in the following paragraphs.
The rate of change of the indoor concentration is assumed to be zero,
this steady-state assumption is due to the nature of the data available to
this and most other TSP monitoring projects. Thus, the first, most basic
assumption of the steady-state TSP model is that the indoor particulate
levels are constant over each 24-h period.
A fundamental difference between gas contaminants and particulate
matter is that only a portion of the TSP matter passes through the cracks
of each structure; the remaining portion settles on the outside surface of
each structure which acts as a barrier. A cleansing factor, f, determines
the portion of TSP that passes through this barrier.
In the indoor environment TSP is generated by two broad mechanisms: TSP
is generated either by a thermal source or by the resuspension and dispersion
of existing particulates. The nature of the data available to this project
-130-
-------
is not suitable for estimating the rates of individual sources; however,
indoor source strengths and resuspension rates have been found to be pro-
portional to the daily family activity. The indoor generation term is
given by S(k), where k is the activity index discussed earlier in Section 2
of this document.
The indoor TSP matter concentrations decrease by exfiltration and by
removal mechanisms. Several removal mechanisms have been identified indoors:
residential filtering devices, adsorption, and gravitational settling. The
total decrease rate of indoor particulate matter by the removal mechanisms
is denoted by R.
The steady-state equation that expresses the above procedures is given
by
vfCout " vCin " RCin + S(k) = 0 (34)
where
3
= TSP concentration indoors, yg/m
v = air exchange rate, air changes/time
f = cleansing factor, pure number
Cout = ^SP concentrat''on outdoors, pg/m
R = removal rate, number/time
S(k) = indoor source strength rate, (vig/m )/time
k ¦ activity Index, pure number.
The first term of Equation (34) is the infiltration term. The next two
terms denote the rate of reduction in the TSP indoor concentrations by
exfiltration and removal mechanisms. The fourth term is the indoor source
strength term.
-131-
-------
Source Estimation--
Simple manipulation of Equation (34) leads to the following equation
S = (v ~ R) C1n - vfCout. (35)
Daily values for and C are obtained from the data sets of eight resi-
dences; a portion of the data is used to establish a physical relationship
between indoor and outdoor TSP levels. The air exchange rate is calculated
daily using the following equation:
v = 0.28 + 0.012 AT (36)
where AT (°F) is the daily temperature difference between indoor average and
outdoor average temperature. The general expression for the air exchange
rate given by Equation (36) has been derived from the analysis of experimental
SFg data obtained by the study. It is a comprehensive equation which includes
detached homes, apartments, mobile homes,and row houses. This v equation is
very similar to an equation obtained by Hunt and Burch (1975) for air
exchange rates under experimental conditions.
The cleansing factor or the fraction of the TSP matter that "passes"
indoors must be distinguished from the infiltration rate. Of the total amount
of particulate that could infiltrate indoors, only a fraction f does because
the remainder is stopped by the structural barrier. J. Alzona (Univ. of
Pittsburgh, private communication, 1976) has elaborated on the physical
meaning of this factor and suggests that its value is smaller than 0.5.
Using the data from this project, different values have been estimated
for this factor; the calculated f values vary between 0.23 and 0.38.
It must be realized that a zero Indoor TSP source is required for an experi-
mental estimation of this factor. The f value used in the steady-state
particulate model is 0.3, which is an assumed value supported by other
studies and calculations on the indoor-outdoor data base.
The "removal" term of the steady-state TSP model is estimated by RCjn,
where R stands for the removal rate of the indoor TSP concentrations. Adsorp-
tion and gravitational settling are the two indoor TSP removal mechanisms.
-132-
-------
Hunt (1972) estimates that the total fraction of TSP deposited per unit time
is given by R = k/(Q + k), where k is a first-order rate constant most nearly
approximating the actual deposition rate, and Q = V^E/V, with the rate of
recirculation, V the volume of the structure, and E the filter efficiency. For
the general conditions found in the indoor residential environment, k values
are between 2.5 and 4.0, and E is approximately equal to one. The removal
rate value assumed for this study is R = 0.8.
Utilizing data from eight residences and Equation (35), the calculated
3
range of indoor source strengths varied between 10 and 200 (yg/m )/24-h.
The calculated source strengths show association with two quantities: the
outdoor TSP levels and the activity index. The source strength is propor-
tional to the outdoor levels because resuspension is a major mechanism of
indoor particulate matter generation. The data set of this project conclu-
sively shows that the resuspended TSP is proportional to the portion that
infiltrates on the same day and not any previous days. The activity index
(k) quantifies the means by which the particulate matter becomes resuspended.
Analyses of the calculated source strengths show that the indoor TSP concen-
trations attributed^ to resuspended particulates are equal to 0.24 kvCQUt.
The second category of indoor particulate sources, the indoor generation of
TSP matter, is also found to be proportional to the activity index and
inversely proportional to the daily air exchange rate. The proportionality
q 2
constant takes the value of 1.5 (yg/m°)/(TIME) . Thus, the total source term
takes the following final form,
S(k) = 0.24 kvCQut + (37)
It is apparent that In the formulation of the steady-state TSP model a
series of complex physical effects, such as the sheltering effect, the removal
of particulate matter by adsorption and gravitational settling, and the indoor
TSP generation processes, have been expressed in terms of admittedly over-
simplified parameters. Realistic representation of these complex dynamic
systems requires more data, additional basic research, and change of emphasis
of the present study; thus, the parameterization approach followed 1n the
formulation of the TSP model 1s preferred.
-133-
-------
I idoor TSP Generation Rates--
The approach followed is an empirical procedure which requires continu-
ous manipulation of the data. Examination of the source strength values from
tie first eight residences leads to the following:
• The resuspension source strength rate is proportional to the
air exchange rate v, the removal factor R, the cleansing factor
f (in this study R = 0.8 and f = 0.3), and the outdoor TSP level.
• The resuspension source strength rate is proportional to the
activity index value for typical and active families.
• The activity index value for inactive houses takes the value
of 4. This is because the measures for low activity days do
not seem to be as realistic as for other days.
• The indoor generation rate of TSP matter is inversely propor-
tional to the air exchange rate and proportional to the activity
index. The constant of proportionality has been estimated to be
1.5 units for every activity unit.
• For residences with air exchange rates equal to or less than
0.2 air changes per hour, the value of the activity index k in
the generation term of Equation 37 is increased to 1.5, its
value obtained from the questionnaire. Thus, the source term
for these very tight houses takes the following form:
S(k) = 0.24 kvC . + 1.5 ^-5^k .
out v
Table 40 is a summary of the relationship between the classes of
families and the two types of indoor TSP generation.
-134-
-------
TABLE 40. INDOOR TSP SOURCE STRENGTH SUMMARY
S = Sr + Sg
Resuspension rate Generation rate
(SR = 0. 24 kwCout) (SG = k(l. S)/u)
Inactive k = 4 k floats
Typical k floats k floats
Active k floats k floats for U>0. 2
k = (1. 5) (Value of daily k index)
for 2*
* Owing to the few examples available, this has not been verified separately.
DISCUSSION AND CONCLUSIONS
The steady-state TSP model developed in this section employs an empirical
approach to estimate indoor levels of TSP matter. Using a portion of the
data base, a two-part indoor TSP source strength term is formulated. The
indoor TSP source strength consists of the resuspension term, which refers
to the reentralnment and dispersion of TSP matter that preexists indoors
and becomes airborne by mechanical procedures, and the generation term,
which refers to sources that produce particulate matter in the indoor environ-
ment. Table 41 Illustrates pairs of observed vs. estimated values for three
different houses corresponding to families with three different activity
1evels.
Figure 33 Illustrates that the estimated value is almost always within
the range of the observed daily values. Thus, while it may not agree with the
observed indoor average, it falls within the observed dally range. An effort
was undertaken to verify the estimated values against the observed Indoor TSP
averages using a portion of the data base that was not utilized In the formu-
lation of the model. On a pair comparison, over 60% of the pairs (observed
indoor averages and predicted values) are w}th1n 30% of each other, an addi-
tional 35% of the pairs are within 50%, and the remaining 5% show larger
differences.
-135-
-------
TABLE 41. COMPARISON OF INDOOR OBSERVED TSP LEVEL, AGAINST TSP
LEVELS ESTIMATED WITH STEADY STATE MODEL
Inactive
Typical
Active
Indoor
Indoor
Indoor
Indoor
Indoor
Indoor
Observed
Estimated
Observed
Estimated
Observed
Estimated
11. 3
22. 4
148. 0
107. 0
68. 3
63. 5
25. 3
27. 7
115. 0
111. 0
101. 3
108. 6
X
X
105. 0
102. 0
108. 7
128. 5
X
X
84. 0
69. 0
89. 0
105. 2
26. 7
20. 5
58. 0
45. 0
77. 7
47. 2
39. 7
34. 3
114. 7
61. 3
72. 0
79. 9
X
X
136. 7
44. 0
81. 7
98. 0
31. 7
29.6
63. 7
49. 5
43.7
32. 7
28. 8
37.7
103. 3
38. 0
4S. 7
74. 6
26. 7
20. 0
X
X
88.7
55. 6
21. 7
19. 2
102. 0
60. 0
83. 5
73. 2
35. 7
62.4
94. 3
142. 0
91.0
65. 7
75. 3
93. 8
137. 0
87. 7
86. 0
78. 8
X
X
88. 7
48. 0
93. 0
45. 2
36. 0
42. 6
56. 7
61. 9
x = not available
In the past the TSP indoor-outdoor association has been given as a ratio
relationship. The present indoor-outdoor study has conclusively demonstrated
that this does not realistically describe the situation and that ratios of
irdoor values to outdoor levels vary from 0.3 up to 3.6 depending on a series
of inputs. Among the most crucial factors that determine the indoor-outdoor
TSP relationship is the family activity. The steady-state TSP model takes
this factor under consideration, and it estimates the observed indoor-outdoor
ratio relationships.
Further work needs to be undertaken. More reliable family index values
are necessary. The questionnaire used and the approach suggested have pro-
vided valuable information; however, the integrity of some of the answers is
questionable. In addition, further work needs to be done in order to estimate
the removal factor and the cleansing factor which have been parameterized in
tMs present study.
-136-
-------
6/10 6/11 6/12 6/13 6/14 6/15 6/16 6/17 6/18 6/19 6/20 6/21 6/22 6/23 6/24 6/25
TIME, month and day
Figure 33. Estimated values of indoor TSP using the steady-state model for the Pittsburgh high-rise apartment 3.
-137-
-------
On the other hand, the steady-state TSP model illustrates how basic
physical principles can be expanded to apply to a variety of conditions and
to estimate indoor TSP concentrations. The model has been successfully used
to predict indoor concentrations when the outdoor TSP concentration and the
family activity index are known. In addition, the model can be used to pre-
dict the indoor TSP levels when only the outdoor levels are known and a
family activity classification is assumed. For such cases the activity index
is given the values of 4, 7, and 10 for inactive, typical, and active
families, respectively.
In conclusion, several features of this phenomenological approach sepa-
rate the steady-state TSP model from previous efforts to estimate indoor TSP
levels: 1) individual indoor source strengths are not estimated; rather, an
overall family activity index is used to relate outdoor TSP levels to indoor
concentrations; 2) the model, in agreement with the observed data, shows that
as a factor indoor resuspension of infiltrated TSP matter is at least as
important as the indoor generation of total suspended particulate matter;
3) previous studies have used a constant ratio value to relate outdoor TSP
levels to corresponding indoor concentrations; the steady-state TSP model
estimates indoor TSP levels and ratio values which vary as a function of the
activity index. The calculated ratio values show good agreement with the
observed ratios. The steady-state TSP model is easy to use, inexpensive, and
does not require complex computer procedures to estimate the indoor TSP
levels.
-138-
-------
SECTION 4
ENERGY CONSIDERATIONS*
The evaluation of the relationships between total energy use for a
residence, its space conditioning energy use, and the conditioning energy
use due to infiltration are important when examining the energy use pro-
files within selected residences. In order to evaluate the total energy
use within a residence, certain data are required.
ENERGY DATA COLLECTION
It is essential that the total power consumption be measured on an
hourly basis. For each residence, measurements are made by employing two
recording ammeters. Each meter is connected such that the positive legs of
the power lines to a residence are monitored. The instrument range normally
employed for the measurement is 0-50 A; however, during the winter heating
season, often a 0-250 A ranae is employed. The recording ammeter chart has
five 12-min divisions per hour. The average percent chart deflection for
each 12-min period is estimated for each strip chart, and five 12-min periods
are added together for each hour. This average percent of the chart deflec-
tion is then multiplied by the instrument range, which results in the estima-
tion of the electric current. The amperage per hour is then multiplied by
the voltage of the positive line, giving a net result in watthours. One
hundred-twenty volts is used as the average line voltage for residential use.
Final reportinq is represented in kilowatthours.
In addition to the above measurement, daily readings of watthour meter
are usually made. This information is reported in kilowatthours per day.
The above procedure identifies the power usage due to electrical equip-
ment, but it does not identify the energy used by equipment operating with
different fuel sources. Oil and gas furnaces are examples of such types,
and measurements are discussed below.
Oil Furnaces
To determine the hourly consumption of heating oil, a recording
ammeter is attached to one leg of the 120 V power line going to the
furnace oil pump. Each time the pump is activated, the duration of its
operation is recorded on the recording ammeter. The oil pumping time
per hour is determined from this recording. This time is then multiplied
by the pump rate and burner nozzle specification which is supplied by the
furnace manufacturer to obtain the volume of oil consumed per hour. A
daily measurement of the oil consumption is made by measuring the volume
of the oil tank and the daily replacement employing a dip stick. This
measurement is used as a quality control check for the hourly measurements.
* Prepared by Hittman Associates, Incorporated.
-139-
-------
Ga> Furnaces
Gas consumption rates from gas furnaces are monitored indirectly by
attaching a recording ammeter on the furnace blower. The furnace thermo-
stat is set at a fixed temperature point during the data collection, and
the furnace operates on the followina cycle:
1. Gas ignites to heat furnace bonnet to set temperature (time
for this activity is measured and is constant for each cycle).
2. When the bonnet reaches its proper temperature, the furnace
blower is switched on, and the activated recording ammeter
measurement begins.
3. Gas combustion continues until the thermostat reaches desired
temperature, then the gas valve is closed electrically.
4. The blower continues to operate until the furnace bonnet cools
to a predetermined temperature and the blower is deactivated.
This portion of the time cycle is measured and is constant.
The gas consumption per minute, while the furnace is in opera-
tion, is read from the gas meter.
From the above data the gas usage per hour is calculated.
After conversion to Btu's, the energy required by a gas or oil "furnace
is added to the energy due to electrical equipment, and the total power
consumption is obtained. Power consumption can then be plotted against
time, so that an hourly profile can be produced. The profile reflects the
energy requirements for space conditioning and inhabitant use..
It is important that the space conditioning energy be compared with
the total power requirements of a residence. This is determined by mea-
suring the hourly conditioning electricity used by air conditioners, heat
pumps, and electric furnaces. This is in addition to the measured values
for gas and oil furnaces. Measurements for electric furnaces, heat pumps,
and air conditioners are discussed below.
Electric Furnace
For an electric furnace, the amperage is monitored with the recording
ammeter. This ammeter is connected to the power line of the resistance
heating coil. Should more than one resistance heating coil be employed
within the furance, and they are wired separately, the amperage of each
line is monitored. Since all electric home heatinq units are 240 V, only
one of the two positive 240-V lines is monitored. The average percent
chart from the recordinq ammeters is determined for each hour, added together,
multiplied by 2 to account for the 240 V supply, then multiplied by the
ammeter range to obtain watts per hour. Kilowatthours are reported.
-140-
-------
Heat Pumps and Air Conditioners
The methods for monitorinq power consumption for heat pumps and air
conditioners are the same except that heat pumps normally have auxiliary
resistance coils. The power consumption for these coils is monitored as
described for the electric furnace. A heat pump or central air condi-
tioner is monitored by employing a recording ammeter on one leg of the
240 V supply line. The calculations to obtain kilowatthours are the
same as described above for the electric resistance furnace.
Utilizing the previous methodology, it is possible to plot the
hourly profiles of total power consumption and space conditioning energy
usage for all the residences being studied. An important question remains
unanswered, How much of the conditioning energy use is due to infiltra-
tion of outside air?
Calculation of Air Infiltration
Based on the physical and climatological data collected at each of
the residences, the energy required in changing the temperature and
humidity of infiltrated air to meet inside requirements can be shown
relative to the total house power and space conditioning energy on an
hourly basis. These relationships are the next topic of discussion.
It has been previously mentioned that indoor air quality is a func-
tion of air exchange rates within residences. Infiltration of outside
ar is considered to be one of the major components of heating and cooling
loads. During the heating season, depending on how well a structure 1s
irsulated, 14 to 53% of the total residential structure heat loss is due
tc infiltration, while in the cooling season, 11 to 27% of the total
structure heat gain is due to infiltration (Hittman Associates, 1977).
Thus, the significance of reduced Infiltration relates directly to the
annual energy use of a residence.
Infiltration 1s commonly known as air leakage into a residence
through cracks and interstices around windows, doors, floors, and walls.
Its magnitude depends on the type of construction, workmanship, condition
of the structure, and climatological environment.
The air flow rate into or out of a residence due to infiltration
or exfiltration depends greatly on the resistance to air flow through
openings in the residence as well as the indoor-outdoor pressure differ-
ence, the indoor-outdoor temperature difference, and the life style of
the inhabitants.
The size of the openings has a great effect on the amount of air flow
in or out of a residence. Openings usually occur around windows, window
frames, doors, door frames, exterior walls, and the ceiling areas. It is
often easier to consider all these crack areas as if they were one opening
through which all the in-out air flow is occurring. The opening is
-141-
-------
commonly called the "equivalent orifice area" (EOA). It is a very effec-
tive measure by which the relative air-tightness of a residence can be
compared with other residences.
Currently, two recommended calculation procedures (crack methods)
give reasonable estimates for the equivalent orifice area (ASHRAE, 1977).
It must be understood that the accuracy of these methods is limited by
the data available on the air leakage characteristics for the variety
of components used in a residence and by the differences that develop
between the components as tested and as installed or constructed.
The first method, though more complex, will result in a better
estimate for the equivalent orifice area. It is based on the openings
in the building envelope and is a constant for a particular residence.
The major openings are the cracks in windows, doors, window and door frames,
walls, and ceilinqs. Therefore, the EOA can be expressed as:
where:
EOA = Equivalent orifice area for the residence
Lw = Crack length for prime windows without storm windows
tw = Crack width for prime windows without storm windows
= Crack length for prime windows with storm windows
t^ = Crack width for prime windows with storm windows.
-142-
-------
n = Flow exponent, ranges from 0.5 and 1.0 (assume 0.6)
Lsw
= Crack
length for storm windows
1'sw
= Crack
width for storm windows
Lwf
= Crack
length for storm windows
Srf
= Crack
width for window frames
Ld
= Crack
glass
length for doors without storm windows, except sliding
doors
*d
= Crack
qlass
width for doors without storm windows, except sliding
doors
Lgd
= Crack
length for sliding glass doors
tgd
= Crack
width for sliding glass doors
= Crack
length for storm doors
td
= Crack
width for storm doors
Ldf
= Crack
length for door frames
= Crack
width for door frames
^-ew
= Crack
length for exterior walls
tew
= Crack
width for exterior walls
Lc
= Crack
length for ceiling
*c
¦ Crack
width for ceiling.
A still useful, but less sophisticated method for estimating the
equivalent orifice area can be represented by the following equation:
EOA - (N„ • A« • Cw) + (N0 . C„) ~ • Cw)
where:
Nw = Number of windows
^ ¦ Area of average window
Nd = Number of doors
Awr ¦ Total wall and roof area.
-143-
-------
The coefficients C , Cp., and C are obtained from Table 42, depending on the
type of windows, d$ors, wall, cKfid roof construction found in a residence.
TABLE 42. ESTIMATED VALUES FOR Cw, CD and Cm
Windows
Cw
Poorly Fitted
12. 0 x lO"3
Average Fitted
8. 0 x 10-3
W e atherstrippe d
6. 5 x 10-3
With Storm Windows
5. 5 x 10-3
Doors
CD
Poorly Fitted
0. 3
Average Fitted
0.25
W e atherstripped
0. 20
With Storm Doors
0. 17
Walls and Roof
c
Poor Workmanship
5. 6 x lO"4
Good Workmanship
2. 8 x lO"4
It is now possible to estimate an effective opening, equivalent to
all the crack area in a residence. This effective openinq or equivalent
orifice area is ranked in Table 43 for each residence included in this
study accordinq to its relative "airtiqhtness."
The EOA by itself is not sufficient in characterizing the structural
parameters necessary for proper estimation of the energy use due to infil-
tration. A second parameter, namely "building size or volume," is incor-
porated with the EOA to reveal a new parameter known as the "orifice
coefficient." As shown below, this parameter is the normalization of the
equivalent orifice area by the building volume, and this results in a
factor that represents not only building tightness but also building size.
Equivalent area of all the openings
Orifice Coefficient (OC, . XlkTOrSflSS "
As a result of this normalization, the residences in Table 43 are pre-
sented in Table 44 in order of their relative tightness and size.
-144-
-------
TABLE 43. EQUIVALENT ORIFICE AREAS
Residence
Building
Type*
Equivalent
Orifice Area (ft^)^"
Air
Tightness
Pittsburgh Apartment
MFLR
0. 53
Tight Residence
Pittsburgh Apartment
MFHR
1.29
k
Pittsburgh Mobile Home 2
MH
1.42
Baltimore Experimental
SFA
1.81
Chicago Experimental
SFD
2.16
Baltimore Conventional
SFA
2. 65
Pittsburgh Mobile Home 1
MH
2.81
Washington Experimental
SFD
3.65
Denver Conventional
SFD
3.89
1
'
W ashington Conventional
SFD
4. 27
Loose Residence
Chicago Conventional
SFD
4. 29
* MFLR = Multi-Family Low-Rise; MFHR = Multi-Family High-Rise; MH = Mobile Home; SFA = Single-
Family Attached; SFD = Single-Family Detached.
t It is assumed that doors and windows are not in the open position.
TABLE 44. ORIFICE COEFFICIENTS
Residence
Building
Type
Orifice
Coefficient (ft"
Chicago Experimental
SFD
0.7
Pittsburgh Apartment*
MFLR
0.8
Pittsburgh Apartment#
MFHR
1.8
Pittsburgh Mobile Home 2
MH
2.0
Baltimore Experimental
SFA
2.3
Washington Conventional
SFD
2.3
Washington Experimental
SFD
2.5
Chicago Conventional
SFD
2.8
Baltimore Conventional
SFA
2.9
Pittsburgh Mobile Home 1
MH
3.3
Denver Conventional
SFD
4.6
~ It is assumed that doors and windows are not in the open position.
-145-
-------
2
Although the Washinqton conventional residence has an EOA of 4.27 ft ,
it had a lower measured infiltration rate than the Baltimore conventional
residence which had an EOA of 2.65 ft . This is_in complete agreement
with the orifice coefficients which are 2.3 x 10" /ft and 2.9 x 10" /ft for
the Washington and Baltimore residences, respectively.
The orifice coefficient is a good representation of building param-
eters, but it is recommended that physical inspection of the residences
be made during future research in this area. This study did not allow
for the cutting of holes, or dismantling of any of the residences in
order to verify potential crack areas. In most instances, engineering
judgment is used in the evaluation of these cracks. For example, a tight
residence may have a window and frame crack width of 1/16" and 1/64"
respectively, while a loose residence may have 5/64" and 3/128" for its
window and frame crack widths.
Once the air exchange rates (air changes/h) are found experimentally
for various residential types, a methodology can be developed to estimate
the amount of infiltration. In this study, tracer gas in the form of SFe
is utilized in determining hourly air exchange rates. SFg is used mainly
because it is nontoxic and decomposes at a relatively high temperature.
By measuring the dilution of the SFg gas with time it is possible to
obtain the air exchanqe rate in a residence. It is released in the return
air duct of a furnace and monitored in the bedroom, living room, and kitchen.
Then the concentration of SFg is measured with time, and the air exchange
rates (air chanqes/h) computed using the following formula:
Q = V
InC0 - InC
At
60
mm
(41)
where:
Q -
V =
Co =
C =
At =
Air change rate (air chanqe/h)
3
Volume of the residence (ft )
SFg concentration at time t = 0
SFg concentration at time ® 't1
Elapsed time between t = 0 and t = 1t1 in minutes
In order to properly characterize Infiltration Into residences, data
on climatic conditions is collected. Measurements are taken on an hourly
basis to provide readings on wind speed, wind direction, inside-outside
temperature, and relative humidity. Equipment and techniques used to
obtain climatic data are discussed in Volume II, Chapter 1.
-146-
-------
In order to calculate the infiltration energy use, the SFg measure-
ments, indoor/outdoor dry bulb temperatures, and indoor/outdoor relative
humidity are utilized. First the SF6 measurements are used directly to
identify the infiltration rates (air chanqe/h). Then, usinq the indoor/
outdoor dry bulb temperatures, the sensible heating and cooling loads are
calculated. Depending on season (heating or cooling) and presence of
humidifiers, relative humidity is used along with the dry bulb temperature
to calculate both sensible and latent heat loads due to infiltration.
Based on the performance of heating/coolinq equipment, the energy use
associated with the infiltration loads can be determined.
CALCULATION OF ENERGY USE
The energy use due to the infiltration is estimated by the following
ec uations:
For Heatinq
" 0.01B7(Q) (V) (Tr-T0)
QH = E
where
Qh = Heating energy use due to infiltration energy
£ = Heating system efficiency
0.075(H -H ) (Q) (V)
where
Qc = Cooling energy use due to infiltration (Btu)
Hr ¦ Indoor enthalpy (Btu/lbm)
Ho = Outdoor enthalpy (Btu/lbm)
Ec = C.O.P. of cooling system.
As previously mentioned, one of the objectives of this study is to
develop an infiltration model that will relate to residential air quality.
The development of such a model includes a surwnary of steps already taken.
First, the building's physical parameters that can affect infiltration
rates are identified. Based on the building's physical parameters, EOA
is computed as defined. The "orifice coefficient" is then computed by
normalizing the calculated EOA. The measured infiltration rates (air
exchange rates) can then be normalized with respect to the calculated
"orifice coefficient." Finally, the overall model has to be specified.
-147-
-------
The infiltration rate is calculated by use of the following model:
0.66
Q = OC
A+B Tr-T0 + CV2
where
Q = Air change rate (air chanqes/h)
OC - Orifice coefficient
Tr = Indoor temperature
T0 = Outdoor temperature
V = Wind velocity (mph)
The parameters A, B, and C have been evaluated empirically for a few
residences. These values can be used as a first estimate for other resi-
dences as well.
The parameters have been evaluated for sinqle-family detached (SFD),
sinale-family attached (SFA), and mobile home (MH) type structures based
on data available from SFg measurements. The model was then used to
calculate the infiltration rates for the hours for which SF6 measurements
were not available.
Data from three SFD structures is used in the evaluation of the
following model. This correlation should be utilized only in the evalua-
tion of single family detached homes.
OC 3 Orifice coefficient
Tt = Indoor temperature
T0 = Outdoor temperature
V = Wind velocity (mph)
Correlation coefficient = 0.95.
Once the air change rate is known from Equation (44), it is a simple
matter of using Equations (42) or (43) in order to evaluate the energy
due to infiltration.
0.66
Q = OC 950 + 6135 Tr-TQ + 592V
where
Q = Air change rate (air changes/h)
-148-
-------
The hourly total enerqy use, conditioning energy use, and infiltra-
tion energy use are plotted in Figure 34 for the Baltimore conventional
residence. As can be seen in this graph, the conditioning energy use
represents a high percentage of the total enerqy use during the entire
day, while the infiltration energy use accounts for 50 to 87% of the
total energy use. The high percentage occurs during the hours at which
the internal load (mainly cooking) and solar radiation heat gain are
highest for the day.
Similar energy use patterns are given in Figure 35 for the Washington
conventional residence. It is noticed here that the conditioning energy
use in the summer also represents a high percentage of the total energy
use during the day, except for those hours in the morning when the outside
temperature drops to its lowest value. The infiltration energy use accounts
for a smaller percentage of the total energy use than that given for the
winter in the previous graph. This is due to two factors: first, the
infiltration rates were low for this residence compared to those of the
Baltimore conventional residence; second, the negative effect of internal
loads in the summer reduces the percentage of infiltration load from the
total load.
Similarly, a preliminary model has been developed for single family
attached structures more commonly known as townhouses. The correlation
developed resembles the following:
Q = OC 3592 + 28539
0.66
VT0
+ 1659V'
(46)
where
Q = Air change rate (air changes/h)
OC * Orifice coefficient
Tr s Indoor temperature
Tq = Outdoor temperature
V = Wind velocity (mph)
Correlation coefficient = 0.9.
Again, energy use data is presented in Figure 36 for the Pittsburgh low-
rise apartment. This graph indicates that the filtration energy use between
the hours of 1500 and 1800 is larger than the total conditioning energy use.
This simply says that the internal load and solar gain during those hours is
higher than that due to conduction and infiltration. Therefore it could be
stated that the infiltration energy use could reach 100% of conditioning
energy use in some residences.
-149-
-------
Figure 34. Energy use profile for Baltimore conventional residence on January 31, 1977.
-------
0400 0800 1200 1600 2000 2400
TIME OF DAY, hours
Figure 35. Energy use profile for W ashing ton conventional residence on July 10, 1977.
-151-
-------
TIME OF DAY, hours
Figure 36. Energy use profile for Pittsburgh low-rise apartment on April 7, 1977.
-------
A model has been also developed for mobile homes, using one of the
homes monitored in this study. A preliminary finding indicates the
existence of the following relationship:
OC = Orifice coefficient
Tr = Indoor temperature
T0 = Outdoor temperature
V = Wind velocity (mph)
Correlation coefficient = 0.9.
I . may seem strange that the "A" term is negative, but H. Ross (U.S. Depart-
ment of Energy, private communication, 1977) has indicated that Princeton
University has come up with a similar finding.
The energy use patterns for mobile home No. 1 in Pittsburgh on
Feburary 17, 1977, are given in Figure 37. This figure Indicates that the
percentage of infiltration energy use 1n this mobile home 1s far less than
the other residences studied. One of the major elements contributing to
this fact is the relatively high skin load in this mobile home.
To summarize, many important relationships have been developed and
presented in regards to energy use and infiltration in residential struc-
tures. For the three types of residences monitored in this study, the
fell owing items have been culminated.
• Total energy use as computed by summing the Btu Input of all
fuel types utilized
• Space conditioning energy use as measured directly
• Infiltration rate (air changes/h) as measured by SFg data
• Identification of physical building paramaters
• Computation of the Equivalent Orifice Area (EOA)
t Normalization of the infiltration rates (evaluated from SFfi
data) with respect to the orifice coefficient
• Evaluation of the parameters for the specific infiltration models.
0.66
Q = OC
(47)
where
Q = Air exchanae rate (air chanqes/h)
-153-
-------
0 0400 0800 1200 1600 2000 2400
TIME OF DAY, hours
Figure 37. Energy use profile for Pittsburgh Mobile I on February 17, 1977.
-154-
-------
It is suqqested that more research be done to empirically verify the
infiltration models presented in this report. Furthermore, additional
experimental data is needed to expand the data base necessary for proper
correlation of the model. Once this has been accomplished, it will be
easier to estimate the infiltration rate of various residential struc-
tures .
-155-
-------
Section 5
RESIDENTIAL AIR QUALITY AND ENERGY
CONSERVATION MEASURES
Two important ideas about energy conservation and indoor air quality
in residences are shaping national decisions on conservation measures.
The first is that a residence shelters its occupants from high ambient
pollutant concentrations. The second is that measures to conserve energy
in a residence do not affect the indoor air quality. The facts are that,
while a residence under certain conditions may shelter its inhabitants
from high concentrations of some pollutants, under different conditions
its occupants may be exposed to concentrations higher than outdoor levels
for other pollutants. Also, energy conservation measures will change the
air quality characteristics of a residence; these changes may affect air
quality either adversely or beneficially.
ENERGY AND COST SAVINGS THROUGH AIR EXCHANGE RATE REDUCTION
Three mechanisms of heat exchange between indoors and outdoors deter-
mine the energy conserved in the residential environment: 1) heat con-
duction; 2) heat transmission through the windows; and 3) heat transmission
through door and window openings or structural cracks. The American Society
of Heating, Refrigerating, and Air Conditioning Engineers (ASHRAE, 1972)
defines air infiltration, the third heat exchange mechanism, as the rate of
air flow into and out of a building without the interference of the building's
occupants. Air infiltration must be distinguished from natural ventilation
which refers to air changes induced by the inhabitants through specified
openings, such as windows, doors, and by ventilators. The majority of resi-
dences do not have mechanical systems that control the air exchange rate.
The air exchange rate measured by the tracer dilution technique used in this
project is a comprehensive term: it estimates the air flow rate into and out
of a residence caused by air infiltration and by intentional air movement
(natural ventilation) through windows and doors. Energy conservation mea-
sures carried out to reduce heat conduction and heat transmission through
the windows do not substantially affect the indoor air quality. These
measures are essential elements in the effort to conserve energy in resi-
dences and will be discussed later in this section.
The physical quantity that associates energy conservation measures with
indoor residential air quality is the air exchange rate. Air infiltration
is the important link in developing a scheme for monitoring both minimal
energy waste and acceptable air quality in residences. The air exchange
rate \> is the signature parameter which may simultaneously represent energy
and air quality measures.
Infiltration heat losses and heat gains in residential structures are
affected by the summation of the following forms of air flow mechanisms:
1) infiltration-exfiltration; 2) infiltration due to door and window
-156-
-------
openings; and 3) infiltration due to operation of the HVAC systems. As noted
earlier, a fourth mechanism, mechanical control of air exchange rates, is
not applicable to residential buildings. Infiltration is the largest
factor affecting cooling and heating loads in residences. During the
heating season, depending on how well a structure is insulated, 14 to 53%
of the total heat losses are due to infiltration. In the cooling season,
11 to 27% of the total structure heat gains are due to infiltration
(Hittman Associates, 1977).
The structural components having the greatest effect on the infiltration
rate are windows and doors because they have the largest potential for air
leakage. Consequently, typical approaches to effectively reduce infiltra-
tion rates are weatherstripping around doors and windows, caulking around
window and door frames, and the addition of storm doors and windows. Infil-
tration losses and gains can be reduced by as much as 30% through the imple-
mentation of these measures. Table 45 illustrates an example of a 1500 ft2
house analyzed in three climatic regions. The air exchange rate reference
level for this residence is 0.8 air changes per hour; this is considered
to be the most frequently occurring rate for existing residential structures.
An air exchange rate of 1.2 air changes per hour is incorporated in
Table 45 for comparative reasons. In order to conserve energy, the infil-
tration rate is first reduced by one-third, then by one-half, and finally
it is reduced to one-quarter of its original reference level of 0.8 air
changes per hour. Reduction of the air exchange rate, an energy conservation
measure, drastically affects the infiltration heat losses during the heating
season, and the infiltration heat gains during the cooling season. The heat
losses or gains are decreased to about 60% of their original values in all
three climatic locations. Table 45 shows the estimated savings that may be
obtained by reducing air infiltration. The first five columns in Table 45
are self-explanatory. The sixth column, Infiltration Energy Savings, pre-
sents the energy savings 1n number of therms associated with the various
reductions in the air exchange rate. The next column, Dollar Savings,
denotes the estimated dollars that could be saved If the designated reduction
occurred during the current year. Maximum Implementation Investment, the
next column, estimates the maximum investment the homeowner would undertake
to save the corresponding number of therms per year over an assumed period
of time. Incorporated in this estimate is an 11% home improvement loan over
5 y and a fuel escalation cost of 5% per year. The last column, Projected
Implementation Cost, indicates the current material and construction costs
required to retrofit the simulated residences and reduce the current air
exchange rate of 0.8 to the desired levels.
It can be seen that the maximum amount of permissible energy conserva-
tion Investment ranges from $500 to a high of $3900. Depending on the type
of HVAC system, local fuel costs, and amount of desired Infiltration reduc-
tion, the most that can be spent must fall within this range for the example
1500 ft^ single-family residence. The material and construction costs range
from $500 to $1600, again depending on the desired amount of reduced Infil-
tration. Present construction and material costs will allow only a one-third
-157-
-------
TABLE 45. ENERGY SAVINGS THROUGH REDUCED INFILTRATION FOR A SINGLE-FAMILY DETACHED RESIDENCE
Air Change
Rate*
Air Chinge
Race Reduction
InflltrattonA
Loads (10s BTU)
HT CL
HVAC
Energy
Forms
Infiltration Energy^
Saving. (Low.) D c
'1Q5BTU1 Saving,
(Lone.) ($)
HT
CL
Maximum*5 Projected
Implementation Implementation
Investment ($) Coit($)
Baltimore
1. 20
428
131
Cas/Elec
(316)
(30)
* (It)
N/A
N/A
Oll/Elec
(302)
(30)
(27)
N/A
N/A
Electric
(217)
(30)
(302)
N/A
N/A
0. 80
235
74
0. 50
Approximately
159
51
Cas/Elec
125
12
$ 44
$ 588
$ 505
One-Third
Oll/Elec
119
12
57
760
505
Electric
85
12
114
1523
505
0.40
One-Half
126
41
Cas/Elec
179
17
$ 63
$ 840
$ 830
Oll/Elec
170
17
81
1083
830
Electric
122
17
164
2182
830
0.20
One Quarter
95
31
Cas/Elec
234
23
$ 83
$1108
$1610
Oll/Elec
219
23
106
1410
1610
Electric
157
23
212
2825
1610
Chicago
1.20
572
98
Cas/Elec
(412)
(22)
S<115)
N/A
N/A
Oll/Elec
(392)
(22)
(173)
N/A
N/A
Electric
(281)
(22)
(385)
N/A
N/A
0.30
325
56
0. 50
Approximately
225
39
Gas/Elec
167
9
$ 47
$ 662
S 505
One Third
Oll/Elec
159
9
90
939
505
Electric
114
9
157
2103
505
0.40
One-Hall
173
30
Gas/Elec
2S3
14
$ 71
$ 948
$ 330
Oll/Elec
241
14
107
1430
330
Electric
172
14
238
3181
830
0. 20
On* Quarter
139
22
Cas/Elec
310
18
$ 88
$1172
$1610
Oll/Elec
295
18
132
1758
1610
Electric
211
18
293
3893
1610
Houston*
1.20
154
282
Gas/Elec
(150)
(61)
$ (99)
N/A
N/A
OU/Elec
Note E
(61)
Note E
N/A
N/A
Electric
(90)
(61)
(118)
N/A
N/A
0.80
73
173
0.50
Approximately
45
126
Gas/Elec
51
26
$ 38
$ 501
$ 505
One Third
OU/Elec
Note E
26
Not* E
Note E
Note E
Electric
31
26
44
593
505
0.40
One-Half
36
109
Gas/Elec
68
36
$ 51
$ 682
$ 830
Oll/Elec
Note E
36
Note E
Note E
Note E
Electric
41
36
60
801
830
0.20
One Quarter
25
88
Gas/Elec
89
49
$ 68
$ 911
$1610
OU/Elec
Note E
49
Note E
Note E
Note E
Electric
53
49
80
1058
1610
* Example of a Southern City.
NOTE A: Weather Bat*
Yearly Heating Degree Oayi
Baltimore 4425
Chicago 6013
Houston 1387
Yearly Cooling Degree Day)
NOTE B: Include, the following HVAC Seasonal Efficiencies
Baltimore
Cm 0.61
Oil 0.64
Electric (ReaMance Heat) 0. 89
Electric (Air Conditioning) 1.90
NOTE C: Energy Costs (I Dollan/Tharat)
Gai Oil
Baltimore 0.24 0.36
Chicago 0.21 0.37
Houston 0.34
NOTE D: Aitumes 9% load tot 30 y,
NOTE E< Oil la art a source of haat in the Hounon area.
Chicago
0.60
0.63
0.88
1.89
Electricity
1.18
1.28
0.78
Baltimore
Chicago
Houston
Houtton
0. 54
Note E
0.90
1.79
1112
983
2570
-158-
-------
reduction in air change rate from the assumed base value of 0.8 air changes
per hour in warm regions. Furthermore, it is only in the colder climatic
regions that an additional reduction in infiltration can be cost-effective
for HVAC systems utilizing the three fuel types. A reduction in the air
exchange rate is seen to be cost-effective if the maximum implementation
investment is greater than or equal to the projected implementation cost.
All values included in this table have been generated by numerical models;
the output is subject to the constraints of the model used for simulations.
The techniques used in arriving at the above estimations are explained
in the literature (Hittman Associates, 1977). The estimates included in
Table 45 may be thought of as indicative of the effects that reduction of
the air exchange rate would have in the Baltimore experimental and the
CMcago conventional residences. The Houston residence is included as an
example of a southern city with different climatic and financial inputs.
The magnitude of the quantities illustrated is not as important as the
relative change estimated. From this perspective the application of presently
known insulation, weatherstripping, and caulking techniques reduce the infil-
tration rate and conserve energy. These energy conservation measures are
ccst-effective if the air exchange rate of a residence is reduced from an
average value of 0.8 air changes per hour to about 0.5 air changes per hour.
Additional energy will be conserved if further reduction of the infiltration
rate is carried out, but it may not be cost-effective.
AIR QUALITY IMPACTS OF ENERGY CONSERVATION
The data base collected by the field monitoring program of this project
and the numerical simulations with the GIOAP model demonstrate that the impact
of reducing the air exchange rate on residential air quality depends on two
additional elements: 1) the strength of an indoor pollutant emission source,
and 2) the chemical nature of the pollutant. The relative Importance of
these factors on the indoor air quality levels will be estimated with numeri-
cal simulations.
Gas cooking appliances are not the only sources of Indoor-generated
gaseous pollutants; however, they are major contributors to the observed
indoor concentrations of such pollutants. CO will be utilized as the inert
pollutant that indicates the effect of reducing air exchange rates on the
residential air quality. Several Input parameters of the GIOAP model are
assumed:
1. A typical 24-h variation of the ambient CO concentrations 1s
illustrated 1n Figures 38 and 39. This Is the average variation
for downtown Los Angeles and may not be typical for a residential
area. However, 1t Is chosen to specifically Illustrate the points
made 1n the following few pages.
2. One burner 1s emitting CO between the time periods of 0700 and
0800 hours, and 1200 to 1300 hours; one burner and the oven are
emitting CO for the 2-h period between 1600 and 1800 hours;
-159-
-------
I
a,
." 20
O
< 18
fS
| 16
§
U 14
o$
8 12
Q "
s
10
8'
6
4
2
0
) I I 1 I I I I I I I I I I I I I I I I I I I
0100 0200 0300 0400 0500 0600 0700 0800 0900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 24
TIME OF DAY, hours
Figure 38. Effect of reducing the air exchange rate in a residence with indoor CO sources.
-------
30
28
26
.24
22
20
18
16
14
12
10
8
6
4
2
0
• Outdoor
O i' = 0. 2
~ v - 0. S
A v = 0. 8
0 v = 1.2
I I I I I I i I I I I I I I I I I I I I I I I
0100 0200 0300 0400 0500 0600 0700 0800 0900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 24<
TIME OF DAY, hours
Figure 39. Effect of reducing the air exchange rate in a residence without indoor CO sources.
-------
during the remaining hours CO emissions are from the pilot
light. The emission rates corresponding to these indoor CO
sources are obtained from the literature. Emissions from the
gas furnace are assumed to be negligible. Residences with
electric cooking appliances are assumed to have zero or very
low CO indoor sources.
•3
3. The volume of the residential structure simulated is 15,000 ft .
4. One representative air exchange rate is used throughout a 24-h
period.
The effect of reducing the air exchange rate in a residence with indoor
CO sources versus one without indoor sources is illustrated in Figures 38
and 39, respectively. Simulations of a residence with an indoor pollutant
source, Figure 39, shows that, in the early morning hours when the simulated
outdoor levels are high and the indoor sources are weak, reduction of the
air exchange rate, denoted by v, decreases the indoor CO concentrations.
However, later during the time period between 1600 and 1800 hours, when the
ambient CO concentrations are low and the indoor emission sources are strong,
the effect is the opposite, and reduction of the air exchange rates leads
to higher indoor CO levels. This morning and late afternoon-early evening
behavior has been noticed in the monitoring study and can be explained by
physical principles. The high outdoor levels remain outside because
reduced air exchange rates imply that less air infiltrates indoors. The
afternoon deterioration of the indoor air quality is due to the "trapping,"
lack of dispersion, of the indoor-generated CO. The "trapping" of the
pollutant in the indoor environment of an electric residence is also illus-
trated 1n Figure 39 which shows that the reduction of the air exchange rate
has a less pronounced effect on the overall indoor air qualtiy. The cumula-
tive 24-h exposure is a parameter of interest because it illustrates the
total effect of reducing infiltration rates. The cumulative exposure is
measured in units of concentration-hours and is equivalent to the area under
the curve of concentration vs. time. Table 46 shows the ambient exposure and
the exposures for the four air exchange rates considered.
TABLE 46. ESTIMATED 24-HOUR INDOOR CARBON MONOXIDE EXPOSURE*
Residences with Gas Cooking Appliances
Residences with Electric
Typical Operation of Indoor
Cooking Appliances
Air Change Rate
CO Sources
No Indoor Sources
Ambient
226
226
1. 2
247
223
O. 8
258
222
0. 5
275
221
0. 2
320
210
* Units ppm-hours.
-162-
-------
CO is an inert pollutant; the effect of reducing the air exchange rate
on the concentrations of a relatively reactive pollutant is quite different.
Owing to the reactive nature of NO?* indoor concentrations of the pollutant
are generally lower than corresponding outdoor levels. If the indoor NO2
emission source is strong, it is likely that for a short period of time,
during the operation of the source, the indoor N0« concentrations will be
higher than the outdoors. Simulations, similar to the ones described in
detail for CO, demonstrate that for a typical daily variation of ambient NO2
concentrations, the outdoor 24-h exposure is 4188 yg/m3-hours, the range of
indoor exposure with gas cooking appliances varies from 2067 to 2873 yg/m3-
hours, corresponding to residences with air exchange rates of 0.2 and 1.2
air changes per hour, respectively. For the same ambient conditions, the
indoor 24-h exposures of an electric house with no NO2 emission rates vary
from 479 to 1972 yg/m3-hours. The low cumulative exposure is manifested in
the residence with 0.2 air changes per hour, while the other end of the range
refers to simulations of a residence with 1.2 air changes per hour.
The data base collected for this project, combined with the results of
numerical simulations carried out with the GIOAP model, demonstrate that
from the perspective of air quality, reduction of the air exchange rate,
an energy conserving measure, may lead to deterioration of the residential
air quality. If the pollutant under consideration is inert and is not
generated indoors, the effect of decreasing the air exchange rate is not
substantial. If the pollutant is inert and, in addition to infiltrating
from outdoors, is generated indoors, reduction of the air exchange rate
increases the indoor contaminant levels. On the other hand, if the pollu-
tant is reactive and is not generated indoors, reduction of the air exchange
rate will decrease the indoor pollutant concentrations. Finally, if the
pollutant is reactive and is generated indoors, reduction of the air exchange
rate will decrease the indoor pollutant concentrations. The major source of
indoor-generated pollutants, cooking with gas appliances, emits both inert
and relatively reactive pollutants. Thus, a decrease of the air exchange
rate in a residence with gas appliances results in an overall increase of
the CO and NO indoor concentrations and a decrease of the NO2 concentrations.
It is apparent that reduction of the infiltration rate leads to energy
conservation and often to deterioration of the residential air quality.
However, further examination gives rise to a "desirable" air exchange rate
which takes under consideration the following factors Involved 1n the complex
residential energy-air quality system:
• The relative changes of the Indoor air quality Induced by the
reduction of the air exchange rate,
t The cost estimates involved in carrying out energy conservation
methods balanced against the financial gains to be realized by
the measure, and
• The national requirements to conserve energy and protect the
welfare of individuals.
-163-
-------
An air exchange rate of 0.5 air changes per hour is suggested as a
practicable level for existing houses with gas cooking appliances. This
air exchange rate is derived on the basis of presently available retrofitting
techniques, cost estimates, and pollutant emission rates. A residential air
exchange rate of the proposed magnitude represents a reduction of the current
average residential rate of 0.8 air changes per hour; it is cost-effective
(recall discussion of Table 43) and does not induce drastic changes in the
indoor air quality. In houses with electric cooking appliances, a major
source of indoor pollutant emissions is eliminated; therefore, energy con-
servation and cost-effectiveness become more important in the deterioration
of a practicable air exchange rate. Regardless of this change of emphasis,
the practicable air exchange rate remains 0.5 air changes per hour because
of pollutants that are not emitted by gas appliances.
Several points must be made with respect to the practicable air exchange
rate:
1. The practicable air exchange has been determined by considering
changes in the magnitude of the pollutant concentrations but not in the
resulting health effects. This approach is preferred because the health
effects due to increasing indoor pollutant concentrations have not been
studied extensively and are not well understood.
2. The numerical value of the practicable air exchange rate should
be thought of as a small range varying between 0.4 and 0.6 air changes
per hour.
3. The practicable rate does not imply that current construction
techniques capable of building dwellings with very low air exchange rates
(0.1 air changes per hour) should not be used; rather it states that in
such super tight-energy conserving residences special care must be taken
to keep indoor pollutant concentrations from reaching levels that exceed
the EPA promulgated NAAQS.
AIR QUALITY CONTROL MEASURES
The practicable air exchange rate is one of many means suggested for
controlling residential indoor air quality. However, techniques can be
applied that may reduce the levels of indoor air pollutants. If 1n fact
these measures are implemented, a further decrease in a structure's air
exchange rate could be possible, which would result in additional energy
conservation. The following discussion investigates several measures to
control residential indoor air quality.
Indoor-generated pollutants are dispersed by infiltrated air. If the
emission rates of gas appliances remained at current levels while sharp
decreases in air exchange rates occur, indoor air pollutants could accumu-
late to potentially dangerous levels. Gas appliances emit NO, N02f and CO
internally, and elevate residential concentrations of these pollutants.
There 1s an agreement among relevant studies on the emission rates of indoor
-164-
-------
pollutant sources; Table 47 (Cote et al. 1974) is an example of pollutant
emission rates from gas appliances. Research must be directed towards
reducing the emission rates of indoor pollutant sources. Substantial
decrease of these emission rates may lead to further reduction of the
practicable residential air exchange rate and conserve additional energy.
One approach towards reducing pollutant levels may be found by venting
pollutant-generating appliances. At first review, this is an attractive
idea, but history has proven otherwise. Approximately 40 y ago, vented gas
ovens/stoves were being produced in this country. There were even isolated
cases of vented gas ovens/stoves being produced circa 1946 because standards
set up by the American National Standards Institute (ANSI) and the American
Gas Association (A6A) called for manufacturers to make them available if they
were so requested by consumers. About 1950 this standard was dropped, and
the vented oven has not been manufactured since. Economics and consumer
demand led to the demise of this device. The draft diverter, which is
necessary for proper stack flow, had to be installed approximately 6 in.
from the stove, with the chimney or stack being installed an additional 6 in.
from the wall. This resulted in the stove being at least a foot away from
the wall. This was necessary to eliminate any possible fire hazards, but,
as a result, it increased costs because of the additional material and
installation procedures required. It also became an undesirable consumer
product because the stove had to be considered a separate element and there-
fore could not be built into the kitchen cabinetry (McGee, 1977). Further-
more, the makeup air required for the venting had to come from the kitchen
area, thus introducing an energy conservation problem.
It has been mentioned that as air exchange rates decrease the concen-
trations of indoor-generated air pollutants increase, thus leading to a
potentially harmful environment. A gas stove/oven combination is a major
source of CO. The installation of a CO detector will alert the inhabitants
of high indoor CO concentrations. This device would sense the levels of CO
in the kitchen and activate an alarm if the levels became dangerous. Venting
of kitchen appliances, opening the windows and curtailing cooking activity
would be some of the necessary steps towards reducing indoor CO levels.
Currently, such devices cost between $950 and $1100. The desire to conserve
energy and protect the public health appears to be a sufficient cause for
using the CO alarms 1n institutional buildings. In addition, the increasing
public awareness of possible health hazards from deteriorating Indoor air
quality combined with possible mass production of the devices will lower
present cost and Increase the accessibility and use of such alarms.
Baumelster (1967) suggests the following techniques for removing gaseous
pollutants from the air:
• Combustion of pollutant gases sometimes coupled with a
catalytic converter
• Absorption of the gas by a solution
-165-
-------
TABLE 47. SUMMARY OF POLLUTANT EMISSION OF GAS APPLIANCES FOR SEVERAL
TYPICAL OPERATING CONDITIONS IN HARTFORD DWELLINGS
Pollutant Emission Rates
Heat Input mg/Hr
Rate
Appliance Operation (Kcal/h) NO NO,, CO
Older Gas Stove
Pilot Lights:
150
6.8
8. 2
62.9
with Cast Iron
1 Burner - High Flame
2700
250.0
140. 0
1031.0
Burners
3 Burners - High Flame
6780
793.0
494. 0
3220.0
Oven;
Transient
2300
361.0
366. 0
4117.0
Steady-State
2200
201.0
161. 0
1166.0
Broiler:
Transient
3000
411.0
369. 0
4050. 0
Steady-State
3800
337.0
184. 0
3108.0
Newer Gas Stove
Pilot Lights:
100
0. 5
1. 9
84. 2
with Pressed
1 Burner - High Flame
3500
450.0
277.0
1785.0
Steel Burners
3 Burners - High Flame
10200
1408. 0
669. 0
3213. 0
Oven:
Transient
4000
1324. 0
316.0
4040.0
Steady-State
2200
171.0
111. 0
3564. 0
Broiler:
Transient
4900
617.0
395.0
4145. 0
Steady-State
3700
503. 0
211.0
2800.0
Unvented Space
Low Flame-Steady-State
2800
214.0
130. 0
1770. 0
Heater
High Flame-Steady-State
6200
837.0
272.0
1978. 0
Domestic Gas
A pprox.
Furnace*
3000
2700.0
1080. 0
* From Compilation of Air Pollution Emission Factors (Revised), U.S. EPA, February 1972.
-166-
-------
t Absorption of the gas molecules to a surface of a solid
# Condensation of a vapor by regulating temperature and pressure.
These methods of cleaning gas from an airstream are often used to
remove pollutants from a waste airstream; however, due to their energy
requirements, high operating temperatures, high initial and maintenance
costs, and large-scale requirements, they are not suited for application
as residential gas cleaners. Research must be directed toward improving
and applying control techniques to remove gaseous pollutants from the
residential environment.
Control of indoor-generated pollutants may be accomplished by a range
hood that filters the pollutants as they are generated. Currently, the
Minimum Property Standards published by the U.S. Department of Housing and
Urban Development (1973) recommends that a range hood must be equipped with
a minimum capacity of 40 ft3/m1n per linear foot of hood length in kitchens.
However, questions have arisen about the efficiency of these devices to
effectively filter out the gaseous pollutants generated. Additionally,
these units generate considerable noise during operation, and thus tending
toward infrequent use. Research is warranted towards improving the efficiency
of the hoods and reducing the noise resulting from their operation.
Additional pollution is created by certain lifestyle activities and
hobbies. Smoking, cooking, and general indoor activities that may result
in reentrainment of settled dusts are major sources of indoor suspended
particulate matter. Generally, TSP may be classified as solid particulates
or liquid particulates.
Solid particulate pollutants consist of dusts, fumes, and smoke. Dusts
are solid particles, typically smaller than 100 ym, that are carried airborne
by natural or mechanical processes. Fumes are solid particles formed by the
condensation from the gaseous state. Generally, the particle size of fumes
is less than 1 ym. Smoke is small solid or liquid particles, from 0.1 to
0.3 ym, generated by the incomplete combustion of organic materials.
Liquid particulate pollutants are made up of mists and fogs. Mists are
small airborne droplets of material, less than 10 ym, that normally are
liquids. They can be formed by spraying, mixing, atomizing, or by chemical
reaction. Fogs are airborne droplets, ranging from 2.0 to 80.0 ym, that are
formed by condensation of vapors.
A separate classification of airborne pollutants is usually created for
biological organisms, which range 1n size from submlcroscoplc viruses to the
largest pollen grains. Bacteria range from 3.0 to 30 ym, spores from 1.0 to
10 ym, and pollen from 10 to TOO ym (Honeywell, 1976).
-167-
-------
The size of particulate pollutants commonly found in residences ranges
from 0.001 to 100 ym in diameter. Visible particles, which make up less than
10% of the total airborne particles, tend to settle on horizontal surfaces.
The remaining 90% of airborne particles are invisible and tend to deposit
by contact with both horizontal and vertical surfaces. Both groups contribute
to the dirt that causes most of household cleaning activity.
Four major methods are used to clean air of particulate matter:
• Impingement systems that use a coarse filtering medium
coated with a viscous substance on which the pollutant
particles adhere,
• Unit or panel filters that use a densely packed medium that
traps particles out of the air stream,
• Diffusion systems that drive particles from the main stream
by random air movements and forces them against fiber filters,
and
i
• Electrostatic precipitators that remove particles by
charging them, then collecting them on an oppositely
charged plate.
Each of the systems has been used in industrial and commercial build-
ings for removing particulate matter of various types and sizes, but only
two systems have been used in residential buildings; the panel filter and
the electrostatic precipitator. The other two systems have not been used,
primarily due to their relatively high operating costs as well as their
large space requirements.
Typically, these air-cleaning devices are not located directly within
the space being conditioned. Therefore, it is necessary to circulate air
from the conditioned space through the air cleaner and back to the space.
In order to provide adequate cleaning, the distribution system should draw
air from and return air to all areas within the residence.
Three operating characteristics distinguish the types of air cleaners:
efficiency, air flow resistance, and dust-holding capacity. Efficiency
measures the ability of the air cleaner to remove particulate matter from
the air stream, averaged over the life of the cleaner. Air flow resistance
is the measure of the static pressure drop across the air cleaner at a given
air flow rate. Dust-hold1ng capacity 1s the measure of the ability of an air
cleaner to hold a particular type of dust. A complete rating of an air
cleaner would then require data on the efficiency, air resistance, and dust-
holding capacity, as well as the effect that dust loading of the air cleaner
has on both efficiency and air resistance.
-168-
-------
Panel filters are flat, narrow units typified by the commonly used
fiberqlass home furnace filter. The media used in panel filters are random
fiber mats or blankets of varying fiber sizes and densities packed 1 or
2 in. thick and held by a rigid frame. In some designs the filter medium
is permanent, requiring periodic cleaning. In some designs the entire
panel is discarded after it has accumulated a dirt load.
Panel filters operate primarily on the straining principle; the
filter fibers are so dense that the air passages are smaller than the
diameter of the particles. Therefore, the particle becomes trapped and
is strained out of the airstream. In addition, some particles are removed
by adsorption to the fiber surfaces.
Operating characteristics of residential panel filters are given in
Table 48. The straining operation of the filters results in a high filter-
ing efficiency as long as the particle size is large (greater than 10pm).
Smaller particles tend to pass through the filter without becoming trapped.
The dense media used in panel filters result in good dust-holding
capacity while monitoring a relatively low resistance to air flow. As
the filter becomes dirt-loaded, its resistance increases from about 0.05 in.
H?0 to a replacement value of 0.5 in. ^0. While the filter itself
requires no energy to operate, as its resistance to air flow increases,
the load on the fan is increased, resulting in an increase in fan energy
requirements.
In addition, the ineffectivenss of the filter to remove particles
smaller than 10 pm would result in a gradual buildup of a layer of par-
ticles on the heat exchanger surfaces of the furnace, reducing its
operating efficiency.
TABLE 48. PERFORMANCE OF PANEL FILTERS
Characteristic
Panel Filters
Arrestance Efficiency
Dust Spot Efficiency
Dust Holding Capacity
Resistance to Air Flow
Initial Cost
Maintenance
Smallest Particle Removed
Power Consumption
85-90%
5-1096
90*180 gm/1000 ft'/min
0.05-0.5 in. H O
2
$ 5-20
Low (periodic replacement)
10,0 vim
0 W
-169-
-------
The initial cost of a panel filter system is low. Maintenance costs
are also low since the only maintenance required is the periodic replacement
or cleaninq of the filter "element. However, since the panel filter is effec
tive only on particles qreater than 10 ym, the units can remove only a
portion of the pollutants that are found in residences.
Electronic air cleaners are units that remove particles from the air
by using the principle of electrostatic precipitation. In residences
using forced air systems, the unit is mounted in the return air duct. In
other residences that do not use a central air system, individual room-
sized units are available.
Residential electronic air cleaners are two stage units. Return air
enters the first stage and passes through a strong electric field that
electrically charges (ionizes) any particles in the airstream. The air
and charged particles then enter the second stage which has an alternating
series of negatively and positively charged plates. The grounded plates
attract and hold the positively charged particles, removing them from
the airstream. In some units, a thin prefilter section made of cleanable,
fibrous material is installed in front of the electronic air cleaner to
remove large particles.
As the unit operates, a layer of dirt is gradually built up on the
collector plates. This dirt includes particles from tobacco tar, cooking
fats, and other viscous contaminants that act as an adhesive. To remove
this material 1t 1s necessary to periodically clean the unit using water
and detergent. Most units are equipped with indicator lights to signal
when the unit needs cleaning.
Operating characteristics of residential electronic air cleaners
are qiven in Taitlfi 49. The units are efficient devices for removing smoke
and fumes as well as atmospheric dust. An Important feature of the elec-
tronic air cleaner 1s its ability to achieve a high degree of air cleaning
efficiency without Introducing excessive resistance to air flow. Actual
resistance 1s about the same for electronic air cleaners and panel filters.
But unlike the panel filter, resistance does not Increase significantly
between the periodic cleanings. Therefore, the energy required to operate
the fan system remains nearly constant. In addition, since the system
is effective in removing a large ranqe of particle sizes, the furnace
heat exchanger surfaces would remain cleaner and would therefore operate
more efficiently.
The Initial cost of an electronic air cleaner 1s high when compared
to panel filters, but this Initial cost will be offset by the benefits
resulting from a cleaner environment 1n the residence. Since the elec-
tronic air cleaner efficiently removes dirt that can soil walls and furnish
1ngs, savings may be realized from a reduction in housekeeping expenses.
-------
TABLE 49. PERFORMANCE OF ELECTRONIC AIR CLEANERS
(ASHRAE 1975)
Characteristic Electronic Air Cleaner
Arrestance Efficiency 70-98%
Du»t Spot Efficiency 70-95%
Dust Holding Capacity High
Resistance to Air Flow 0.10-0.30 in. H^O
Initial Cost $150-300
Maintenance Low (periodic cleaning)
Smallest Particle Removed 0.01 ym
Power Consumption 20-60 W
Electronic air cleaners do require power to operate. This electric
use, which averages 20 to 60 W, would result in an increase in the
annual electric bill by less than $8.50, subject to market pricinq.
The two cleaning devices previously discussed offer varying costs
and efficiency of operating. While neither 1s best suited for all appli-
cations, a weighing can be made of the fallowing characteristics of each:
Advantages
Panel Filters
• Low Initial cost
Low maintenance
No power require-
ments
Electronic Filters
High efficiency on
large and small
particles
High efficiency on
smoke
Constant resistance
to air flow
Disadvantages
Low efficiency on
small particles
Resistance to air flow
Increases with dust
loading
High Initial cost
Requires power to
operate
-171-
-------
Advantages Disadvantages
Electronic Filters (Continued)
• Low maintenance
• Automatic indication
when unit needs cleaning.
Figure 40 shows some of the most common particulates found in the
residential environment and the effective range of panel and electronic
air filters for removing particulate matter.
o
8
o
PARTICLE DIAMETER, ym
o
o
Electromagnetic
Waves
Pollutants
Effectiveness of
\u Cleaners
C
c
X-Rays
Visible
XX
y Far Infrared
Ultraviolet Near Infrared
V
.Fumes
W
>c
Microw av :
Dust
Smog
X
Spray
XZ2D
V
ToE acco
C
Smok^
^""^|?lant Spores
^ollen^
Atmospheric Dust
>
K~»
( Bacteria
C3>
^ Electrostatic Precipitators
Panel Fitters
Figure 40. Characteristics of particulate pollutants.
-172-
-------
Energy Conservation Measures
As more energy conservation techniques are applied to residential
structures in the form of insulation, storm doors, storm windows, weather-
stripping, and caulking, the greater is the decrease in infiltration. Today
residential structures are being built or modified to resist infiltration
her»t losses during the winter and lower infiltration heat gains during
the summer. This is very beneficial in cutting energy usage of residential
structures, but it creates the potential problem of increased indoor air
pollution. However, several energy conservation measures can be implemented
with minimal effects on indoor air quality levels.
Energy management techniques in residential structures will result
in substantial reductions of energy consumption. In addition, structural
modifications that reduce conduction/radiation heat losses can be applied
to obtain considerable energy savings while exerting almost no effect on
the air exchange rate of the residence.
One such structural modification that can effectively reduce energy
consumption, while leaving air infiltration rates virtually unchanged, 1s
the reduction in thermal conduction through the roof/celling surfaces.
The heat loss through the roof/ceiling accounts for 8 to 14% of the total
structure heat loss in both cold and warm regions during the heating season.
The magnitude (not percentage) of these losses is larger in colder climates
and, therefore, will have a greater Impact on total energy cost. In a pro-
gram performed for ERDA (Energy Research and Development Administration),
Hittman Associates (1977) found that heat gains through the ceiling repre-
sent 2 to 5% of the total structure heat gains during the cooling season.
There are two items which can be implemented to improve the thermal per-
formance of roof/ceiling.
First, the thermal conduction through the roof/ceiling can be reduced,
and secondly, ventilation of the attic can be improved. Reducing the con-
duction 1s easily accomplished by the addition of more insulation. The
insulation may be In the form of batts, blankets, loose-fill, or foamed-
1n-place. The addition of Insulation increases the thermal resistance
of the roof/celling interchange. This thermal resistance, measured as
the R-value, 1s the reciprocal of the heat transfer coefficient or the
Inverse of thermal conductance. 011 or gas heating systems in colder
climatic regions should have an insulation value of R-30, while electric
heating system losses in similar climatic regions should have R-38 1n the
ceiling. Under the present fuel price schedules, new homes with R-19 in
the ceiling cannot justify the dollar savings resulting from any addi-
tional amounts of insulation, except 1n the coldest climatic regions of
in electrically heated homes.
-173-
-------
Attic temperatures can build up to 30 °F or more above the outdoor
temperature. This can increase the heat gain through the ceiling during
the summer if proper ventilation is not provided. If mechanical ventila-
tion through the use of an exhaust fan is installed in the attic, an
adequate amount of outside air must be supplied or the exhaust fan will
draw cool conditioned air from the space below, thereby increasing infil-
tration (Professional Builder, 1974).
Energy conservation can also be obtained by the reduction of thermal
conduction through the structure's walls. The heat loss through the walls
accounts for 12 to 19% of the total structure winter heat loss in both
cold and warm regions. The magnitude (not percentage) of those losses
will have a greater impact on total energy costs for residences in the
colder climates. During the cooling season, 3 to 5% of the total struc-
ture heat gain is due to heat flow through the walls (Hittman, 1977).
Insulating exterior walls is cost-effective only if there are high local
fuel costs, severe climatic conditions, no existing wall insulation, or
the structure is electrically heated. As a retrofit measure, the cost
and benefits are difficult to determine because exterior walls usually
cannot be insulated without puncturing the sheathing or the interior wall
surface. Usually, walls are only insulated if there is an air space of
at least 3 or more inches. Even with foamed insulation, problems arise
because of possible moisture accumulation, settling, shrinkage, and
inability to fill voids which minimize the effectiveness of insulation.
As a retrofit Item, additional wall Insulation 1s one of the least cost
effective of all those available.
A third structural modification which can be applied to residential
structures 1s the insulation of floors. The heat loss through the floors
represents 6 to 22% of the total structure heat losses. During the
cooling season, the floors are sources of 1 to 11% of the structure heat
losses (Hittman, 1977). The contribution of a floor over a nonheated
basement or enclosed crawl space to the heating load is quite different
from any other structural component because it is proportional to ground
tenperature, which 1s warmer than outdoor air in the winter, and usually
cooler in the summer. A heat loss through the floor In the summer is
a '•esult of the cooler ground temperature, and this helps to lower the
cooling loads. However, leaving the floor uninsulated to take advantage
of the floor's cooling effect in summer causes large heat losses during
the winter. Since the detrimental effect of heat losses in the winter is
larger than the beneficial effect of the heat losses in the summer, insula-
ting the floor has a net beneficial effect in cooler or more severe climatic
regions, especially when electric heat is utilized.
During the heating season, the windows affect the overall heat balance
in residential structures by losing heat to the outside through conduction
(which varies from 23 to 37% of the total structural heat losses) and
gaining heat by allowing solar radiation (which varies from 9 to 27%
-174-
-------
of the total structural heat gains) to penetrate into residences (Hittman,
1977). The combined effects are a net heat loss during the heating
season. Increasing the window R-value (i.e., adding double or triple
glazing and/or storm windows) will effectively decrease conduction heat
losses, and also lower radiation heat gains. A proportionally greater
amount of the heat losses will be decreased than the heat gains, thus
reducing the total heating load that must be met by the heating system.
During the cooling season, window conduction heat gains reinforce the
radiation heat gains, and as a result, make windows one of the largest
structural components affecting cooling loads.
As a further illustration, Figures 41 and A2 graphically show how
heat gains and losses are reduced for a 1200 ft single-family detached
residence in the Baltimore area utilizing increased insulation and the
infiltration parameters are described in Table 50. Techniques used to
arrive at these comparative diagrams are found in the relevant literature
(Hittman, 1977). These graphical illustrations show the relative change
in the magnitude of heating and cooling loads that can be realized through
various energy conservation measures.
ENERGY MANAGEMENT IN RESIDENCES
HVAC Controls
Application of energy management techniques 1n residential structures
can result in substantial reductions of energy consumption. Management
techniques which can be applied to residential structures are modifica-
tions to the heating, ventilating, and air conditioning (HVAC) systems
and their mode of operation. The following discussion will investigate
the potential for applying these modifications.
A simple adjustment of the thermostat control is the first, easy,
and rather effective energy management step. Nighttime temperature set-
back on a thermostat control 1s a very easy and effective method to reduce
energy usage. Utility companies and manufacturers have Indicated that
for each 1 °F that the thermostat is turned (town, a savings of 1% of
the annual fuel bill will result. Lowering the thermostat by 10 #F
during the night may reduce annual fuel bills by 10$ in northern
climates and 30% In southern climates. Nighttime setback can be a
manual or automatic operation. Automatic thermostats Include either
a clock or a clock-triggered resistor. The latter is used to retrofit
manual thermostats to automatic operation, while the former is best
used for new Installations.
An existing HVAC system can be retrofitted with an automatic flue
control to reduce energy consumption. Flue control in the form of an auto-
matically controlled flue damper conserves energy by retaining heat in the
heat exchanger of a residential furnace during nonoperatlonal periods so as
to minimize the off-cycle heat loss through the stack. If It Is assumed
that the flue damper can cut as much as 8055 of the off-cycle heat losses
-175-
-------
X
&
§ iOO
1 "50
£ 200
C
O
150
100
S
33 so
0
so
100
150
200
2S0
300
Si*50
400
gj450
5
—500 J
1/3 ^
uo
9 550
f-
2 600
T*
**•4
650
700
750
800
850
900
1000
.60
40
20
>"*
V5
as
TV PI c\ l
INSULATED
Q RESIDENCE
SO
3
Q
WELL
INSULA TED
RESIDENCE
Internal
1-0 ads
Window
Radiation
Inter;
Loads
nal
20
-40
JSO
-80
J00
20
40
J 60
JSO
200
220
¦240
260
180
<£
i
N.
(/)
Roof
Roof
Floor
Doors
Walls
Floor
/
Window
Conduction
/
Infiltration
/
/¦
/
Walls
/
/
/
/
s
Window .
Conduction
/
/
/
/
/
/
/
/
/
/
/
/
Infiltration
V
/
/
s
/
8
Figure 41. Single-family detached-Baltimore, heating (Hitman Associates, 1977).
-176-
-------
8
Figure 42. Single-family detached—Baltimore, cooling (Hittmaa Awociatei, 1977).
-177-
-------
TABLE SO. SPECIFICATIONS FOR A TYPICAL AND WELL INSULATED RESIDENCE (HITTMAN ASSOCIATES, 1977)
Typical Insulated Residence
Specifications
Component Conductance Comments
Roof
0.065
6" Loose Fill Insulation (R-13)
Doors
0.413
Wood Doors
Floor
0. 200
Uninsulated
Walls
0. 084
3-1/2" Glass Fiber Batts (R-ll)
Windows
1.01
Single Glazing
Infiltration P arameters
Yes No
W e atherstripping
X
Storm Windows
X
Storm Doors
X
Caulking X
Well Insulated Residence
Specifications
Component
Conductance
Comments
Roof
0.031
12" Glass Fiber Insulation (R-38)
Doors
0.102
Metal/Foam Sandwich and Storm Doors
Floor
0.033
10" Glass Fiber Insulation (R-30)
Walls
0.038
6" Glass Fiber Insulation and
2" Rigid Foam (R-31)
Windows
0.420
Double Glazing and Storm Windows
Infiltration
W e atherstripping
Storm Windows
Storm Doors
Caulking
Yet
X
X
X
X
No
-178-
-------
through the stack, a 4 to 7% reduction in the annual fuel consumption
will result in either northern or southern climates, though certainly
the magnitude of these savings will be noticed only-in the colder regions
regions (Hittman Associates, 1977).
A technique that can be applied to the HVAC combustion unit is that
of the intermittent ignition. Currently, many gas residential heating
systems are equipped with pilot lights. The substitution of these with
electronic ignition can result in a 3 to 6% savings in the annual fuel
bill in northern climates and a 15 to 22% savings in the southern
climatic regions.
Another technique that can be implemented with existing HVAC systems
is to lower the furnace fan shutoff temperature. By allowing the fan to
run longer after the furnace has shut off, the heat that would otherwise
escape up the chimney is available for space heating. The cost of running
the fan is in many instances offset by the 2-3% annual fuel savings.
A technique that has successfully been employed in large office and
commercial buildings and that should be investigated for residential struc
tures is that of a variable zonal ventilation system. This system yields
increased air exchange rates in individual rooms on a need basis. In the
residential environment, high levels of pollutants are generated in the
kitchen and other activity rooms. Use of the variable zonal ventilation
system in these rooms can reduce elevated pollutant concentrations by
mechanical ventilation near the point of production before emissions can
contribute significantly to the general household pollutant levels. This
approach may decrease total residential energy consumption. The major
disadvantage of such a system Is its expense, which is too high to make
it feasible for single family residential use.
Other modifications can be applied to the heating, ventilating, and
air conditioning systems in addition to those previously mentioned. The
majority cannot be considered as retrofit items but should be considered
whenever complete replacement of an existing system or the installation
of a new one is necessary. A 11st of some of these Items follows:
t Size heat Installation for peak winter load (most Installa-
tions are now sized for twice the peak load)
• Place all new ductwork within the conditioned space (up to
10% annual fuel savings)
• Use outside preheated combustion and dilution air (requires
ductwork to furnace or placing the furnace outside)
§ Increase the Coefficient of Performance (C.O.P.) of air con-
ditioning and heat pump systems (requires increased coll
effectiveness, higher compressor efficiencies, and higher
motor and fan efficiencies).
-------
Indoor air pollution is a complex function of internal volumetric air
change. A high air exchange rate often cleanses the indoor air of pollu-
tant buildup; however, increased air exchange rates result in additional
energy consumption. A balance must be obtained between energy consumption
and residential indoor air quality.
Problems associated with energy conservation measures and air quality
in residences have been outlined in this section. In addition, the number
of scenarios suggested demonstrates that substantial reductions of the
energy consumed by a typical household are possible with minimal impact
on the indoor residential air quality.
-180-
-------
REFERENCES
Altshuller, A.P. 1973. Atmospheric Sulfur Dioxide and Sulfate, Distri-
bution of Concentration at Urban-Nonurban Sites in the United States.
Environ. Sci. Tech. 7(8):709-12.
American Conference of Governmental Industrial Hygienists (ACGIH). 1976.
Threshold Limit Values for Chemical Substances and Physical Agents in
the Workroom Environment with Intended Changes for 1976. American
Conference of Governmental Industrial Hygienists, Cincinnati, Ohio.
American Society of Heating, Refrigerating and Air-Conditioning Engineers,
Inc. 1972. ASHRAE Handbook of Fundamentals. American Society of
Heating, Refriqerating and Air-Conditioning Engineers. 3rd Edition
1976; New York, N.Y.
American Society of Heating, Refrigerating and Air-Conditioning Engineers,
Inc. 1975. ASHRAE Equipment Handbook. American Society of Heating,
Refrigerating and Air-Conditioning Engineers, New York, N.Y.
American Society of Heating, Refrigerating and Air-Conditioning Engineers,
Inc. 1977. 1977 Fundamentals. American Society of Heating, Refrigerat-
ing and Air-Conditioning Engineers, Inc., New York, N.Y.
Andersen, I., G.R. Lundqvist and L. Molhave. 1975. Indoor Air Pollution
Due to Chipboard Used as a Construction Material. Atmos. Environ.
9:1121-27.
Baumeister, T. 1967. Standard Handbook for Mechanical Engineers. McGraw-
Hill Book Company, New York, N.Y.
Bevington, P.R. 1969. Data Reduction and Error Analysis for the Physical
Sciences. McGraw-Hill Book Company, New York, N.Y.
Calder, K.L. 1957. A Numerical Analysis of the Protection Afforded by
Buildings Against BW Aerosol Attack. BWL Technical Study No. 2.
Office of the Deputy Commander for Scientific Activities. Fort Detrick,
Md.
Colucci, A.V. 1976. Sulfur Oxides: Current Status of Knowledge - Final
Report. EPRI Report No. EA-316. Prepared for Electric Power Research
Institute under Research Project 681-1 by Greenfield, Attaway & Tyler,
Inc., San Rafael, Calif.
Cote, W.A., W.A. Wade, III, and J.E. Yocom. 1974. A Study of Indoor Air
Quality. EPA-650/4-74-042. U.S. Environmental Protection Agency,
Research Triangle Park, N.C.
-181-
-------
Dravnieks, A. 1977. Organic Contaminants in Indoor Air and Their Relation
to Outdoor Contaminants - Final Report on Phase I: Development of
Methodology and Initial Field Trials. Research Project No. 183. Pre-
pared for American Society of Heating, Refrigerating and Air Conditioning
Engineers. IIT Research Institute, Chicago, 111.
Federal Register. 1972. Analytical Method for Detection of Asbestos
Fibers. FR 37(110):11320-22.
Henderson, J.J., F.B. Benson, and D.E. Caldwell. 1972. Indoor-Outdoor
Air Pollution Relationships: A Literature Review. No. AP-112. U.S.
Environmental Protection Agency, Research Triangle Park, N.C.
Hittman Associates, Inc. 1977. Development of Residential Buildings
Energy Conservation Research, Development and Demonstration Strategies.
HIT-681 Final Report. Prepared for the U.S. Energy Research and Develop-
ment Administration, Washington, D.C.
Hollowell, C.D., R.J. Budnitz, G.D. Case, and G.W. Traynor. 1976.
Combustion-Generated Indoor Air Pollution - I. Field Measurements 8/75-
10/75. LBL Report No. 4416. Prepared for U.S. Energy Research and
Development Administration. Contract W-7405-ENG-48. Lawrence Berkeley
Laboratory, Berkeley, Calif.
Honeywell, Inc. 1976. Electronic Air Cleaners, Theory and Fundamentals.
Honeywell, Inc., Minneapolis, Minn.
Hunt, C.M., B.C. Cadoff, and F.J. Powell. 1971. Indoor Air Pollution
Status Report. NBS Report 10-591. National Bureau of Standards
Project 4214101. Gaithersburg, Md.
Hunt, C.M. 1972. Simple Observations of Some Common Indoor Activities
as Procedures of Airborne Particulates. In ASHRAE Symposium: Cleaner
Indoor Air - Proqress and Problems. CI-72-1. Cincinnati, Ohio,
Oct. 19-21.
Hunt, C.M., and D.M. Burch. 1975. Air Infiltration Measurements in a
Four-Bedroom Townhouse Using Sulfur Hexafluoride as a Tracer Gas.
ASHRAE Transactions, 81 (Part I):186-201.
Johansson, T.B., R.E. VanGrieken, J.W. Nelson, and J.W. Winchester. 1975.
Elemental Trace Analysis of Small Samples by Proton Induced X-Rsy
Emission. Florida State University, Tallahassee, Fla.
McGee, E. 1977. Private communication. Gas Appliance Manufacturers
Association Technical Committee on Gas Ranges. Magic Chef. Cleveland,
Tenn.
-182-
-------
TECHNICAL REPORT DATA
'(Please read Instructions on the reverse before completingj
1. REPORT NO.
2.
3. RECIPIENT'S ACCESSION*NO.
4. TITLE AND SUBTITLE
INDOOR AIR POLLUTION IN THE RESIDENTIAL ENVIRONMENT
Volume I - Data Collection, Analysis, and Interpretation
5. REPORT DATE
August 1978
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Demetrios J. Moschandreas, Ph. D.; John W. C. Stark, M.S.;
James E. McFadden, B. A.; Sallie S. Morse, B. S.
8. PERFORMING ORGANIZATION REPORT NO.
GEOMET Report No. EF-688
9. PERFORMING ORGANIZATION NAME AND ADDRESS
GEOMET, Incorporated
15 Firstfield Road
Gaithersburg, Maryland 20760
10. PROGRAM ELEMENT NO.
^.caM'iueT/aftAN'f n6.
EPA Contract No. 68-02-2294
12. SPONSORING AGENCY NAME AND ADDRESS
U. S. Environmental Protection Agency
Environmental Research Center
and
U. S. Department of Housing and
Urban Development
Office of Policy
Development Research
13. TYPE OF REPORT AND PERIOD COVERED
FINAL
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. A#StRACV i
A 24-month study was undertaken to characterize the indoor residential air quality. Seventeen residential dwellings
were monitored, each for a 14-day period. Air samples were collected from four locations: one outdoor site adjacent to
the building; and three indoor sites, the kitchen, bedroom, and living room. "Continuous" sampling was carried out for
CO, SO2, NO, NOj, CO2, O3, CH^, and THC. TSP, RSP, SO^, NOj, Pb, ammonia, and aldehydes were monitored
intermittently. Aerosol samples were collected for elemental analysis by the Proton Induced X-ray Emission (PIXE) tech-
nique. In addition, data on energy parameters, infiltration rates, and family activities were obtained by observations,
field experiments, and daily questionnaires, respectively. Each residence was monitored with minimal interference in the
daily activities of the occupants; thus, the residential pollutant concentrations were determined under real-life conditions.
The air quality of the residential environment was determined to be markedly different from the outdoor ambient air
quality. Three classes of air pollutants have been identified with respect to indoor-outdoor pollutant relationships:
1) concentrations of CO, NO, CO2, NMHC, and aldehydes in the residential environment are often higher than outdoors;
2) TSP and RSP are sometime! lower and sometimes higher indoors than outdoors; and 3) indoor concentrations of SO2, O3,
SO|, and NOj are almost always lower than the corresponding outdoor pollutant concentrations. Indoor concentrations
of TSP, Oj, NMHC, and CO have been observed to equal or exceed the National Ambient Air Quality Standards for these
pollutants. The observed elevated levels of indoor pollutant concentrations are attributed to indoor pollutant sources.
(continued)
17.
KEY WORDS AND DOCUMENT ANALYSIS
a.
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATi Field/Group
Residences
Air Pollution
Energy Conservation
Health Effects
Numerical Models
Residential Monitoring
Energy -Environment
Considerations
18. DISTRIBUTION STATEMENT
Unclassified
20. SECURITY CLASS (fhkpaf)
21. NO. OF PASES
201
Unclassified
22. falCS
EPA Perm 2220-1 (»-73)
-------
16. ABSTRACT (Continued)
Two numerical models formulated in the course of this study are discussed in this document. The GEOMET Indoor-
Outdoor Air Pollution (GIOAP) model simulates indoor conditions and estimates indoor gaseous pollutant concentrations as
a function of outdoor levels, air exchange rates, indoor source strengths, and pollutant decay rates. Unique characteristics
of the GIOAP model include the transient term, a first-order chemical decay term and fine time resolution. Estimated
concentrations of CO, NO, NO2, and NMHC were within 25% of the observed values. Ozone predictions did not meet the
predetermined GIOAP model validation criteria but were realistic and often within 2 ppb of the observed Oj concentrations.
The second model formulated during this study, the Steady State TSP model, is an empirical model which estimates indoor
TSP levels as a function of outdoor levels, air exchange rates, removal mechanisms, and indoor TSP source terms. The
indoor TSP source terms include a reentrainment term and a pollutant generation term, both of which are functions of the
family activity index generated from information obtained by the questionnaires. Estimations made using the Steady State
TSP model compare favorably, always within 50% and most often within 30%, with the observed levels.
Finally, the relationship between energy conservation measures and air quality in the indoor environment is examined.
The parameter that associates energy conservation measures and air quality in residences is the air exchange rate of the
structure. Reduction in the air exchange rate will conserve energy, but it may lead to deterioration of the residential
air quality. A practicable air exchange rate of 0,5 air changes per hour was determined by numerical simulations. This
level is a cost-efficient reduction from the current average residential air exchange rate of 0.8 air changes per hour. Thus
it will conserve energy. In addition, it will not substantially deteriorate the indoor air quality. A number of scenarios
that conserve energy in residences and do not affect indoor air quality are also discussed in this study.
The observed indoor air pollutant.concentrations were, on the average, not very high; however, persistent moderate
and, at times, elevated pollutant levels were observed. The lack of studies concerning the health implications of such
levels is briefly discussed.
------- |