600R95180
        GEOMET Report Number IE-2603

               March 9,  1995
      ESTIMATION OF DISTRIBUTIONS FOR
       RESIDENTIAL AIR EXCHANGE RATES
               FINAL REPORT
               prepared for

              Patrick Kennedy
        Exposure Assessment Branch
Economics,  Exposure and Technology Division
 Office of  Pollution Prevention  and Toxics
   'U.S. Environmental Protection Agency
            401 M Street, S.W.
         Washington,  D,C.   20460
                   under

      EPA Contract Number 68-D9-0166
        Task Number 3-19,  Subtask 4
                    and
      EPA Contract Number 68-D3-0013
        Task Numbers 1-11 and 2-44
                prepared by
                   t
         GEOMET.Technologies,  Inc.
          20251 Century Boulevard
        Germantdwn,- Maryland  20874

-------
                            DISCLAIMER
     This document has been reviewed.and approved for publication
by the Office of Pollution Prevention and Toxics, U.S.
Environmental Protection Agency.  The use of trade names or
commercial products does not constitute Agency endorsement or
recommendation for use...
                               11

-------
                         TABLE  OF CONTENTS

                                                             Page

 1.    INTRODUCTION AND SUMMARY        .      '                   1

 2.    AIR  EXCHANGE MEASUREMENTS                                3
      2.1.  General  Considerations,                            3
      2.2.  Perfluorocarbon  Tracers                            4

 3.    DATA RESOURCES                               „           9
      3.1.  Database of-PFT  Ventilation Measurements           -9
      3.2.  Extracts and  Supplements  for the              ,    11
           PFT Database

 4.    ANALYSIS AND RESULTS            .      .  '         '       13
      4.1.  Data Exclusions  and Supplements                   15
      4.2.  Assignment  of Weights                             15
      4.3.  Summary Statistics   /                         .17

 5.    REFERENCES     .                         .           '25

APPENDIX
      Relationship Between Interzonal Airflows
      and House Volume and Air Exchange
                               111

-------
                         LIST OF FIGURES

                                                             Page

1    Airflows for multiple-zone systems                       6

2    Regions defined by the U.S. Census Bureau               13

3    Frequency .Distribution of Estimated Residential         2.0
     Air Exchange Rates for All Regions Combined
                               VI

-------
                         ACKNOWLEDGEMENTS
     This report was prepared by GEOMET Technologies, Inc.,
Germantown, Maryland, for the EPA Office of Pollution Prevention
and Toxics, Exposure Assessment Branch  (EAB) , under EPA Contract
No. 68-D3-0013  (Task 2-44) with Versar, Inc., Springfield,
Virginia.  The EPA-EAB Task Manager was Patrick Kennedy; his
support and guidance are gratefully acknowledged.

     In addition to the authors of this report, .a number of
Versar/GEOMET personnel have contributed to this task over the
period of performance.  These individuals are shown below:

     Program Management  -         Gayaneh Contos, Versar

     Task Management     -         Leo G.  Schweer, Versar
                                             Markeis, GEOMET
     Editing             -         Kristeen Hinkley, GEOMET

     Word Processing     -         Joy Warren,  GEOMET
                                   Glenda Heredia,  GEOMET
                               VII

-------
 1.         INTRODUCTION AND SUMMARY

      This report describes the information sources,  analysis
 methods  and estimates of national and regional distributions for '
 annual average air exchange rates measured in United States
 residences.   All the measurement  results derive from techniques
 involving constant release and time-integrated sampling of a
 family of compounds known as perfluorocarbon tracers (PFTs).
 This  technique,  developed at Brookhaven National Laboratory
 (BNL), is described in Section 2  of  this report.   Section 2 also
 provides a general discussion of  air  exchange,  including
 governing factors and alternative measurement methods.

      Components  of the PFT database used for this analysis,
 although deriving from more than  one  field study,  are unified by
 common reliance  on the BNL measurement  protocol.   Further,  all
 ~\ a VN/-N va •*- r^->~T.r T.T^-v-V  -n -«^ *-3 v*-v- •? T*I -> •*".*. ^~*~^  d^d"1""Cii3 'jLjrdi-i^inM C-L-Lk^J.
 checking)  was  accomplished in a single  laboratory.   Prior to the
 availability of  the PFT database,  relatively little  work had been
 directed toward  estimating a national distribution for air
 exchange rates,  primarily  due to  the  lack of national-scale
 studies  featuring unified  protocols.  The review chapter on
 infiltration in  the.current  edition of  the ASHRAE  Handbook of
 Fundamentals  (ASHRAE 1993),  for example,  briefly summarizes a
 number of  air  exchange  studies  that involved as  many as  300
 homes.   Of these,  only  two relatively early  studies  (Grimsrud et
 al. 1982, Grot and Clark 1979)  are cited as  giving a national
 perspective, and these  studies  were restricted  in  terms  of  the
 type of  housing.

     Since 1982,  the  PFT technique has  been  used to  measure
 airflow  rates  in more than 4,000  occupied residences in
 accordance with  BNL  protocols.  Under EPA sponsorship, these
measurement results  were compiled by BNL into data files that
were used by Versar  (1990)  in developing a database  and  user
 interface to allow various  researchers  to access the data.
 Pertinent features of the  database and  interface are outlined in
Section  3 of this  report.  The  structures  represented in the
database do not  constitute  a  random sample of residences across
the United States.   Rather,  they  represent a  compilation of
 residences visited in the  course  of about  100. separate field-
research projects  by.various  organizations,  some of  which
 involved random  sampling and  some of which involved  judgmental or
 fortuitous sampling.  Certain areas of  the country (e.g.,
Southern California) are heavily  represented in  the  database  used
 for .analyses, whereas other  areas have  little or no
representation.   Similarly,  some  seasons are  more  heavily
represented than others.

     The geographic  imbalance in  the PFT database  was partly
compensated by seeking  additional PFT measurement  results (e.g.,
from recently completed studies)  for areas .with  limited

-------
representation.  Further compensation was obtained by applying
weighting factors in the analysis.  These factors were developed
in such a way that state-specific results would contribute to
resulting estimates in proportion to their respective numbers of
occupied housing units, as determined from the 1990 census of
population and housing.  The weighting factors, as well as
criteria for excluding certain cases from the analysis, are
presented in Section 4 together with the results of the analysis,

     The results indicate that a value of 0.18 air changes per
hour (ACH)  would be appropriate as a conservative number when
modeling inhalation exposure, and that a value of 0.45 ACH would
be appropriate when a typical air exchange rate is desired for
modeling inhalation exposure.  These values correspond to the
10th and 50th percentile, respectively,  for the estimated
national distribution of annual average air exchange rates.

-------
 2.        AIR EXCHANGE MEASUREMENTS

.2.1..      General  Considerations

     Air  exchange,  the balanced flow of air into and out of a
 building, is  an  important determinant of indoor air quality.  For
 pollutants  of indoor origin, air exchange can serve to dilute
 indoor  levels.   Pollutants of outdoor origin, however, can also
 be brought  indoors.  Air exchange is comprised of three
 processes:


     •    Infiltration - airflows through random cracks,
          interstices and other unintentional openings in the
          building envelope;

     •    Nacurai  ventilation - airflows through doors, operable
          windbws,  and other designed openings in the building
          envelope; and

     •    Forced ventilation - controlled airflows driven by
          mechanical systems.


     While natural  ventilation and forced ventilation contribute
at times in homes,  infiltration is the dominant mechanism for
residential air exchange.  Infiltration is the inadvertent air
leakage of an otherwise sealed building.  The use of natural
ventilation is seasonal, and forced ventilation systems
 (excepting small local exhaust fans for spot ventilation in
kitchens and baths) are more often found in larger, non-
residential buildings.

     Basic principles relating, to the physics, modeling and
measurements of air exchange have been summarized elsewhere
 (ASHRAE 1993; Hutcheon and Handegord 1989; Sherman 1990).
Infiltration airflows are driven by pressure differences created
by wind action and  indoor-outdoor temperature differences.  The
driving pressure, in turn,  acts upon the leakage sites
distributed over the building surface.  Two basic avenues of'
measurement have evolved for air exchange:  (1)  pressurization
testing to measure  the effective leakage area of the building
coupled with modeling to calculate air exchange, and (2) tracer-
gas techniques to monitor the dilution effects of air exchange.

     Pressurization testing measures the effective leakage area
and thus provides  information on a more or less constant property
of the building.  While calculation methods exist to relate
effective leakage area to air exchange  (see ASHRAE 1993), results
are predicated on empirical assumptions (e.g., physical
distribution of leakage area on the building surface) and
meteorological considerations.

-------
     Tracer-gas  techniques,  on  the  other hand,  provide a more
direct measure of  air  exchange.   Tracer-gas  techniques are
generally based  on mass-balance  considerations  with explicit
assumptions that (1) well-mixed  conditions prevail,  and (2)
outdoor concentrations are negligible,  and take the form:
                     dC   _   St_     Q_c
                     dt   ~   V      V   Ct
where
     Ct =  tracer-gas concentration at time t (e.g., mg/m3)
     Sc =  tracer-gas release rate at time t (e.g., mg/h)
     V  =  volume of tested airspace (e.g., m3)
     Qt =  exiting airflow at time t (e.g., m3/h)


     The ratio of exiting  airflow (Q, m3/h) to indoor volume
 (V, m3)  gives the air exchange rate  (I,  h"1) .  Three basic tracer-
gas strategies have  evolved  to measure air exchange :
 (1) concentration decay,  (2). constant concentration,  and
 (3) constant injection.  The concentration-decay method  involves
releasing a small amount of  tracer  gas, mixing it  into the  full
air space and monitoring the change in concentration  with time.
From equation 1, the concentration  data are related to the  air
exchange rate by the analytical solution:


        C2  =  Cx  e  -1 (t>-t<),  or I  =   Ln  
-------
 tracer technique,  a  kit  was  developed by  researchers  at
 Brookhaven National  Laboratory  (BNL).  The  field components--
 miniature  perfluorocarbon  tracer  (PFT) permeation sources and
 passive samplers--are  inexpensive and reusable, require minimum
 training of field  personnel,  and can provide integrated
 measurements over  days or  months.
  '                     0
      The PFT technique developed by BNL is  an offshoot of the
 constant-inject ion,,  steady-state method of  measuring  air
 infiltration that  has  been in use for many  years.   In that
 scheme,  a  tracer is  emitted,into a building at a constant known
 rate  and its concentration is allowed to  reach a steady-state
 level.   This value is  measured and then converted to  a flow of
 outdoor air into the building through the use of a  single-zone,
 mass-balance model.

      BNL has also  extended this method to address multizone
 -czGurcnr.anto by deploying  a  dirrerent PFT in each zone.
 Measurement  of the concentrations of each tracer in all zones of
 the building permits the calculation of the infiltration and
 exfiltratien of outdoor air  for each zone of the building as well
 as the  flows between zones.   The system uses miniature permeation
 tubes as tracer emitters and passive samplers termed  Capillary
 Atmospheric  Tracer Samplers  (CATS) to collect the perfluorocarbon
 tracers.   The passive  samplers are returned to the laboratory for
 analysis where the tracers are separated by gas chromatography
 and quantified by  an electron capture detector (GC/ECD).  A
 detailed description of this  technique has been presented
 elsewhere  (Dietz et  al. 1986).

     As  shown'in Figure 1, a  single-zone building,  because it is
 flow-coupled only  to the outdoors, has two flows.   A  two-zone
building,  on the other hand,  has each zone coupled to the
outdoors as  well as  to each  other, giving a total of  six flows.
Generalizing, a building composed of N well-mixed zones will have
 (N-l)  interzonal airflows  for each zone plus an indoor/outdoor
 flow pair  for each zone,  giving a total of N«(N-1)  + 2N,  or
N»(N+1)  flows. The airflows  are  computed  by inserting the
measured tracer concentrations and the known tracer emission
rates into the system of mass-balance and flow-balance equations
 for each zone of the building.

     Although the  PFT methodology is relatively simple to
implement,  it is subject to  errors and uncertainties.  The
general performance  of the sampling and analytical aspects of the
system are quite good.  That  is,  sample collection by the CATs
allied to  GC/ECD analysis will measure the correct time-weighted
average tracer, concentration to within a few percent  (Dietz et
al. 1986).    Nonetheless,  significant errors can arise when
conditions in the  measurement scene greatly deviate from
 idealizations calling  for  constant, well-mixed conditions.
Principal  concerns focus on  the effects of naturally  varying air
exchange and the effects of  temperature in  the permeation source.

-------
Single-Zone
System

Two-Zone
System *






*.


Three-Zone
System
                             -V
 N-Zone  System Requires  (N +  1) • N Airflows
  Figure 1.  Airflows for multiple-zone systems,

-------
      Sherman (1989)  carried out  an error  analysis  of  the  PFT
 methodology using mathematical models  combined  with typical
 weather data to calculate how an ideal sampling system would
 perform in a time-varying environment.  He  found that for simple
 single-story (ranch)  and two-story plus basement (colonial)
 layouts,  seasonal measurements would underpredict  seasonal
 average air'exchange by 20 to 30 percent.   Underprediction can
 occur because the PFT methodology is measuring  the effective
 ventilation (the product of ventilation efficiency and air
 exchange),  and the temporal efficiency will generally be  less
 than  Unity over averaging periods of this length.  Sherman also
 noted,  however,  that while the bias could have  an  impact  on    »
 determining air exchange (absent knowledge  of ventilation
 efficiency)  for calculating energy loads, the effective air
 exchange  term is directly relevant to  determining  average indoor
 concentrations resulting from constant sources.

      Sherman (1989)  also noted that while multiple tracers
 improve estimates of average effective ventilation in multiple-
 zone  buildings,  the  uncertainty  associated  with a  given
 interzonal  airflow is 10 to 20 percent  of the zone total,
 reducing  the quantitative utility of smaller estimated airflows.
 As with the  effective air exchange concept, however,  the  airflow
 estimates  correspond to  effective interzonal transport and are
 directly  relevant to determining average indoor concentrations in
 one zone  resulting from  a constant source in another.

      Leaderer et  al.  (1985)  conducted  a series  of  experiments in
 a room-sized environmental  chamber to  evaluate  the practical
 impacts of varying air exchange  and the temperature response of
 the permeation  sources.   The negative bias  anticipated in the
 measured  (effective)  versus  actual  air  exchange  as conditions
 varied diurnally  between 0.4  and 1.5 ACH was evident  but  minor
 (3 to 6 percent),  most likely due  to the mechanical mixing in the
 chamber and  the  relatively  short'integration time  (72 h).
 Similarly, cycling temperature diurnally over an 8 C° range
 (holding air  exchange  steady at  0.6 ACH) would  cause
 concentrations  changes of about  20  percent  as emissions
 fluctuated.   The  investigators found; however,  that using a  time-
weighted average  temperature  to  define  the  emission rate  reduced
 the temperature bias  to  essentially zero.

     The PFT  measurement  system  is  also subject  to potential
 complications related to choices  made when  deploying  the  tracers.
 If a  field researcher deploying  tracers unknowingly treats a
well-mixed zone as two separate  zones,  then the  computational
 solution for  the  ventilation flows  becomes  very unstable.  In the
mathematical  literature,  the  system is  termed ill-conditioned.
Researchers  at  BNL developed a means to reliably detect and
accommodate  such  unstable situations using  the  matrix of  tracer
 concentrations  to compute a  condition  number  (D'Ottavio et al.
 1988).  Through the  use  of  the condition number, large flow
 errors  for homes  with two or more  well-mixed zones can be

-------
errors for homes with two or more well-mixed zones can be
eliminated by combining .the well-mixed zones into a single zone
prior to the computations.  As described later,- a number of
measurement results were encountered in this analysis for which
the automatic zone-reduction procedure had been applied at BNL.

-------
 3.         DATA RESOURCES

      Most  of  the  measurement  results used  for this analysis were
 extracted  from a  database  of  PFT ventilation measurements
 developed  by  Versar (1990) based on data files received  from  BNL
 under a  contract  with the  USEPA Office of  Toxic Substances.
 Selected features of  this  database are described in Section 3.1.
 Information extracted from the database was supplemented with PFT
 measurement results from three additional  field studies.  The
 need  for and  nature of these  supplements are discussed in
 Section  3.2.
       i
 3.1.       Database of PFT  Ventilation Measurements

      The PFT  database consists of six distinct files:
      {j.;   me Project file, which' contains information about  the
           institution that performed the measurements and the
           general location of the houses associated with the
           project;

      (2)   The House file, which contains information associated
           with the entire house such as the house setup code, the
           start and stop dates for the measurement, and the
           overall air exchange rate for each house;

      (3)   The Zone 1 file, which contains information unique  to
           the first zone in the measured building, including  the
           zone volume, the tracer used, and the infiltration  and
           exfiltration flow rates for the zone;

      (4)   The Zone 2 file, which contains information unique  to
           the second zone, if any;

      (5)   The Zone 3 file, which contains information unique  to
           the third zone, if any; and

      (6)   The Zone 4 file, which contains information unique  to
           the fourth zone, if any.


     Many  houses are associated with a particular project record
(i.e., there are many-to-one relationships between the House  file
and the Project file).  Each house is associated with one Zone 1
record since each house must have at least one zone in which  PFT
measurements were taken.  Therefore, a one-to-one relationship
exists between the House file and the Zone 1 file.  In addition,
each house may or may not be associated with a record from the
Zone 2, Zone 3, or Zone 4 files depending on how many zones the
house was  divided into for the ventilation measurement.  For
example, a house in which two measurement zones were defined  by

-------
the researchers would appear in the House file and in the Zone  1
and Zone 2 files, but this house would not appear in the Zone 3
or Zone 4 files.

     Each house calculation can be identified by a unique field
that exists in the House file and all Its associated zone files.
This field is labeled HOFILE in the House file, Z1FILE in the
Zone 1 file, Z2FILE in the Zone 2, Z3FILE in the Zone 3 file, and
Z4FILE in the Zone 4 file.  The House and Zone files can always
be related through this unique field.  Further details on the
specific contents of these files are provided in a separate
report (Versar 1990).

     Some houses appear in the database more than once for a
particular set of measurements because zone-reduction techniques
were applied by BNL.  In certain cases where researchers
deploying PFTs unknowingly divided a well-mixed zone into two'
separate zones, relatively large errors were associated with the
airflow rates computed by BNL.  The.large flow errors for homes
with two o"r more well-mixed zones were eliminated by combining
the well-mixed zones into a single zone prior to performing
computations on airflow rates.

     In the quality-control review of the database,  Versar and
GEOMET developed six "flag" fields to identify cases where zone-
reduction procedures were applied as well as other cases that
users might want to avoid.  The names and explanations of these
six flags are as follows:                                        '


     (1)   Zone Reduction Flag - A code of "1" indicates cases
          where zone-reduction procedures were applied (such
          houses are also represented in the database in reduced
          form); a code of "0" indicates cases where zone-
          reduction procedures were never applied as well as the
          final reduced form for cases were zone-reduction
          procedures were applied,-

     
-------
      (4)   Sums Flag -  "1"  indicates cases where the sum of
           airflow rates  into  and  out of each zone do not agree
           within one percent;

      (5)   Negative Flows Flag - "1" indicates cases where one or
           more negative  airflows  has been computed for a house;
           although physically impossible, negative airflow rates
           can result from  calculations because of measurement
           uncertainties; and

      (6)   Flow Error Flag  - "1" indicates a case for which one or
           more zones has an error associated with an estimated
           airflow rate that is as large as, or larger than, the
           flow itself; thus,  the  estimated airflow rate cannot be
           confidently distinguished from zero.


     Two additional  sets of important information were included
among the  files  provided by BNL.  First, a description of the
house structure  was  included  in a specially designed house setup
code.  This code indicates where  each house zone is located,
whether the house has a  fireplace, and whether the house has a
basement,  crawl  space, attic,  or  garage as part of the measured
space.  Such  information is useful in understanding (1) how the
zoning of  a house was conceptualized by the field researchers and
(2) cases  where  unconditioned  zones, which may have atypically
high ventilation rates,  were  included in the overall ventilation
measurement for  the  house.

     The second  major addition was information about the projects
for which  the  ventilation measurements were made.  On average,.
each project comprised approximately 50 house measurements.  A
group of ventilation measurements was associated with a
particular project name  if all the measurements were conducted by
the same institution and if all the homes were located in the
same geographic  area (i.e., within about 100 miles of each
other).  Each  project has a unique project code, a full
description of the location of the homes, and a short description
of the project.   In  addition,   information is provided about the
institution that  made the measurements, such as the names, phone
numbers and full  addresses of  contact people at the institution.

3.2.       Extracts and Supplements for the PFT Database

     The software developed by Versar for accessing the PFT
database provides three  essential types of functions:
(1) retrieving and sorting data;  (2) viewing data files; and
(3) creating reports and output files.

     The function for creating reports and output files is
relatively powerful.  User-selected reports can be directed to a
printer or disk  file, enabling the user to examine information

                                11 •

-------
 for the chosen houses in greater detail outside the database.
 The user can also request that certain information about the
 chosen houses be written to an ASCII file.   The ASCII file is
 written in a relatively straightforward format (i.e., single
 record for each house consisting of fields  in fixed locations
 with a blank delimiter between fields).   This format is
 compatible with a number of commonly used software programs
 (e.g.,  Lotus,  dBase,  SPSS/PC)  and programming languages,
 providing the user with the capability to .further manipulate or
.analyze the data outside the database.   This feature is important
 because the PFT database and software have  no built-in capability
 for statistical analysis.                          ;      •

      The option to create an ASCII file was chosen to enable
 analysis outside the  PFT database.  This option allows the user
 to  create an ASCII file either (1) from the House,  Zone 1 and
 Zone 2  files or (2)  from the Zone 3 and Zone 4 files.  Because
 the focus of the analysis was on whole-house air exchange rates,
 as  opposed,, to airflows among zones within residences, the
 selection was made that created a file  from the House,  Zone 1 and
 Zone 2  files.   The key information in this  file needed for data
 analysis included the state in which measurements were taken,  the
 project code,  the measured whole-house  air  exchange rate,  and the
 date when each measurement was initiated.   Prior to extracting
 this information (together with additional  information included '
 in  the  file that was  not needed for analysis),  certain cases were
 excluded.   The nature of and reasons for these exclusions are
 discussed in Section  4.

      The primary intent of the analysis  was to provide an
 estimate of the national distribution of residential air exchange
 rates.   Preliminary analysis of the data indicated that (1)
 measurements from the Los Angeles area  (Wallace 1987,  Wilson et
 al.  1986)  constituted a large fraction  (more than one-third)  of
 the database entries  and (2)  there were  few measurements from
 states  in the  north-central and southern regions  of the country
 (as defined by the U.S.  Census Bureau;  see  Figure 2).   The
 measurements from the Los Angeles area were supplemented with PFT
 icoulLb f-LGiu Lvno icx-ciii- lield studies for the California energy
 Commission (Berkeley  Solar Group 1990 and ADM Associates 1990)
 that were conducted primarily in other  areas of the state.
 Measurements for the  southern region were supplemented with PFT
 results from a 1991 study in Florida (GEOMET 1992).   Potential
 supplemental sources  of information for  the north-central  region
 of  the  country also were explored,  but  none were  found.
                                                 i
      The supplemental data were appended to the ASCII file
 extracted from the PFT database,  with critical information  such
 as  state,  air exchange rate,  and measurement date entered in
 appropriate field locations.   Data analysis,  described in the
 next section,  was applied to the supplemented ASCII file using
 SPSS/PC software.

                                12

-------
          WEST
                               NORTH CENTRAL
                                                NORTHEAST
                   HAWAII
Figure 2.   Regions defined by the U.S. Census  Bureau.
                           13

-------
     Other researchers* analyzing the PFT database discovered
some errors in the data from the SoCal field project  (Wilson et
al. 1986).  These errors, which pertained to the house volume,
zone-specific volumes and air exchange rate for some of the
houses, were found to be traceable to the original data files
provided to Versar by BNL.  Corresponding data for these fields
were obtained from the original SoCal field researchers and added
to the PFT database.  The air exchange rate was recalcu? ated by
dividing the house infiltration airflow rate from the PFT
database by the house volume provided by the researchers.  In
cases where the recalculated air exchange rate differed from the
one in the original database by more than 10 percent, a flag was
generated and the air exchange rate provided by the original
researchers was used instead.  Based on a series of contacts with
original researchers for other field projects who were able to
locate their own copies of PFT measurement results, it was
determined that any errors in the PFT database appear to be
confined' to the SoCal project.
           Muhilan  Pandian,  University of Nevada Las Vegas,
letter •communication dated August 1, 1994.

                                14

-------
 4.         ANALYSIS AND RESULTS

 4.1        Data Exclusions  and Supplements

      The  analysis  procedure  involved  (1) exclusion  of  certain
 cases that could have  biased the  results and  (2)  assigning
 weights to the results from  each  state  to  compensate for
 geographic imbalance in the  locations where PFT measurements  were
 taken.  The original PFT database compiled by Versar from  BNL
 data  files contains 4,590  measurement results.  However, as noted
 in Section 3,  some homes are represented more than  once  (i.e.,
 essentially redundant  results for whole-house air exchange rates)
 because zone-reduction procedures were  applied; in  such
 instances,  only the reduced  case  was used  (i.e.,  the case  with
 the smallest apparent  measurement error).  Residences  outside the
 U^iLcJl .jwctLco  WC.LC: excluded,  as were nomes with unconditioned
 spaces  (which  could lead to  artificially high air exchange
 rates).   Also  excluded were  small field studies involving
 repeated  measurements  in research ho'uses,  based on  descriptions
 provided  in the Project file.

      After the  above exclusions,  3,174  cases  remained  for
-analysis.   EPA-sponsored TEAM studies involving a random cross-
 section of  houses  in California,  Maryland  and New Jersey involved
 repeated  measurements  in a subset of the houses,  typically for
 two sequential  12-hour intervals  (daytime  and nighttime).  The
 house code  assigned by the original researchers* enabled
 identification  of  these sequential measurements,  which were
 averaged  to provide a  single  air  exchange  rate for  each house.
 Following this  averaging process,  which eliminated  285 repeat
measurements, there were 2,889 cases for analysis.   Supplemental
measurements from  California  and  Florida,  described in the
previous  section,  increased  the number of  cases to  2,976.

     These  cases were  first  examined in terms of  state-specific
 frequency distributions.  A  few negative values that were  found
were excluded from further analysis (such  values  are physically
impossible).  Also  excluded  were  several apparent outliers--three
for Texas and one  each for California and  Maryland.  In each
case,  the excluded  values were at  least seven times greater than
the next highest value for the state.   After  excluding these
outliers,  2,971 cases  remained for analysis.

4.2       Assignment of  Weights

     The  remaining  data were  examined by state and  by  the  four
geographic  regions  defined by the U.S.  Census Bureau (see
Figure 2).  As  shown in Table 1,  measurements taken in California
accounted  for nearly half of  all  measurements in  the database and
           Lance Wallace,  U.S.  Environmental Protection Agency,
Warrenton, VA, personal communication.

                                15

-------
Table 1.  Original and Weighted Numbers of PFT Measurements,
                     by Region  and  State
Region/State
West Region
Arizona
California-Los Angeles
California-Other
Colorado
Idaho
Montana
Oregon
Washington
Region Total
;North Central • -Region,
Minnesota
Wisconsin
Region Total
Northeast Region
Connecticut
New Jersey
New York
Region Total
South Region
Florida
Maryland
Texas
Region Total
Total , All Regions
Number of Valid
PFT Measurements

25
1,388
69
9
78
51
212
235
2,067
• '
63
57
120

25
9
527
561

t>
18 .
163
42
223
2,971
Weight Applied

0.680
0.046
0.928
1.778
0.051
0.078
0.066
0.098


1.825
2.246


0.960
6.000
0.241

'.'''."•.'.,•
7.611
0.288
3.857


Resultant
Number
of Cases
l
17
64
64
16
4
' ' 4
14
23
206

115
128
243

24
54
'. 127
205

137
47
162
346
1,000
                             16

-------
 two-thirds of the measurements in the  west  region.   To prevent
 this preponderance of results for California  from excessively
 influencing summary statistics,  weights  were  assigned to  each
 state's results so that the resultant  number  of  cases would
 represent each state in proportion to  its share  of occupied
 housing units,  as determined from the  1990  census of population
 and housing (U.S. Bureau of the Census,  1991).   Further,  for
 California,  the weights for the Los Angeles area and remainder  of
 the state (which are nearly equal in housing  share)  were  set so
 that each of these areas would contribute equally to the  weighted
 results.   The weights for each,state also reflected each  region's
 share of  occupied housing units.   The  weights were constructed  in
 such a way that the resultant number of  cases, when summed across
 regions,  would total exactly 1,000 (see  Table 1).

 4.3       Summary Statistics

      Summary statistics for estimated  annual  average air  exchange
 rates,  based on weighting of cases as  described  above,  are given
 in  Table  2 for each of the four geographic  regions  and for all
 regions combined.   The air exchange rates are in units of air
 changes per  hour (ACH).   Up through the  65th  percentile,  the
 distributions  are fairly similar across  geographic  regions (i.e.,
 the values for each region are within  0.1 ACH of the value for
 all regions  combined).   The primary differences  across regions
 are lower-than-average values for the  north-central  region and
 higher-than-average values for the northeast  region,  as reflected
 by  both the  arithmetic and geometric means.


      For  purposes of modeling inhalation exposure  in residential
 settings,  the  10th percentile value of 0.18 ACH  for  all regions
 combined  seems  appropriate when  a conservative air  exchange rate
 is  desired.  None of the region-specific values  at  this
 percentile deviates markedly from 0.18.  When a  typical air
 exchange  rate  is desired for modeling, the  median  (50th
 percentile)  or  geometric mean is  a preferred  measure of central
 tendency  because neither is excessively,influenced  by extreme
 values  at the upper tail of the  distribution.  The median value
.for all regions combined is 0.45  ACH and the  geometric mean is
 0.46  ACH.  Thus,  0.45  ACH seems  appropriate as a typical  air
 exchange  rate.

      In applying conservative or  typical values  of  air exchange
 rates for modeling purposes,  it  is important  to  keep in mind the
 limitations  of  the underlying data base.  Although  the estimates
 are based on thousands of measurements,  the residences
 represented  in  the data base are  not a random sample of the
 United  States  housing stock.   The houses are  not geographically
 balanced,  but  efforts have been made to  compensate  for this
 imbalance.   Further,  as discussed later, the  results are  not
 equally balanced among different  times of the year,  and air
 exchange  rates  often vary seasonally because  of  changes in

                                17

-------
    Table 2.  Summary Statistics* for Estimated 'Annual Average
               Air  Exchange Rates (in ACH) by  Region
Statistic
Unweighted
No. of Cases
Weighted
No. of Cases
Arithmetic
Mean
Arithmetic
Standard
Deviation
Geometric
Mean
Geometric
Standard
Deviation
West Region
2067
206
0.66
0.87
0.47
2.11
North
Central
Region
120
243
0.57
0.63
0.39
2.36
Northeast
Region
561
205
0.71
'0.60
0.54
2.14
South
Region
223
• 346
0.61
0.51
0.46
2.28
All
Regions
2,971
1,000
0.63
0.65
0.46
2.25
Percentiles of Distribution
5th
10th
15th
20th
25th
30th
35th
40th
45th
50th
55th '
OULil
65th
70th
75th
80th
85th
90th '
95th .
Maximum
0.16
0.20
0.23
0 . 27
0.30
0.30
0.33
0.37
0.40
0.43
0.48
U . J^.
0.60
0.68
0.76
0.83
0.96
1.25
1.96
23.32 .
0.12
0.16
0.18
0.20
0.22
0.23
0.26
0.29
0.32
0.35
0.38
^ . -1 •_
0.51
0.55
0.62
0.75
1.00
1.49
1.92
4.52
0.19
0.23
0.27
0.32
0.32
0.36
0.39
0.42
0.48
0.49
0.53
: . : :
0.72
0.91
0.99
1.19
1.26
1.33
1.87
5.49
0.15
0.16
0.19
0.21
0.24
0.31
0.35
0.41
0.48
0.49
0.56
: - r
0.65
0.70
0.73
0.86
1.06
1.21
1.58
3.44
0.15
0.18
0.20
0.23
0.26
0.30
0.33
0.36
0.40
0.45
0.49
n r A
0.60
0.66
0.76
0.90
1.09
1.26
1.74
23.32
* All  numbers reflect weighting by state as  shown in Table 1,
                                 18

-------
 driving forces  such as  indoor-outdoor  temperature  differences.
 Despite such limitations,  the  estimates  in  Table 2 are  believed
 to  represent the  best available  information on  the distribution
 of  air  exchange rates across United  States  residences throughout
 the year.

      Components of  the  PFT database  used for this  analysis,
 although deriving from  more than one field  study,  are unified by
 common,  reliance on  the  BNL measurement protocol.   Further,  all
 laboratory work and primary data analysis (including error
 checking) was accomplished in  a  single laboratory.  Prior to the
 availability of PFT database,  relatively little work had been
 directed toward estimating a national  distribution for  air
 exchange rates, primarily due  to the lack of national-scale
 studies featuring unified  protocols.   The review chapter on
 infiltration in the current edition  of the  ASHRAE  Handbook  r^f
 r^iJaiiiciii-calfi. v'/iortK/ir, ly^j; , ror  example,  briefly summarizes a
 number  of air exchange  studies that  involved as many as 300
 homes.   Of these, only  two relatively  early studies  (Grimsrud et
 al.  1982, Grot  and  Clark 1979) are cited as giving a national
 perspective", and  these .studies were  restricted  in  terms of  the
 type of  housing.

     The estimated  annual  average distribution  for all  regions
 combined is  shown in graphical form, as  a histogram with
 intervals of 0.1  ACH, in Figure  3.   The  shape of the histogram
 generally is consistent with that from a lognormal distribution.
 Nearly  50 percent of the residential air exchange  rates are in
 the relatively  narrow range of 0.1 or  less  to 0.4  ACH,  but  the
 tail of  the  distribution is quite extended  and  includes about
 3 percent of residential air exchange rates  that exceed 2.0 ACH.

     Summary statistics for estimated seasonal average  air
 exchange rates  are  given in Table 3.  Seasons were defined  by
grouping December,  January, and  February measurements-for winter;
March, April and  May for spring; and so  on.  Caution should be
exercised in interpreting  or applying the seasonal statistics
because  (1)   the seasons are not  equally  represented in  terms of
unweighted or weighted numbers of cases,  and (2) attempts to
compensate for  geographic  imbalance were  not applied on a season-
specific basis.   The winter and  spring measurements each account
 for more than a third of the total observations whereas the fall
measurements account for less than 10 percent.  Limitations
related  to seasonal  imbalance apply  equally to the annual
estimates previously given in Table  2.   Although,  in concept,
efforts  could have  been made to  adjust for  both geographic  and
seasonal imbalance  in developing the annual  estimates,  such an
effort would have been severely  hampered by the limited
 information  available for  numerous combinations of geographic
 location and time of year.
                                19

-------
      0.1 0.2 0.3 0.4  0.5 0.6 0.7 0.8 0.9 1.0 1.1  1.2 1.3 1.4 1.5  1.6 1.7 1.8 1.9  2.0
                        AIR EXCHANGE INTERVAL, ACH
>2.0
Figure 3.   Frequency Distribution of  Estimated Residential
        Air  Exchange  Rates for All Regions Combined.
                                20

-------
        Table  3.   Summary Statistics*  for  Estimated Seasonal
                    Average Air Exchange Rates
Statistic
Unweighted
No. of Cases
Weighted
No. of Cases
Arithmetic
Mean
Arithmetic
Standard
^ c: *„ _L. a v~ j. s-^ii
Geometric
Mean
Geometric
Standard
Deviation
Winter
1162
439
0.67
0.60
0.52
2.01
Spring
1160
387
0.65
0.65
0.45
2 . 41
Summer
507
121
0.57
0.88
0.34
' 2.48
Fall
142
53
0.38
0 . ^1
0.31
1.88
Percentiles of Distribution
10th
25th
50th
75th
90th
0.21^
0.34
0.51
0.86
1.26
0.19
0.27
0.43
0.76
• 1.50
0.15
0.16
0.26
' 0.63
1.45
0.14 -
0.22
0.30
0.48
0.74
*A11 numbers reflect weighting by state as shown.in Table 1.
                               21

-------
     Seasonal and annual average air exchange rates are
summarized by region and by state in Table 4  (arithmetic mean and
standard deviation) and Table 5  (geometric mean and standard
deviation).  The unweighted and weighted numbers of cases for
each area-season combination are also shown in Table 4.  In most
cases the seasonal means are below one ACH.   The major exception
is the Los Angeles area, for which both the arithmetic and the
geometric mean for summer are above one ACH  (most likely due to
open-house configuration under moderate climate conditions).  New
Jersey also has arithmetic/geometric means of one ACH or higher
for winter, but these statistics are based on a limited set of
nine homes.

     As noted previously, the estimates provided in this section
are believed to represent the best available information on
distributions of air exchange rates but are limited by both
geographic and seasonal imbalances.  Such limitations could
conceivably be overcome by conducting additional measurements in
areas and at times that have received limited coverage to date.
Given sufficient resources, a preferred alternative would be to
select- a nationally representative sample of residences and
conduct week-long measurements that are equally distributed
throughout the year in each region of the country.

     In addition to the air exchange rate for the house,
estimates of air flows among zones within a house may be needed
to model inhalation exposure,  because the release of airborne
chemicals is often confined -to one zone of the house.   Given the
variety of situations represented in the PFT database,  it is
difficult to choose typical or conservative values for interzonal
airflow rates.  Consequently,  an effort was made to model the
relationship between interzonal airflow rates and whole-house air
exchange rates.  This analysis effort,  described in the Appendix,
was conducted for two situations--bedroom area versus remainder
of the house and kitchen area versus remainder of the house.
                                22

-------
     Table  4.   Arithmetic Statistics for  Estimated  Seasonal  and
       Annual  Average  Air Exchange Rates,  by Region and State
Region/State
West
Region 	 	
Arizona
California-
Los Angeles
California-
Other
Colorado
Idaho
Montana
[ . i
Oregon
Washington
North Central
Region
Minnesota
Wisconsin
Northeast
Region :
Connecticut
New Jersey
New York
South
Region
Florida
Maryland
Texas
Winter
0.51 ± 0.88*
(752,72)
,^^^_MB^B^— ^MB— mMBMM—i^^H
0.40 + 0.23
(11,7) %
0.77 + 1.72
(375,17) .
0.56 + 0.33
(16,15)
0.24 +0.07
(2,4)
0.50**
no. i ^
0.21 ± 0.22
(27,2)
a
0.49 ± 0.45
(128,8)
0.32 + 0.19
(169,17)
0.36 ± 0.24
(28,53)
0.27 ± 0.12
(23,42)
0.70 +0.28
(5,11)
0,76 :+. 0.50 :
(342,144) '
0.44 + 0.31
(14,13)
1.12 ± 0.46
(9,54)
0.55 ± 0.40
(319,77)
. 0::75:::;±::"d.:55 ;:
,140,170) : ;
0.43 + 0.14
(8,61)
0.60**
(4,1)
0.93 + 0.62
(28,108)
Spring
0.60 ± 0.56
(692,49)
^HM«^H««B^«MH^BH^^BMM^MM
0.49**
(1,1)
0.81 + 0.68
(551,25)
0.31 + 0.02
(3,3)
0.33 ± 0.22
(6,11)
^0.55 ± 0.66
f ~*. •-, "> '
0.20**
(4,0)
0.51 ± 0.35
(51,3)
0.36 ± 0.23
(41,4)
0.66 ± 0.70
(87,181)
0.34 ± 0.32
(35,64)
0.83 + 0.79
(52,117)
0.61 ± 0.78
(210,;59)
1.25 + 1.63
.(11,11)
--
0.47 + 0.31
(199,48)
;;:o .;68 :;± o.:47 •;
•.' (171,: 99). •;
.
0.59 + 0.44
, (157,45)
0.75 + 0.49
(14,54)
Summer
0.98 ± 1.14
(492,58)
•*«BIWM^HI«WMI^M.^^_
0.36 + 0.20
(10,7)
1.62 + 1.62
(449,21)
0.68 + 0.50
(33,31)
--
.--
--
-
--
--
--
--
0.82**.
(5,1) :.
--
--
0.82**
(5,1)
0.19 ±:0.:05
v 01 0/61) :.
0.19 + 0.04
(8,61)
0.28**
(2,1)
--•
Fall
0.47 ± 0.38
(131,27)
MBMM^HIBHMBIMMHBMHHM^^^—
0.41 ± 0.36
(3,2)
0.85**
(13,1)
0.52 + 0.28
(17,16)
0.16 ± 0.00
U',2)
0.36**
119,1)
0.25 + 0.25
(20,2)
0.63 ± 0.42
(33,2)
0.39 ± 0.18
(25,2)
0.13 ± 0.03
<5,9)
0.13 + 0.03
(5,.9)
--
:0.23**
(4,:l)
'--
--
0.23**
(4,1)
,0.36 ± 0.12
, (2,15)
0.36 ± 0.12
(2,15)
--
.--
All Seasons
0.66 ± 0.87
(2067,206)
I^^^BMHMB>*«M^H^HIMI^HH»KIIBd
0.39 + 0.21
(25,17)
1.06 ± 1.39
(1388,64)
0.60 + 0.41
(69,64)
0.29 + 0.19
(9,16)
n 4Q j. n c-
(78~,4)
0.23 ± 0.17
(51,4)
0.52 ± 0.40
(212,14)
0.34 ± 0.19
(235,23)
0.57 ±0.63
(120,243)
0.30 ± 0.26
(63,115)
0.82 +0.76
(57,128)
0.71 ± 0.60
(561,205')
0.80 + 1.15
(25,24)
1.12 ± 0.46
(9,54)
0.52 ± 0.37
(527,127)
0.61 ± 0.51
(223,346)
0.31 ± 0.16
(18,137)
0.59 ± 0.44
(163,47)
0.87 ± 0.58
(42,162)
Arithmetic mean + standard deviation  (unweighted and weighted number
cases); all numbers reflect weighting by state as shown in Table 1.
Based on one or fewer cases after weighting.
of
                                23

-------
  Table 5.  Geometric Statistics for Estimated Seasonal and
   Annual Average Air  Exchange  Rates,  by  Region  and State
Region/State .
West
Region
Arizona
California-
Los Angeles
California-.
Other
Colorado
Idaho
Montana
Oregon
Washington
North Central
Region
Minnesota
Wisconsin
Northeast •:.
Region
Connecticut
New Jersey
New York
South '
Region • :
Florida
Maryland
Texas , -
Winter
0.39 ;± .1.93*
0.35 ± 1.62
0.54 ± 2.10
0.49 ± 1.70
0.23 ± 1.37
0.31**
0.18 ± 2.04
0.41 ± 1.81
0.28 ± 1.75
0.30 ± 1.75 •'..
0.25 ± 1.48
0.65 ± 1.49
.0.:59 ± :2.12. ':
0.37 ± 1.81
1.00 ± 1.71
0.45 .± 2.03
0.63 ±.1.77
04] 0.1 at
0.56**
0.80 ± 1.70
Spring
0,46 ± 2.00 ;
0.49**
0.65 • ± 1.97
0.31 ± 1/07
.0.28 ± 1.63
0.41 ± 3.29
0.20**
0.46 ± 1.70
0.31 ± 1.82
0.44 ± 2.46
. 0.27 + 2.00
0.57 ± 2.43
.0:44 ±2:00 :;
0.67 ± 3.10
--
0.40 ± 1.82 .
0.49 ± 2.72

0.42 ± 2.88
0.55 ± 2.57
Summer
0.67 ± 2.31
0.31 ± 1.76
1.12 ±2.46
0.56 ± 1.88
--
-•-
--

--
--.-.. . •'.
' --
.
0.64** : -.•';
--
—
0.64**
.0.18 ± 1:.38 •
2 . ic x i . 11
0.05**
--
Fall
0.34 ±. 1.81
0.34 ± 2.26
0.45**
0.47 ± 1.60
0.16 ± 0.00
0.28**
0.21 ± 2.57
0.55 ± 2.05
0.36 ± 1.57
0.13 ± 1.25.
0.13 ± 1.25
--
0.22**
--
--
0.22**
.0.34 ± 1.43
u . 0-i ± J. . 4^
-- •
--
All Seasons
0.47 ± 2.11
0.34 ± 1.67
0.73 ± 2.27
0.50 ± 1.76
0.25 ± 1.58
0.34 ± 2.60
0.19 ± 1.84
0.44 ± 1.75
0.29 ± 1.71
0.39 ± 2.36
0.24 ± 1.82
0.58 ± 2.35
0.54 ± 2.14
0.48 ± 2.46
1.00 ± 1.71
0.43 ± 1.95
0.46 ± 2.2E
0.28 ± 1.62
0.41 ± 2 . 99
0.71 ± 2.05
Geometric mean ± standard deviation; all numbers reflect
weighting by state as shown' in Table 1.

Based on one or fewer cases after weighting. '•
                             24

-------
 5.         REFERENCES

 ADM Associates.   1990.   Pilot Residential Air Exchange Survey.
 Task  2;   Pilot  Infiltration Study.  Indoor Air Quality Assessment
 Project.   ADM Associates,  Inc., Sacramento, CA.  Prepared  for
 California Energy Commission under  Contract No. 400-88-020.

 ASHRAE.   1993.   Infiltration and Ventilation.  In 1989 ASHRAE
 Handbook  of Fundamentals.• American Society of Heating,
 Refrigerating and Air-Conditioning  Engineers, Atlanta, pp. 23.1-
 23.20.

 Berkeley  Solar Group.  1990.  Occupancy Patterns and Energy
 Consumption in New California Houses  (1984-1988).  Berkeley Solar
 Group and Xenergy, Oakland, CA.  Prepared for California Energy
 Commission under Contract No. 400-87-015.
                                               t
 D'Ottavio,  T.W.,  G.I. Senum, and R.N. Dietz.  1988.  Error
 Analysis  Techniques for Perfluorocarbon Tracer Derived Multizone
 Ventilation. Rates.  Building and Environment 23:4, pp. 187-194.

 Dietz, R.N., R.W.  Goodrich, E.A. Cote, and R.F. Wieser.  1986.
 Detailed  Description and Performance of a Passive Perfluorocarbon
 Tracer System for Building Ventilation and Air Exchange
 Measurements.  Measured Air Leakage of Buildings.  ASTM STP 904,
 H.R.  Trechsel and P.L. Lagus, Eds., American Society for Testing
 and Materials,  Philadelphia, PA, pp. 203-264.

 Dixon, W.J., and  F.J. Massey, Jr.   1969.  Introduction to
 Statistical Analysis.  McGraw-Hill, New York, NY.

 GEOMET.   1992.   Evaluation of Radon-Resistant Construction
 Practices  in New  Homes in Florida.   Report No. IE-2588, GEOMET
 Technologies, Inc., Germantown,  MD.  DCA Contract No. 91 RD-41-
 14-00-009.
                                                              \
Grimsrud,  D.T.,  M.H. Sherman., ard R.C. Sonderegger.  1982.
 Calculating Infiltration:  Implications for a Construction
Quality Standard.  Proceedings of the ASHRAE-DOE Conference on
 the Thermal Performance of the Exterior Envelope of Buildings II,
Las. .Vegas, NV.

Grot,  R.A. and R.E. Clark.  1979.  Air Leakage Characteristics
and Weatherization Techniques for Low-income Housing.
 Proceedings of the ASHRAE-DOE Conference on the Thermal1
 Performance of .the Exterior Envelopes of Buildings, Orlando, FL.

Hutcheon,  N.B.  and G.O.P. Handegord.  1989.  Building Science for
 a Cold Climate.    Construction Technology Centre Atlantic, New
Brunswick, Canada.
                                25

-------
Leaderer, B.P., L. Schaap, and R.N. Dietz.  1985.  Evaluation of
Perfluorocarbon Tracer Technique for Determining Infiltration
Rates in Residences.  Environmental Science & Technology. Vol.
19, No. 12, pp. 1225-1232.

Sherman, M.H.  1989.  Analysis~.of Errors Associated with Passive
Ventilation Measurement Techniques.  Building and Environment.
Vol. 24, No. 2, pp. 131-139.  -

Sherman, M.H.  (Ed.).  1990.  Air Change Rate and Airtightness in
Buildings.  ASTM STP 1067, American Society for Testing and
Materials, Philadelphia, PA.
                                 i
U.S. Bureau of the Census.  1991.  Statistical Abstract of the
United States: 1991.  Washington, DC.  Table No. 1284, p. 727.

Versar.  1990.  Database of PFT Ventilation Measurements:
Description and User's Manual.  Versar Inc., Springfield, VA.
Prepared for USEPA Office of Toxic Substances Under Contract No.
68-02-4254, Task No. 39.

Wallace, L.A.  1987.  The Total Exposure Assessment Methodology
(TEAM)  Study:  Summary and Analysis (Volume 1).   Report Number  .
EPA/600/6-87/002a,< Washington, DC.

Wilson, A.L., S.D. Colome, P.E. Baker, and E.W.  Becker.  1986.
Residential Indoor Air Quality Characterization Study of Nitrogen
Dioxide. Phase I.  Volume 2;   Final Report.  Prepared for Southern
California Gas Company, Los Angeles, CA.
                                26

-------
                Appendix

Relationship Between Interzonal Airflows
    and House Volume  and Air Exchange

-------
 A.I.      Introduction

      Estimates of the house>volume and indoor-outdoor air
 exchange rate may not be sufficient for modeling inhalation
 exposure because the release of many airborne chemicals may be
 confined to certain areas or zones within the house.  In such
 cases,  the .airborne concentration within the zone of release
 generally will be higher than in other areas of the house,
 especially during the release period and until such time as the  .
 chemical has been completely mixed throughout the house.

      The communication of air between one zone of the house and
 the remainder is governed by a number of factors, including
 (1)  communication barriers such as walls, (2) door and window
 openings that promote air movement to,  from and within the house,
 _. ,-; ^ f*3\  ^^^-^j-^-^.^^-^^^^-, ^ £ ^^^_ _—, „ —,	.4~,i~ j -. ~ - '. ., ^ _. -, ._-v -_ -. ^ •-.  _i ~" _ _ ' .
 .'._.„ L „ .  '-,,,• — —	..._ . _...•,_.. .^ *_* ^ » —_ <^ ^. iw* *_; o* N— * * L* »j j- w j- w w *w± d J. j.
 heating/cooling systems and portable or stationary fans.  It is
 hypothesized that the rate of air movement within a.house will
 have some re-lationship to the indoor-outdoor air exchange rate,
.because  the air exchange rate will tend to be higher (1) when
 doors or windows are open,  which will  also promote air movement
 within the house,  or (2)  when temperatures are colder outdoors,
 which will place an increased demand on heating systems that move
 air either mechanically or through convective heat transfer.

 A. 2.       Methods.

     An  empirical relationship between interzonal airflows and
 the  indoor-outdoor air exchange rate was developed using the PFT
 database described in the main body of  this  report.   Two
 situations were examined: (1)  bedrooms,  for  which communication
 with the remainder of the house may be  restricted because of
 relatively small communication pathways  (doors),  and (2) the
 kitchen,  which generally has a more open communication path to
 adjacent areas such as the  living or dining  room.   The PFT
 database was searched for cases where  researchers labeled either
 the  kitchen or bedroom(s) as a separate  zone,  using the same
 exclusion criteria as described in Section 4 of this report.
 Based on these criteria,  408 bedroom cases and 360 kitchen cases
 were located.   The information extracted for these cases was the
 house volume (m3) , the air exchange rate  (h"1) and the  airflow
 rate (m3h"1)  between  the bedroom (s)  or the  kitchen and  the
 remainder of the house.              .

     This analysis could easily be confounded by different
 volumes  across houses or by inequalities in  the airflow rates
 into and out of the zone under study.   Consequently, the inflows
 and  outflows for each house were averaged,  and the averaged flow
 was  normalized by dividing it by the house volume:
                                A-l

-------
           QN =  (Q» + Qai)/2                                   (A_1}
                      V


where
           QN    is  the normalized airflow between zone 1 (kitchen
                or  bedroom)  and zone 2  (remainder of the house)
           Q12   is  the measured airflow rate from zone 1 to zone 2
           Q21   is  the measured airflow rate from zone 2 to zone 1
           V    is  the volume of the house.

     The  normalized interzonal airflow rate,  QN, has  the same
units  (h"1, or air  changes per hour) as the air exchange rate.
The relationship of QN to the  air exchange  rate  (I) was obtained
by linear regression analysis, yielding an equation of the-
following form:

        .   QN = a + bl                               '          (A-2)

where  -             .

           a is  the regression intercept  (h"1)
           b is  the regression slope.

A.3.       Results  and Applications

     For  the bedroom case,  the following relationship was
obtained  from the  regression analysis:

           QN = 0.078 + 0.311  (R2 = 0.56)           •           (A-3)

The relationship for the kitchen  case  was:

           QN = 0.046 -I- 0.391  (R2 = 0.58)                      (A-4)

Thus,  more  than 50  percent  of  the variance  across houses in  the
normalized  interzonal  airflow  rate  was explained ,by each
regression  equation.

     Based  on the  relationships given  above,  characteristic
airflow rates can  be postulated for two-zone  situations
conceptualized as  "bedroom  versus remainder of the  house"  and
"kitchen  versus remainder of  the  house."  Consider, for example,
a house with a volume  of 408 m3 (national average,  as determined
from information reported from the  Residential Energy. Consumption
Survey conducted by the U.S. Department  of  Energy*)  and an air
exchange  rate of 0.45  h'1  (national  average, as listed in Section
           USDOE.   1992.   Housing Characteristics 1990.  Report
No. DOE/EIA-0314(90), U.S. Department of Energy, Energy
Information Administration, Washington, DC.

                               A-2

-------
4 of this report) .   From  equation A-3, QN for the bedroom would
be 0.078 + 0.31 x  0.45, or  0.2175 h'1.  Multiplication by the
house volume yields  an  interzonal airflow rate of  88.7  m3h"1.  By
a similar process, the  airflow rate between the kitchen and the
remainder of the house  is calculated from equation'A-4  to be
90.4 m3}!'1.

     One cautionary  note  is in order when using the relationships
described above.   Some  or many of the researchers  contributing
measurements to the  PFT database may have defined  a zone as a
group of adjacent  bedrooms, i rather than an individual bedroom.
If so,  then the interzonal airflow rate for an individual bedroom
is likely to be lower than indicated by equation A-3.   Similarly,
the living room, which  like the kitchen has a fairly open
communication with the  rest of the house but also has a larger
volume  than the kitchen, might be expected to have a hiaher
                   race cnan indicated Cy equation A-4.
                               A-3

-------
  6027M01	.	
  REPORT DOCUMENTATION
             PAGE
  4. Tide and Subtitle
1. REPORT NO.
    Estimation of Distributions for Residential Air Exchange Rates
                                                            3. Recipient's Accession No.
                                                            5. Report Date
                                                             March 1995
                                                            6.
  7. AuthorU)
       M.D. Koontz,  H.E. Rector
                                                            8. Performing Organization Rept. No.
                                                             GEOMET Report No.  IE-2603
  a. Performing Organization Name and Address

       GEOMET Technologies, Inc.
       20251 Century Boulevard
       Germantown, MD  20874
                                                                                           10. Project/Task/Work Unit No.
                                                            11. Contract(C) or Grant(G) No.
                                                            (O 68-D9-0166
                                                                68-D3-0013
  12. Sponsoring Organization Name and Address

       United States Environmental Protection Agency
       Office of Pollution Prevention and Toxics
       Economics, Exposure and Technology Division
       Washington. DC  20460
                                                                                           13. Type of Report & Period Covered
                                                                      Final Report
                                                            14.
  15. Supplementary Notes       •

      The EPA Work Assignment Manager was Patrick Kennedy.
  16. Abstract (Limit: 200 words)
      This report describes the information sources, analysis methods and estimates of national and regional
      distributions  for annual average air exchange rates measured in United States residences.  All the measurement
      results derive from techniques involving constant release and time-integrated sampling of a family of
      compounds known as perfluorocarbon tracers (PFTs).

      The results indicate that a value of 0.18 air changes per hour (ACH) would be appropriate as a conservative number
      when modeling inhalation exposure, and that a value of 0.45 ACH would be appropriate when a typical air exchange
      rate is desired for modeling inhalation exposure. These values correspond to the  10th and 50th  percentile,
      respectively,  for the estimated national distribution of annual average air exchange rates.
 17. Document Analysis a. Descriptors

      Exposure Assessment
      Indoor Air Quality Monitoring
      Indoor Air Quality Modeling

  b. Identifiers/Open-Ended Terms
      Air Exchange
      Air Leakage
      Tracer Gas

  C. COSATI Reid/Group
 IS.JWailability Statement

      istribution Unlimited
                                             19. Security Class (This Report)
                                                 Unclassified
                                                                            20. Security Class (This Page)
                                                                                Unclassified
21. No. of Pages
        42
                                                                                                        22. Price
(See ANSI-Z39.18)
                                                   See Instructions on Reverse
                                                                       OPTIONAL FORM 272 (4-77)
                                                                           (Formerly NTIS-35)
                                                                        Department of Commerce

-------