EPA/600/R-15/271 I July 2016
www.epa.gov/homeland-security-research
United States
Environmental Protection
Agency
oEPA
Incorporating a Capability for
Estimating Inhalation Doses in TEVA
SPOT
Office of Research and Development
Homeland Security Research Center
-------
This page left Intentionally Blank
-------
EPA/600/R-15/271
July 2016
Incorporating a Capability for Estimating
Inhalation Doses in TEVA-SPOT
United States Environmental Protection Agency
Office of Research and Development
National Homeland Security Research Center
Cincinnati, Ohio 45268
2
-------
Contents
1 Introduction 6
2 Showering 7
2.1 Introduction 7
2.2 Behavior: Background 8
2.3 Behavior: Approach 9
2.3.1 Showering Frequency 9
2.3.2 Showering Duration 9
2.3.3 Showering Start Times 9
2.4 Behavior: Options 12
2.5 Estimating Doses 12
2.5.1 Aerosolization 14
2.5.2 Volatilization 17
2.5.3 Discussion 19
3 Humidifier Use 20
3.1 Introduction 20
3.2 Parameters 21
3.2.1 Behavior 21
3.2.2 Humidifier Description 22
3.2.3 Environment 22
3.3 Estimating Doses 22
3.3.1 Mass-Balance Model 22
3.3.2 Empirical Model 23
3.4 Summary 23
4 Additional Calculations 23
4.1 Reporting Results for CxT 24
4.2 Reporting Separate Results for C and T 25
4.3 Examination of Importance of Individual Exposure Events 25
References 27
3
-------
List of Figures
Figure 1: Air Concentration of Contaminant in a Shower 8
Figure 2: Cumulative Distribution of Starting Times for Single Showering Events 10
Figure 3: Estimated Weighted Probability Densities of Starting Times for Individuals with Two
Showering Events 13
Figure 4: Histogram for Time Separation between Showering Events for Cases with Two Events
14
Figure 5: Ratio of Shower-Related Inhalation Doses for Contaminated Aerosols Estimated by the
Empirical and Mass-Balance Approaches 17
List of Tables
Table 1. Summary of Options for Showering Behavior 15
Table 2. Parameters Required for Estimating Doses for Showering 20
Table 3. Parameters Required for Estimating Doses for Humidifiers 24
4
-------
Disclaimer
The U.S. Environmental Protection Agency (EPA) through its Office of Research and Development
funded, managed, and collaborated in the research described herein under an interagency
agreement with the U.S. Department of Energy through Contract DEAC02-06CH11357 with
Argonne National Laboratory. It has been reviewed by the Agency but does not necessarily reflect
the Agency's views. No official endorsement should be inferred. EPA does not endorse the
purchase or sale of any commercial products or services.
Questions concerning this document and its application should be addressed to the following
individual:
Robert Janke
National Homeland Security Research Center
U.S. Environmental Protection Agency
26 West Martin Luther King Dr.
Cincinnati, Ohio 45268
Janke.robert@epa.gov
5
-------
1 Introduction
Various uses of water have the potential to result in exposures to contaminants present in the
water via the inhalation pathway. Contaminants can be present in aerosols generated during
water use; volatile contaminants can be released to the air by such use. Considerable effort has
been devoted to studying chronic exposures to volatile contaminants in drinking water. Aerosol
production associated with water use has also been studied; the motivation again has been
chronic exposures to contaminants. A detailed examination of chronic inhalation exposures to
contaminants in drinking water has been provided in a monograph edited by Olin (Olin 1999).
During contamination events in a water distribution system (WDS), the potential exists for short-
term inhalation exposures to elevated concentrations of contaminants. Such acute exposures
have not received the same attention as has been devoted to chronic inhalation exposure to
contaminants such as disinfection byproducts. In particular, there do not appear to be any studies
of the system-wide inhalation exposures that could occur during a contamination event in a WDS.
Various domestic uses of water can release volatile contaminants or generate aerosols or do
both. The largest inhalation exposures to volatile contaminants likely result from showering
(Wilkes et al. 1992). A screening-level assessment of inhalation exposure doses indicates that
ultrasonic and cool-mist humidifiers and showering likely are the largest sources of exposures to
contaminants contained in aerosols (Hines et al. 2014).
This report presents the approach (the software design) that was used to incorporate inhalation
models in the U.S. Environmental Protection Agency's Threat Ensemble Vulnerability
Assessment, Sensor Placement Optimization Tool (TEVA-SPOT) (U.S. EPA 2015). The software
design outlined here and incorporated into TEVA-SPOT provides the capability for estimating
inhalation doses that result from the most important sources of contaminated aerosols and
volatile contaminants during a contamination event. TEVA-SPOT has had the capability to provide
estimates of ingestion doses associated with a contamination event for the population served by
a WDS. The approach presented here now provides TEVA-SPOT with a comparable capability for
inhalation exposures. This report does not provide derivations of equations, attempt to justify
use of specific values for parameters, or provide any recommendations for users of the approach.
These are all subjects that will be addressed more appropriately elsewhere. The purpose of this
report is two-fold: (1) to document the inhalation models that have been incorporated in TEVA-
SPOT and (2) to provide some brief background for the models.
In the approach used in TEVA-SPOT, individuals using water from a distribution system are
located at the nodes (junctions) in the system at which there is a nonzero demand for water.
Contaminant concentration during a contaminant event is determined at all system nodes using
EPANET (Rossman 2000). Concentration varies with time and location during an event.
Consequently, the behavior of individuals with respect to how they use water is important
because it determines the times at which possible exposures to contaminated water can occur.
Inhalation doses for individuals that result from exposures to contaminated water can be
6
-------
determined using the approach presented in this document. These individual doses can then
aggregated to obtain system-wide results.
Section 2 discusses the approach implemented in TEVA-SPOT for estimating inhalation doses for
showering. Section 3 presents the approach used for estimating inhalation doses from humidifier
use. Section 4 discusses some additional calculations that apply to both showering and humidifier
use. References are provided at the end of the document.
2 Showering
2.1 Introduction
Including a capability for estimating inhalation doses associated with showering in TEVA-SPOT
requires a model for showering behavior and a model for estimating contaminant concentrations
in the air in a shower. Contaminants may be present in the air in a shower due to the release of
volatile contaminants contained in the shower water or the formation of aerosol particles by the
shower head. In either case, the air concentration in a shower will vary with time approximately
as shown in Fig. 1. The air concentration will tend to approach an equilibrium value if the shower
is on for sufficient time and will decrease when the shower is turned off. The equilibrium
concentration is determined by the relative sizes of the source and loss rates for the contaminant.
Contaminant mass can be lost due to air exchanges between the shower stall and the adjacent
room and by deposition and other processes.
Inhalation dose is determined by the air concentration of the contaminant in the shower, the
length of time an individual remains in the shower while it is on, the length of time an individual
remains in the shower after it is turned off, and the individual's breathing rate. Air concentration
is determined by contaminant concentration in the water entering the shower and the physical
processes affecting the generation and loss of the contaminant in the air in the shower stall.
Water concentration is determined by the concentration in the distribution system, which varies
with time. Estimating the inhalation dose for an individual requires specifying the behavior of the
individual, namely the frequency at which showers are taken, the times when they are taken,
and their duration.
Models for behavior that describe showering frequency, duration, and timing are outlined and
discussed in Sections 2.2 - 2.4, followed in Section 2.5 by a presentation of approaches for
estimating inhalation doses based on several models for the physical processes that determine
the air concentrations of a contaminant in a shower.
The following discussion outlines how inhalation doses are determined for individual receptors
who shower using water from a distribution system. Results for individual receptors are
aggregated by scenario and ensemble as is already done in TEVA-SPOT for ingestion dose.
7
-------
shower on
t?f exit shower
c
o
c
CD
u
c
o
u
<
t1( shower off
Time
Figure 1: Air Concentration of Contaminant in a Shower. The vertical dotted lines indicate when
the shower is turned on, turned off, and when the individual exits the shower.
2.2 Behavior: Background
Given the limitations in available data, a number of assumptions are necessary when developing
a model that includes showering frequency, duration, and timing. In particular, the major
assumptions are the following: (1) showering duration is independent of showering frequency
and timing; (2) the behavior of each individual is the same for each day in a simulation; and (3)
the timing of grooming events as reported in time-use studies serve as an adequate proxy forthe
timing of showering events. Because less than 1% of the population takes more than two showers
per day (Wilkes et al. 2005), it will be assumed that all individuals take two or less showers per
day. Finally, because we generally have no demographic information on the individuals at
receptor locations (the network nodes), we do not consider how any parameters might vary with
demographic factors such as age or gender.
The approach used in TEVA-SPOT to account for receptor behavior associated with showering is
similar to that used to account for behavior related to ingestion of tap water. However, the model
used for ingestion of tap water has only two parameters involving behavior, namely timing and
volume; the model used for inhalation requires three. Each individual at each network node
needs to be assigned a daily number of showers (0, 1, or 2). If an individual is assigned one or
more showers, values for shower duration and starting time(s) also need to be assigned.
Estimates for showering frequency and duration are available (Wilkes et al. 2005). The American
Time Use Survey (ATUS 2013) provides data on when grooming events occur. These events
include showering, which is a major subset of grooming activity. The occurrence of grooming
events sets bounds on when showering can occur, but not all grooming events involve showering.
8
-------
Wilkes et al. (2005) provide results only for the length of time a shower is turned on. No similar
results appear to be available for the length of time an individual spends in the shower stall after
the shower is turned off. This quantity has a default value of zero, but can be provided by the
user, if desired.
2.3 Behavior: Approach
A detailed model for behavior is presented in this section. Options included in TEVA-SPOT for
modeling behavior are presented in Section 2.4 and also include more simplified models.
2.3.1 Showering Frequency
Showering frequency is determined using results presented in Table II of Wilkes et al. (2005),
which are based on the analysis of data from the National Human Activity Patterns Survey.
Considering individuals of all ages, for the day of the survey 22% reported not taking a shower,
60% reported taking one shower, and 18% reported taking two or more showers. (As noted
above, less than 1% take more than two showers per day.) The showering frequency to assign to
each individual at each node is determined by drawing a random number from U(0,1). If the
number does not exceed 0.22, the individual does not shower. If the number is greater than 0.22
but does not exceed 0.82, the individual takes one shower per day. If the number exceeds 0.82,
the individual takes two showers per day. Showering frequency is the same for each day in a
simulation. An option is provided that allows frequencies to be specified by the user.
2.3.2 Showering Duration
Showering duration is determined using the results presented in Table V or Fig. 4 of Wilkes et al.
(2005) that were developed by analyzing data from the Residential End Uses of Water Survey.
From this source the distribution of showering durations is approximately lognormal. On a
logarithmic scale the parameters for the distribution are ju = 1.92 and a = 0.493. (Given in
minutes, they would be exp(1.92) = 6.8 and exp(0.493) = 1.64.) The showering duration for an
individual is determined by drawing a random number from a lognormal distribution with these
parameters. To avoid a possibility of very long showering durations, any duration that exceeds
60 min is assigned a value of 61 min. (For the distribution being used, the probability that the
showering duration will exceed 60 min is less than 0.0001.) Showering duration is assumed to be
the same for both showers if an individual takes two showers per day and is assumed to remain
the same for each day in a simulation.
2.3.3 Showering Start Times
Estimates for showering start times are determined using results obtained from ATUS for the
timing of grooming. For individuals taking one shower per day, a starting time is obtained using
an empirical probability distribution based on ATUS timing data for grooming. When a second
shower is taken, its timing is influenced by the time of the first shower. To avoid issues related
to trying to account for this dependency using probability distributions, random samples are
taken of actual reported starting times for events in ATUS data. Starting times for an individual
9
-------
remain the same for all days in a simulation. No information is available on the time delay
between the start of a grooming event and the start of a shower. It is assumed to be zero.
One Event:
Fig. 2 shows the empirical, weighted cumulative distribution for starting times for single
grooming events for 2003-2012 ATUS data. It is based on the 54,094 events reported by the
54,094 individuals reporting one grooming event.
A text file ("cdf2003-12singles.txt" and included with this report), was developed using ATUS
data, that contains tab-separated values for the starting times and cumulative probabilities
plotted in Fig. 2. There are 101 rows in the file. The first entry in each row is the cumulative
probability (0 to 1.0) and the second entry is the corresponding starting time (0.0 to 24.0 hours).
.a
re
-Q
o
I—
Q_
CD
*+->
_ro
E
2
u
1.0 -
0.8 -
0.6 -
0.4 -
0.2 -
0.0 -
ATUS 2003-12 Events
Single Events Only
N = 54,094
1 1 T
5 10 15 20
Starting Times (hours, local time)
Figure 2: Cumulative Distribution of Starting Times for Single Showering Events
In TEVA-SPOT, random starting times are determined by inversion using the results plotted in Fig.
2 and contained in the text file. Inversion is accomplished by fitting a spline function (call it g) to
the values for starting time (T) and cumulative probability (P) in Fig. 2, so that T = g(P). A random
starting time, t, is given by t = g(p), where p is a random number drawn from U(0,1). t is
distributed according to the empirical distribution in Fig. 2. Random starting times are
determined in this way for all individuals at all nodes who have one showering event per day.
10
-------
Two Events:
A second text file ("two events 2003-12.txt" and included with this report) was developed that
contains data for all 36,652 ATUS respondents who reported two grooming events in 2003 to
2012. Results in this file are used in TEVA-SPOT to generate random starting time for individuals
who take two showers per day. The file has 36,652 rows and five tab-separated columns. The
first column contains the year the data were collected and the second column contains the ATUS
identifiers used for the respondents. The third column contains the starting times in hours local
time for the first event and the fourth column contains the starting time in hours local time for
the second event. The fifth column provides the ATUS weights for the respondents. Weights are
needed to compensate for the manner in which sampling and data collection were carried out in
ATUS. For example, some demographic groups were oversampled to ensure an adequate sample
size and more sampling was done on weekends than weekdays. Response rates varied by
demographic group and day of the week. The weight for a particular respondent is the number
of person-days that the results for the respondent represent in the entire U.S. population. The
total number of person-days for the U.S. is the size of the U.S. population multiplied by the
number of days in a year.
For some respondents the starting time for the second event is numerically smaller than the
starting time for the first. For example, the time of the first event could be 10:00 hours and the
time of the second event could be 1:00 hours. Such cases occur when the second event begins
after 24:00 hours. The survey period was from 04:00 hours on the first day for which respondents
provided information to 04:00 hours on the second. In cases in which the starting time of the
second event is numerically smaller than the starting time of the first event, the order of the
events is reversed when assigning starting times.
Estimated weighted densities for starting times for the events included in the file are shown in
Fig. 3. The figure is provided for illustration purposes only; the two distributions are not
independent.
Weights for respondents can vary substantially. The ratio of the maximum to minimum weights
in the file is about 251. Consequently, consideration of the weights is important when results in
the file are used.
Weighted sampling with replacement is used to obtain starting times for individuals who take
two showers per day. Starting times for two events for an individual are obtained by randomly
selecting one of the 36,652 cases in the file, considering the weights for the cases. Sampling is
done with replacement when more than one individual is considered.
The distribution of the time separation between events for cases with two events is shown in Fig.
4. The maximum separation between events is 12 hours. For more than 81% of the cases the
separation between events is 6 hours or longer. However, in some cases the difference between
reported starting times for the two events can be small. Of the 36,652 cases with two events, 24
11
-------
(0.07%) have a time separation of 5 min or less, 73 (0.2%) have a time separation of 10 min or
less, 164 (0.4%) have a time separation of 15 min or less, 514 (1.4%) have a time separation of
30 min or less, and 1,186 (3.2%) have a time separation of 60 min or less.
To avoid cases in which time separations are smaller than seems realistic for the time separating
two showers, or in which the time difference is smaller than the shower duration, an option has
been included that allows users to reject any cases in the data file in which the time separation
is less than some specified value (e.g., 0.5 or 1.0 hours).
2.4 Behavior: Options
TEVA-SPOT allows users to choose among various options for ingestion behavior; to provide
similar flexibility for considering inhalation exposures, several options related to inhalation are
also provided. Using a label similar to those used for ingestion, the most detailed option is PD1,
for which the timing is specified by a probabilistic model, as is the duration. The simplest option,
FM, has a fixed time (or times for individuals taking two showers) and a fixed duration. The
intermediate option, FD, has a fixed time(s) and a duration described by a probabilistic model.
For PD1, the models for durations and times are described in Sections 2.3.2 and 2.3.3,
respectively. For FD, the model for duration is described in Section 2.3.2. The time for a shower
for individuals taking one shower is either 06:30 or 21:30 hours local time. These times are the
most common times when people who take one shower per day actually shower (on the basis of
estimated peaks in the weighted histograms for starting times for grooming events). For
individuals taking one shower, 70% take it at the first time and 30% take it at the second time (on
the basis of the fraction of individuals taking showers in the intervals from 02:00 to 14:00 hours
and from 14:00 to 02:00 hours, respectively). For individuals taking two showers, the most
common times are also 06:30 and 21:30 hours local time. For option FM, the times for showers
are the same as for FD; the duration for all showers is 7.7 min, the average shower duration.
The approaches used in TEVA-SPOT for the three options for behavior are summarized in Table
1. For each individual, behavior is the same for all days in a simulation. The only parameters that
a user can supply for the behavior models are shower frequencies and the minimum time
separation between two showering events. Selection of an option (e.g., PD) for behavior is always
required.
2.5 Estimating Doses
Contaminants may be present in shower air contained in aerosols or as a gas. For aerosols, a
mass-balance model and an empirical model for estimating doses are available in TEVA-SPOT.
These models are discussed in Section 2.5.1. For volatile contaminants, two approaches based
on mass balance are available, one using a mass-transfer-coefficient model and the other a
1PD, FM, and FD are labels, not abbreviations. The first letter indicates whether the timing model is probabilistic (P)
or uses a fixed time (F). The second letter indicates whether the duration model is probabilistic (D) or uses a fixed
duration (M).
12
-------
transfer-efficiency model. These are discussed in Section 2.5.2. Volatile chemical contaminants
may be present as a gas and chemical, biological, and radiological contaminants may be
contained in aerosols. If a chemical has even low volatility, the contribution of the portion of the
mass of the chemical present in aerosols generated during showering to total inhalation dose can
be neglected.
in
r\l
o
1st Event
2nd Event
o
r\l
o
in
T—I
o
o
LO
o
d
o
o
d
0 5 10 15 20 25 30
Starting Times (hours)
Figure 3: Estimated Weighted Probability Densities of Starting Times for Individuals with Two
Showering Events (note that the two distributions are not independent and this figure is provided
for illustration only). For cases in which the starting time of the second event is numerically
smaller than the starting time of the first event, 24 hours was added to the starting time of the
second event before densities were determined.)
Using models for the physical processes affecting contaminants in a shower, air concentrations
of a contaminant in a shower stall were determined separately for aerosols and volatile
chemicals. These concentrations were integrated over the duration of a shower and combined
with an average breathing rate to obtain the estimates for inhalation dose presented here and
available in TEVA-SPOT. These doses are the mass of the contaminant that enters the body by
inhalation and are actually potential inhalation doses because not all of the contaminant mass
will necessarily remain in the body; some is exhaled.
Doses are determined separately for each showering event and for each receptor because
contaminant concentration varies with time during a simulation. Doses for each of the separate
13
-------
showering events for each individual receptor are added to obtain a cumulative dose for each
receptor for the period of the simulation. These cumulative results for individual receptors can
be binned, as is done for ingestion doses, so that the number of receptors with cumulative
inhalation doses in each bin specified by the user are available for each scenario evaluated.
Alternatively, dose-response data for a particular contaminant can be provided to TEVA-SPOT to
allow estimation of health-effect end points. Section 4 discusses a capability that has been added
to TEVA-SPOT that allows the user to determine the relative contribution of the most important
showering event for each individual to the cumulative inhalation dose for the individual. This
capability allows the user to examine the degree of conservatism involved in using cumulative
doses.
o
o _
o
m
>-
u
c
cu
=5
CT
O)
o
o
o
r\i
o
o _J
o
o
o —
ID
O —
Time Difference between E vents (hours)
Figure 4: Histogram for Time Separation between Showering Events for Cases with Two Events
(N = 36,652)
2.5.1 Aerosolization
The more detailed model for aerosolization is based on transient mass balance and requires the
user to specify an empirically determined value for aerosol generation rate. The user also may
specify an air exchange rate for the shower stall. Rates for other removal processes (e.g.,
settling or impaction) can be included in the value used for air exchange rate. The second
model for aerosolization uses a purely empirical constant to relate contaminant concentration
in the air to contaminant concentration in the water entering the shower.
14
-------
Table 1. Summary of Options for Showering Behavior
Option3
Parameter
Model or values for parameter
All
Shower frequency
60% take one per day, 18% take two per day,
22% take none. User can specify frequencies
PD, FD
Shower duration
~ L(ju = 1.92, a - 0.493), with any durations
greater than 60 min set equal to 61 min
FM
Shower duration
7.7 min
PD
Times - one event
CDF in Fig. 2
PD
Times - two events
Sample times of actual events
User can specify minimum time separation,
if desired; default separation = 0
FD, FM
Times - one event
06:30 (70%) or 21:30 (30%) hours, local time
FD, FM
Times - two events
06:30 and 21:30 hours, local time
aPD, FM, and FD: The first letter indicates whether the timing model is probabilistic (P) or uses a fixed time (F). The
second letter indicates whether the duration model is probabilistic (D) or uses a fixed duration (M).
Mass-Balance Model:
The mass-balance model assumes complete, immediate mixing of the contaminant in the air in
the shower stall. Contaminant concentration in the shower air as a function of time was
estimated using an approach similar to that used in Zhou et al. (2007). This time-varying
concentration was integrated to give the inhalation dose, D, (in mg or number of cells [#]) for one
showering event:
ex p(—k2T2))
(1}
where B is the breathing rate (m3/min), G is the aerosol mass generation rate (mg/min),/is the
mass or number fraction of the contaminant in the water (see discussion below), Vs is the volume
of the shower stall (m3), fa is the removal rate for aerosols (min"1) while the shower is on, Ts is
the duration of the shower (min), Ti is the time an individual remains in the shower after it is
turned off (min), and fa is the removal rate for aerosols (min"1) after the shower is turned off.
The removal rates fa and fa a re equal to the sum of the air exchange rate for the shower stall and
the loss rate of particles due to deposition and other processes. In general, values for the various
rates will depend on whether the shower is on or off.
D = BGf.'ik, i;,l I,
II
e\v\ —k
rsr:<
11 -c-xju-A^rji (i
15
-------
If 7~2 = 0, Eq. 1 reduces to the following:
BG f l
D — ——\TS — —{1 - exp{-ki%))
ki Vs k\ '
(2)
If ki = foand Tj> 0, Eq. 1 reduces to the following:
D = f|[T» ~ ^-(i - i)]
(3)
Contaminant air concentration in the shower stall is proportional to the aerosol mass generation
rate (G), which determines the mass of aerosols in the air. Only a fraction of this aerosol mass is
contributed by the contaminant; this fraction is /, the fraction of the water mass that is
contaminant. It is assumed that the fraction of aerosol mass that is contaminant is the same as
the fraction of incoming water mass that is contaminant. If the contaminant concentration in the
shower water is 1 mg/L, then/= 10"6and has no units. If the contaminant in the shower water is
an organism or spore, then/is the number of organisms or spores per milligram of shower water.
If the concentration of organisms or spores in the shower water is 1 per L, then, since 1 L has a
mass of 1 kg,/= 10"6 mg-1. In this case,/has units of #/mg./is determined during the simulation
for each receptor location (node).
Empirical Model:
Using an empirically determined constant (b) that gives the ratio of contaminant concentration
in air and in water (Pandis and Davidson 1999), inhalation dose (in mg or number of cells [#]) for
one shower event can be estimated using the following equation:
where B is the breathing rate (m3/min), Cwis the contaminant concentration in water (mg/L or
#/L), Ts is the duration of the shower (min), and the ratio b has units of L/m3. This empirical model
is available for use in TEVA-SPOT.
This approach assumes that equilibrium conditions exists throughout the duration of the shower
and that transient effects related to turning the shower on and off can be neglected. Inhalation
dose after the shower is turned off is assumed to be zero. If this approach is based on an
equilibrium air concentration that is the same as the equilibrium air concentration predicted by
the mass-balance approach, it will overestimate the inhalation dose relative to that estimated by
the mass balance approach because it assumes that the contaminant air concentration is at the
equilibrium value during the period immediately after the shower is turned on, when air
concentration is actually smaller than the equilibrium value. For short duration showering events,
equilibrium conditions may not occur during most of the event, if at all.
D — bBCwTs
(4)
16
-------
The ratio of the estimated dose obtained using the empirical approach to the estimated dose
obtained using the mass-balance approach (with Tj= 0) is shown in Fig. 5, asa function of shower
duration, for several values of the removal rate. For a shower of average duration, the empirical
approach yields estimated doses that are about 2 to 3 times larger than those obtained by the
mass-balance approach, forvalues of the removal rate in the range of about 0.1 to 0.3 min-1. The
dose estimated by the empirical method approaches that estimated by the mass-balance method
for long shower durations. Values for b are not contaminant specific. However, very few values
for b have been published.
o _
rsi
a! ^ —
-------
Mass-Transfer-Coefficient Model:
Concentration of the contaminant in shower air was determined following an approach outlined
in Little and Chiu (1999). This concentration was then integrated over the duration of the shower
to obtain an inhalation dose.
The flow of water through the shower is assumed to be simple plug flow: velocity is constant
across the flow at any location from the shower head to the floor of the shower. The perimeter
of the stream of water is constant.
To simplify the expression for inhalation dose, si (units are mg/min/m3) and r (units are min-1)
are defined as follows:
»i = 1 • < it'\-KoiA/Q))
1» (5)
r = TiW//T;(1 _ <'M~Ko,.a/Q)) + (6)
where Q is the water flow rate for the shower (L/min), Co is the water concentration of the
contaminant when it enters the shower (mg/L), Vs is the volume of the shower stall (m3), Kol is
the overall mass-transfer coefficient (L/min/m2) based on the liquid-phase concentration, A is
the interfacial area (m2) that the mass transfer flux passes through, H is Henry's Law constant
(dimensionless) for the contaminant2, and kon is the air exchange rate for the shower (min-1)
when it is on.
Using these definitions, and letting k0ff be the air exchange rate for the shower when it is turned
off, the inhalation dose, D, (in mg) for volatiles for one showering event is given by the following:
Bs i 1 1
D = [Ts - - (1 - ex p(-rTs)) + -— (1 - exp{-rTs)) (1 - exp(-koffT2))]
r r K0ff
(7)
where, again, B is the breathing rate (m3/min), Ts is the duration of the shower (min), and Tj is
the time an individual remains in the shower stall after the shower is turned off (min).
2 The factor of 1000 (units L/m3) is present because water concentration is in mg/L and air concentration is in mg/m3.
His the ratio of equilibrium gas-phase and aqueous-phase concentrations of the contaminant. Henry's law constants
depend on temperature and are generally given for a reference temperature (298.15 K or 25 °C). The user can
provide a value for H for the reference temperature or provide a temperature-adjusted value, considering the
temperature of the shower water.
18
-------
If 7~2 = 0, Eq. 7 reduces to the following:
D
Bs
1
(1 — exp(—rTg))\
r
r
(8)
Transfer-Efficiency Model:
Mass-transfer coefficients may not always be available and transfer efficiencies have been
measured in some studies. The transfer efficiency is the weight fraction of the contaminant that
is volatilized from the shower water. Air concentration of contaminant was determined following
the method used by Xu and Weisel (2003) and integrated over the duration of the shower to
obtain an inhalation dose. Using a transfer efficiency, the inhalation dose (mg) for volatiles for
one shower event is given by the following:
where S2= QTeCo/Vsand Te\s the average transfer efficiency for removal of volatile chemicals from
shower water. 7~ecan vary from 0 to 1. If Tj = 0, Eq. 9 reduces to the following:
2.5.3 Discussion
Table 2 summarizes the parameters that a user needs to provide to use the various approaches
for estimating inhalation dose. The table also provides default values for most parameters. The
user also needs to specify the bins to be used to report results for inhalation dose. If health-effect
end points (e.g., fatalities) are needed, the user must specify the appropriate dose- response
information.
The default value for breathing rate provided in the table is the mean value for most age groups
given in Table 6-2 of EPA's Exposure Factors Handbook (U.S. EPA 2011) for light intensity activity.
The value is the recommended short-term exposure value for males and females combined.
For all the models used for estimating doses, dose is proportional to the contaminant
concentration in the water feeding the shower. Consequently, doses can be estimated for unit
concentrations of a contaminant foreach receptor before beginning a simulation and these doses
can then simply be multiplied by water concentrations determined during the simulation foreach
showering event.
Bs¦
1
(10)
19
-------
Table 2. Parameters Required for Estimating Doses for Showering
Process
Model
User-provided
parameters3
Default values
Aerosolization
Mass balance
B
0.012 m3/min
(Eq.l)
G
6.0 mg/min
Vs
2 m3
ki
0.3 min"1
ki
0.1 min"1
t2
0
Empirical
b
No default
(Eq.4)
B
0.012 m3/min
Volatilization
Mass transfer coeff.
B
0.012 m3/min
(Eq.7)
Q
9 L/min
Vs
2 m3
H
-b
KolA
-b
kon
0.15 min"1
koff
0.075 min"1
Ti
0
Transfer efficiency
B
0.012 m3/min
(Eq.9)
Q
9 L/min
Te
0.8
Vs
2 m3
kon
0.15 min"1
koff
0.075 min"1
t2
0
Parameters are defined in the text. bThese parameters are contaminant specific.
3 Humidifier Use
3.1 Introduction
The approach for estimating inhalation doses associated with the use of an ultrasonic or cool-
mist humidifier is essentially the same as that used for estimating inhalation doses associated
with showering. The only differences are that the values of the parameters are different and
transient effects are neglected. These humidifiers are very efficient generators of aerosol
particles. Therefore, even though the rate at which water is used in a humidifier is much less than
that in a shower, the rate at which aerosols are generated can be larger. Because of the relatively
small volume of water used in a humidifier, they are less important as a source of volatile
contaminants than a shower, which can remove a sizeable fraction of volatile contaminants from
a much larger volume of water.
20
-------
Unfortunately, relatively little information appears to be available on humidifier use.
Consequently, an approach to estimating inhalation doses that is suitable for performing
sensitivity analyses is needed. This section describes the capability added to TEVA-SPOT for
carrying out such analyses.
Relative to showering, the duration of exposures to contaminants released by a humidifier can
be very long. The average duration of a shower is less than eight minutes. Exposures related to a
humidifier can last eight hours or more. Consequently, the transient effects associated with
turning on and turning off a humidifier are less important than such effects are for a shower. The
approach presented here for humidifiers is based on the assumption that contaminant air
concentration is constant and at its equilibrium value throughout the entire exposure period.
Turn-on and turn-off periods are neglected.
Estimating inhalation doses for humidifiers requires information on three different topics: (1) the
behavior of the user of the humidifier, (2) the characteristics of the humidifier, and (3) the
environment in which the humidifier is used. Information on the last two topics is available.
Limited information is available for user behavior.
This section outlines how inhalation doses can be determined for a single use of a humidifier (a
humidifier event) that has been filled with contaminated water from a distribution system.
Humidifiers are assumed to be filled daily immediately before they are operated. Individuals who
used humidifiers are assumed to use them every day during the course of a simulation. The only
difference between the daily events for an individual is that contaminant concentration in the
water used to fill the humidifier will generally change. Total inhalation dose for a receptor for a
simulation is determined by summing the doses for the separate, daily events. Results for
individual receptors are combined for each scenario to obtain impacts by dose level as is done
for ingestion dose. The user needs to specify the dose bins to use to reports impacts by scenario.
The various parameters needed to estimate inhalation doses are discussed in Section 3.2 and the
approaches used to estimate dose are given in Section 3.3.
3.2 Parameters
3.2.1 Behavior
Several parameters need to be quantified to describe behavior: (1) What fraction of the
population uses a humidifier, (2) When is water added to the humidifier, and (3) How long is the
humidifier used. Good information is not available on these parameters.
The fraction of the population that uses a humidifier can vary from 0.0 to 1.0. All individuals using
water from the network are potential users of humidifiers and use is assigned randomly to the
fraction of the population specified. The default value is 0.2. Limited information is available on
the fraction of the population that uses a humidifier. TEVA-SPOT allows the user to easily vary
this and other parameters in order to perform sensitivity analyses.
21
-------
Receptors are assumed to be exposed to contaminants released by a humidifier for some
duration of time specified by the user. The default assumption is that the humidifier is used while
sleeping and the duration of exposure is eight hours.
No data are likely to be available on when humidifiers are filled with water. In order to determine
contaminant concentration in the fill water, a time must be specified. The default assumption is
that humidifiers are filled immediately before use and that this occurs at 22:00 hours.
Behavior is assumed to be the same for every day in a simulation. Only one event involving
humidifier use is assumed to occur in a day.
Although not strictly a parameter related to behavior, a breathing rate for receptors is needed.
This rate needs to be consistent with the assumed use of the humidifier. If the humidifier is used
while sleeping, then the appropriate breathing rate is the average short-term rate for adults
while sleeping. This value is 0.3 m3/h, which is the default value. Values for breathing rates are
provided in Table 6-2 of U.S. EPA (2011).
3.2.2 Humidifier Description
The humidifier is assumed to convert all water in the humidifier into inhalable aerosol particles.
The only parameter needed to describe the humidifier is the rate at which water is used. The
default rate is 0.5 L/h.
3.2.3 Environment
The only parameters needed to describe the environment are the volume of the room in which
the humidifier and the receptor are located and the total loss rate for the aerosols. The default
values are 30 m3 (1,060 ft3) and 1.0 h"1, respectively.
3.3 Estimating Doses
Two approaches are available in TEVA-SPOT for estimating inhalation doses resulting from
exposure to contaminated aerosols generated by a humidifier. The first approach is based on
mass balance and requires the user to specify the water use rate for the humidifier and the
removal rate for aerosols. The second approach uses a purely empirical constant to relate
contaminant concentration in the air to contaminant concentration in the water used in the
humidifier. As is the case for inhalation doses estimated for showering, the estimated doses
associated with humidifier use are also potential inhalation doses and equal the mass of
contaminant that enters the body by inhalation.
3.3.1 Mass-Balance Model
The mass-balance model assumes complete mixing of the contaminant in the air of the room and
neglects transient effects. The inhalation dose, D (mg or number of cells) for one humidifier event
is shown in Eq. 11:
22
-------
(11)
where B is the breathing rate (m3/h), G is the aerosol volume generation rate (L/h), Cw is the
contaminant concentration in the humidifier water (mg/L or #/L), Th is the duration of exposure
(h), k is the removal rate for aerosols (h_1), and V is the volume of the room (m3). The removal
rate is the sum of the air exchange rate for the room and the loss rate for the aerosols.
The aerosol generation rate equals the water use rate. All water is assumed to be converted into
inhalable aerosols and released into the room, which is a conservative assumption. The
contaminant concentration in the water equals the contaminant concentration in the water in
the distribution system at the location of the receptor at the time the water is withdrawn for use
in the humidifier.
3.3.2 Empirical Model
Using an empirically determined constant (b) that gives the ratio of contaminant concentration
in air and in water, inhalation dose (in mg or number of cells) for one humidifier event can be
estimated using the following equation:
where the ratio b has units of L/m3.
3.4 Summary
The various parameters that a user needs to be specify when estimating humidifier related doses
and their default values are summarized in Table 3. The user also needs to specify the bins to be
used to report numbers of individuals with inhalation doses greater than the threshold values
specified for the bins. Additionally, the user can specify the dose-response information needed
to determine associated health-effect end points.
Sections 2 and 3 present approaches for estimating inhalation doses associated with showering
or humidifier use. This section discusses options for (1) presenting results in terms of exposure
duration and air concentration of contaminant rather than dose and (2) providing results that
can be used to evaluate the relative importance of the major exposure event associated with
either showering or humidifier use that an individual experiences during a contamination event.
An individual may experience multiple exposure events during the period in which a WDS is
contaminated.
D — bBCwTh
(12)
4 Additional Calculations
23
-------
Table 3. Parameters Required for Estimating Doses for Humidifiers
Model
User-provided parameters3
Default value
Mass balance
B
0.3 m3/h
(Eq.ll)
G
0.5 L/h
V
30 m3
k
1.0 h"1
Th
8 h
No default
Empirical
b
value
(Eq. 12)
B
0.3 m3/h
Th
8 h
Both
Fraction of population
using a humidifier
0.2
Time water is added
22:00 hours
aThe user-provided parameters in this column are defined in the text.
For some contaminants there is a preference for using contaminant concentration (C) and
exposure duration (T) rather than dose as the metric of interest. Two options are available in
TEVA-SPOT that allow the user to examine consequences in this alternative way. The approaches
used in these options are based on the assumption that the great majority of an individual's dose
is accumulated during one exposure event. This may not be a good assumption; however, the
dose estimates available to the user give only cumulative values, so the user cannot be certain
about the validity of the assumption. Therefore, an additional option is included that allows the
user to examine the degree to which cumulative dose is the result of one or more exposure
events.
4.1 Reporting Results for CxT
For the case in which inhalation dose is the result of a single exposure event of duration T (min)
with a constant air concentration of C (mg/m3) for an individual with a breathing rate B (m3/min),
the dose (mg) is given by:
D = BCT (13)
Therefore, if D is known, CxT is simply D/B. Note that for showering, T = Ts+ Tj. Generally, values
of CxT are of interest only for some chemical contaminants.
Assuming that the doses determined by the various approaches given in Section 2.5 or Section
3.3 are the result of a single exposure event (or at least a series of events in which one is
24
-------
dominant), an estimate for CxT for the dominant event can be determined for each receptor by
dividing the cumulative dose calculated for the receptor by the breathing rate. This is done by
TEVA-SPOT in parallel with dose calculations, with a separate binning of CxT values. The user
needs to specify the bins to be used. Alternatively, if no dose calculations are being done, values
for CxT can be obtained by simply performing the dose calculations with a value of B = 1. The
approach outlined here gives the product of the sum of the average concentrations for each of
the exposure events for an individual multiplied by the event duration (which is the same for all
events of the same type for a given receptor). If exposure to the contaminant occurs during only
one exposure event, then this approach gives the product of the average air concentration of the
contaminant for that event times the duration of the event.
4.2 Reporting Separate Results for C and T
For some contaminants, the health effects resulting from exposure to the contaminant are the
same, or approximately the same, for different events if the product CxT remains the same, even
if the values for C and T vary from event to event. In other words, the consequences do not
depend on dose rate. This has been the implicit assumption in all calculations involving multiple
exposures spread over a long simulation. However, for some contaminants the values of C and
T, and not just their product, are important. Therefore, TEVA-SPOT allows reporting of results for
both C and T and not just their product or dose.
If values for C and T are to be reported, these values should be for the major event for a receptor
during the simulation. The major event is the one for which the water concentration of the
contaminant is largest. Therefore, for each receptor for the event with the maximum water
concentration the exposure duration (Ts+ Tj for showering and 7~bfor humidifier use) and the
average air concentration in the shower stall or the room in which a humidifier is used is retained
JL
and then binned for the scenario. The average air concentration equals fir for the event. The
binning is two dimensional because the two quantities C and T are needed. The results are the
number of receptors with values of C and T who are in each bin. The user needs to specify the
bins for both C and T.
4.3 Examination of Importance of Individual Exposure Events
The results obtained using the preceding two options are only useful if the great majority of
inhalation dose is accumulated during a single exposure event. This option provides the ability to
determine if this is a good assumption for a particular evaluation for a network.
As noted above, the event of most importance for each individual is the one with the largest
water concentration of the contaminant. The relative contribution of this event is the dose
estimated for the event divided by the total dose determined for the individual for the entire
simulation. For each individual who showers or uses a humidifier and who accumulates an
inhalation dose greater than zero the relative contribution of the major event is determined. Its
value will always be less than or equal to one. These values are binned for all individuals for each
25
-------
scenario using 10 equally spaced bins extending from 0 to 1.0. The binned results are reported
for the scenario and also combined to yield binned results for the entire ensemble.
Many receptors will receive little or no dose and are of no particular importance for these
calculations. The user needs to specify multiple thresholds for dose above which binned relative-
importance values are to be reported. For example, if the dose threshold specified is 1.0 mg, the
results will be, for each of the scenarios evaluated and for the ensemble of scenarios, the
numbers of individuals receiving an inhalation dose of this size or greater whose relative-
importance values for the major exposure event are in each of the 10 bins that extend from 0 to
1.0. Using a number of dose thresholds it is possible to determine if the relative importance of
major exposure events changes as the dose threshold changes.
26
-------
References
American Time Use Survey (2013). American Time Use Survey User's Guide, Bureau of Labor
Statistics, June.
Hines, S.A., et al. (2014). "Assessment of relative potential for Legionella species or surrogates
inhalation exposure from common water uses." Water Research, 56:203-213.
Little, J.C., and Chiu, N. (1999). "Transfer of volatile compounds from drinking water to indoor
air", in Olin, S.S. (ed), Exposure to Contaminants in Drinking Water: Estimating Uptake through
the Skin and by Inhalation, CRC Press, Boca Raton, FL, pp. 90-94.
Olin, S.S. (ed.) (1999). Exposure to Contaminants in Drinking Water: Estimating Uptake through
the Skin and by Inhalation, CRC Press, Boca Raton, FL.
Pandis, S.N. and Davidson, C. (1999). "Aerosols and water droplets", in Olin, S.S. (ed), Exposure
to Contaminants in Drinking Water: Estimating Uptake through the Skin and by Inhalation, CRC
Press, Boca Raton, FL, p. 107.
Rossman L.A. (2000). EPANET 2 Users Manual. Cincinnati, OH: U.S. EPA, Office of Research and
Development, National Risk Management Research Laboratory Report, EPA/600/R-00/057.
U.S. Environmental Protection Agency (2011). Exposure Factors Handbook, 2011 Edition (Final),
Washington, DC, EPA/600/R-09/052F.
U. S. Environmental Protection Agency (2015). Models, Tools and Applications for Homeland
Security Research. http://www2.epa.gov/homeland-securityresearch/models-tools-and-
applications-homeland-security-research#tab-l. Accessed on 22 October 2015.
Wilkes, C.R., Small, M.J., Andelman, J.B., Giardino, N.J., and Marshall, J. (1992). "Inhalation
exposure model for volatile chemicals from indoor uses of water." Atmospheric Environment,
26A(12): 2227-2236.
Wilkes, C.R., Mason, A.D., and Hern, S.C. (2005). "Probability distributions for showering and
bathing water-use behavior for various U.S. subpopulations." Risk Analysis, 25(2): 317-337.
Xu, X., and Weisel, C.P. (2003). "Inhalation exposure to haloacetic acids and haloketones during
showering." Environmental Science and Technology, 37(3): 569-576.
Zhou, Y., Benson, J.M., Irvin, C., Irshad, H., and Cheng, Y-S. (2007). "Particle size distribution and
inhalation dose of shower water under selected operating conditions." Inhalation Toxicology,
19(4): 333-342.
27
-------
vvEPA
United States
Environmental Protection
Agency
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT NO. G-35
Office of Research and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
------- |