&EPA
United States
Environmental Protection
Agency
Theoretical and Experimental
Analysis of Important Parameters
for Determining the Impact of a
Biological Attack on a Building
FINAL REPORT
-------
-------
EPA/600/R-08/052 December 2008 www.epa.gov/ord
FINAL REPORT ON
Theoretical and Experimental Analysis of
Important Parameters for Determining the
Impact of a Biological Attack on a Building
Contract No. GS-10F-0275K
Task Order 1105
Prepared for
Joseph Wood and Les Sparks, Project Officers
U.S. ENVIRONMENTAL PROTECTION AGENCY
Research Triangle Park, NC
Prepared by
Brian E. Hawkins, Ph.D. and Kent C. Hofacre
"WARNING - This document may contain technical data whose export is restricted
by U.S. law. Violators of export control laws may be subject to severe legal penalties.
Do not disseminate this document outside the United States or disclose its contents to
non-U.S. persons except in accordance with applicable laws and regulations and after
obtaining any required authorizations."
BATTELLE COLUMBUS OPERATIONS
505 King Avenue
Columbus, Ohio 43201-2693
Office of Research and Development
National Homeland Security Research Center, Decontamination and Consequence Management Division
-------
Disclaimer
This report is a work prepared for the United States government by Battelle. In no event
shall either the United States government or Battelle have any responsibility or liability
for any consequences of any use, misuse, inability to use, or reliance on the information
contained herein, nor does either warrant or otherwise represent in any way the accuracy,
adequacy, efficacy, or applicability of the contents hereof.
-------
Table of Contents
Glossary vii
Executive Summary ix
1.0 Introduction 1
2.0 Objective 1
3.0 Scope 1
4.0 Model Development 3
4.1 General Approach 3
4.2 Simplified Model Development 3
5.0 Experimental Methods 7
5.1 Test Design 7
5.2 Test Building Description 8
5.3 Release Methods 12
5.4 Sampling Methods 13
6.0 Preliminary Efforts 15
6.1 Airflow Measurements 15
6.2 Leakage Tests 20
7.0 Experimental Results 21
8.0 Discussion of Experimental Results 41
9.0 Model Impact Analysis 43
9.1 Impact Analysis Approach 43
9.2 Impact Analysis Limitations 47
9.3 Preliminary Impact Analysis Results 48
9.4 Parameter Space Map Impact Analysis 50
9.4.1 "Large" Notional Building Parameter Space Map Results 51
9.4.2 "Small" Notional Building Parameter Space Map Results 54
9.5 Functional Analysis Guidelines 59
9.5.1 Contaminant Transport Dominated by HVAC Mechanisms 60
9.5.2 Contaminant Transport Dominated by Interzonal Leakage 60
9.5.3 Perfect Filtration 61
10.0 In-Room Air Cleaners 63
11.0 Conclusions and Recommendations 65
12.0 References 67
Appendix A. SF6 Experimental Methods and Results A-l
Appendix B. Preliminary Simulation Results B-l
Appendix C. Data Quality C-l
-------
Lists of Figures
Figure 1. Well-Mixed Zone Model Diagram 4
Figure 2. Test Building HVAC Region Diagram 9
Figure 3. Alterations to the Upper Floor (3rd Floor) of the Test Volume 10
Figure 4. Alterations to the Lower Floor (2nd Floor) of the Test Volume 11
Figure 5. Photograph of the Paniculate Eductor Release Mechanism 12
Figure 6. Particle Size Distribution of Visolite® Aerosolized with an Eductor Release Mechanism
Operating at a Gas Flow Rate of 100 1pm as Measured by an Aerosizer® 13
Figure 7. Photograph of MetOne Handheld Optical Particle Counter 13
Figure 8. Typical Paniculate Data Gathered by a MetOne During Testing 14
Figure 9. Photographs of Flow Measurement Using a.) a Balometer and b.) an Anemometer 15
Figure 10. Sampling Scheme for "Large" Notional Building 24
Figure 11. Sampling Scheme for "Small" Notional Building 25
Figure 12. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with
Moderate Recirculation, Low Leakage, Moderate-Low Filtration, Standard Makeup Air, and
Standard Infiltration 27
Figure 13. Demonstration of Repeatability by Comparison of Experimental and Model-Predicted Data
for "Large" Notional Building with Moderate Recirculation, Low Leakage, Moderate-Low Filtration,
Standard Makeup Air, and Standard Infiltration 28
Figure 14. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building
with Moderate Recirculation, Low Leakage, Moderate Filtration, Standard Makeup Air,
and Standard Infiltration 28
Figure 15. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building
with Moderate Recirculation, Low Leakage, Low Filtration, Standard Makeup Air,
and Standard Infiltration 29
Figure 16. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building
with Moderate Recirculation, Low Leakage, High Filtration, Standard Makeup Air,
and Standard Infiltration 29
Figure 17. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building
with Moderate Recirculation, High Leakage, Moderate Filtration, Standard Makeup Air,
and Standard Infiltration 30
Figure 18. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building
with Moderate Recirculation, High Leakage, Low Filtration, Standard Makeup Air,
and Standard Infiltration 30
Figure 19. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building
with Moderate Recirculation, High Leakage, High Filtration, Standard Makeup Air,
and Standard Infiltration 31
Figure 20. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building
with Moderate Recirculation, Low Leakage, Moderate Filtration, Standard Makeup Air,
and Standard Infiltration 31
Figure 21. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building
with Moderate Recirculation, High Leakage, Moderate Filtration, Standard Makeup Air,
and Standard Infiltration 32
Figure 22. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with Moderate
Recirculation, Low Leakage, High Filtration, Standard Makeup Air, and Standard Infiltration 33
Figure 23. Comparison of Experimental and Model-Predicted Data for " Small" Notional Building with Moderate
Recirculation, High Leakage, High Filtration, Standard Makeup Air, and Standard Infiltration 33
Figure 24. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with Moderate
Recirculation, High Leakage, Low Filtration, Standard Makeup Air, and Standard Infiltration 34
-------
Lists of Figures (Continued)
Figure 25. Comparison of Experimental and Model-Predicted Data for " Small" Notional Building with Moderate
Recirculation, Low Leakage, Low Filtration, Standard Makeup Air, and Standard Infiltration 34
Figure 26. Photographs of Various Door Positions for the "Large" Notional Building (i.e., the door between
B113 andB114) and Recirculation Rates (i.e., low or moderate) with Corresponding Interzonal
Leakage Rates 35
Figure 27. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with
Low Recirculation, Very Low Leakage, Moderate Filtration, Standard Makeup Air, and
Standard Infiltration 36
Figure 28. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Low
Recirculation, Low Leakage, Moderate Filtration, Standard Makeup Air, and Standard Infiltration 36
Figure 29. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Low
Recirculation, High Leakage, Moderate Filtration, Standard Makeup Air, and Standard Infiltration 37
Figure 30. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Low
Recirculation, Low Leakage, Low Filtration, Standard Makeup Air, and Standard Infiltration 37
Figure 31. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Low
Recirculation, Low Leakage, High Filtration, Standard Makeup Air, and Standard Infiltration 38
Figure 32. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with
Low Recirculation, Moderate Leakage, Moderate Filtration, Standard Makeup Air,
and Standard Infiltration 39
Figure 3 3. Comparison of Experimental and Model-Predicted Data for " Small" Notional Building with Low
Recirculation, Moderate Leakage, Low Filtration, Standard Makeup Air, and Standard Infiltration 39
Figure 34. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with Low
Recirculation, Moderate Leakage, High Filtration, Standard Makeup Air, and Standard Infiltration 40
Figure 35. Illustration of Exposure-Based Metric Inadequacies for a Hypothetical Case of Various Filtration
Efficiencies, Equal Zone Volumes, Makeup Air, Moderate Recirculation, Infiltration,
and High Leakage 44
Figure 36. Normalized Exposure at 30 Minutes Versus Model Input Parameters for a "Large"
Notional Building 45
Figure 37. Normalized Time to Critical Exposure Versus Model Input Parameters for a "Large"
Notional Building 45
Figure 38. Normalized Exposure at 30 MinutesVersus Model Input Parameters for a "Large" Notional Building ...48
Figure 39. Impact Analysis Results for Various Zone 3 Volumes Under 1ACH Makeup Air, 5 ACH Recirculation,
0.3 ACH Infiltration, 30% Filtration, and 1 ACH Interzonal Leakage 50
Figure 40. "Large" Building (14,600 m3) Parameter Mapping Results for a Recirculation Rate of 7 ACH 53
Figure 41. "Large" Building (14,600 m3) Parameter Mapping Results for a Recirculation Rate of 5 ACH 53
Figure 42. "Large" Building (14,600 m3) Parameter Mapping Results for a Recirculation Rate of 3 ACH 53
Figure 43. "Small" Building (1,000 m3) Parameter Mapping Results for a Recirculation Rate of 7 ACH 57
Figure 44. "Small" Building (1,000 m3) Parameter Mapping Results for a Recirculation Rate of 5 ACH 57
Figure 45. "Small" Building (1,000 m3) Parameter Mapping Results for a Recirculation Rate of 3 ACH 57
Figure 46. Parameter Space Map of the Dominant Parameter, or Parameter with the Highest Impact Score,
for a "Small" Building (1,000m3) 58
Figure 47. Photograph of In-Room Air Cleaner Used During Testing (Whirlpool Model AP4503HO) 63
Figure 48. Comparison of Experimental Data for an In-Room Air Cleaner for a "Large" Notional Building
with Moderate Recirculation, Low Leakage, Moderate Filtration, Standard Makeup Air,
and Standard Infiltration 64
-------
List of Tables
Table 1. Moderate Recirculation Test Matrices 7
Table 2. Low Recirculation Test Matrices 8
Table 3. Lower Floor Airflow Measurements Under Moderate Recirculation 16
Table 4. Upper Floor Airflow Measurements Under Moderate Recirculation 17
Table 5. Moderate Recirculation Airflow Summary 17
Table 6. Lower Floor Airflow Measurements Under Low Recirculation 18
Table 7. Upper Floor Airflow Measurements Under Low Recirculation 19
Table 8. Low Recirculation Airflow Summary 19
Table 9. Moderate Recirculation Test Matrix Results Key 22
Table 10. Low Recirculation Test Matrix Results Key 23
Table 11. Comparison of Experimental and Predicted Performance Metrics for the Zone of Interest for the
"Moderate" Recirculation Condition 42
Table 12. Comparison of Experimental and Predicted Performance Metrics for the Zone of Interest for the
"Low" Recirculation Condition 42
Table 13. Typical Change in Parameters for Use in Calculating Scale Factors 46
Table 14. Normalized Impact Scores for the Base Case for a "Large" Notional Building 47
Table 15. Impact Analysis Results for Various Zone 3 Volumes Under 1ACH Makeup Air, 5 ACH Recirculation,
0.3 ACH Infiltration, 30% Filtration, and 1 ACH Interzonal Leakage 49
Table 16. Typical Parameter Values and Model Parameter Ranges 49
Table 17. Impact Analysis Results for the "Large" Building (14,600 m3) Parameter Space Map 52
Table 18. Impact Analysis Results for the "Small" Building (1,000 m3) Parameter Space Map 56
Table 19. Comparison of Experimental Performance Metrics for the Zone of Interest for the "Moderate"
Recirculation Condition With and Without an In-Room Air Cleaner 64
-------
Glossary
ACH
Filtration or Filtration Efficiency
Infiltration
Leakage
Makeup Air
Recirculation
Zone 1
Zone 2
Zone 3
Air change rate expressed as airflow in units of volume per hour divided by the zone
volume in identical volume units (ASHRAE, 1991).
The nominal efficiency with which a filter removes particles from an air stream.
Uncontrolled inward leakage of outdoor air into a building through cracks and interstices
in any building element. Normally caused by pressure effects or differences in air density
(ASHRAE, 1991). Also known as building air infiltration.
Uncontrolled and unplanned flow between two zones of a building. Normally caused
by slight pressure differences between the zones. Also know as interzonal leakage or
room leakage.
Air brought into a building from outside to replace that exhausted. Air intentionally
brought into a building that was not previously circulated through the building
(ASHRAE, 1991). Also known as outdoor air.
Air taken from a zone and returned to the zone after being passed through conditioning
elements (ASHRAE, 1991). Also known as recirculated air.
Zone of release
Zone of interest
The lumped zone, which represents the bulk of the building, or the rest of the building.
-------
-------
Executive Summary
Although it is intuitive that building and heating, ventilation,
and air conditioning (HVAC) system parameters play a major
role in determining how an agent disperses after an indoor
biological attack, how and to what extent those parameters affect
the transport of contaminants is not well understood. A better
understanding of how building and HVAC parameters affect
the transport of an agent will improve the ability to mitigate
the effects of a biological attack and aid in determining where
resources in building protection are best directed. Therefore, the
objective of this project was to theoretically and experimentally
evaluate the effects that modifications to HVAC and building
design and operation would have on the spread of biological
agents (in aerosol form) within a building.
To this end, a three-zone model representation of a building
was developed to determine which HVAC and building
parameters are most important, and how accurately they
need to be known, in determining the impact of an indoor
bio-agent attack. The three-zone model consists of a zone
of release and zone of interest, which are adjacent and of
equal size, as well as a "lumped" zone, which represents
the remainder of the building. All three zones are serviced
by a single HVAC system with common recirculation.
The potential impact of an attack was quantified in terms of
two performance metrics: the cumulative exposure to the agent
after 30 minutes and the time to reach a critical exposure level
(referred to as Ct) for that agent. The "critical exposure level"
was arbitrarily selected. Other values of critical exposure may
have an effect on the outcomes estimated in this report. A
sensitivity analysis method was also developed for determining
the required level of accuracy for each building and HVAC
operating parameter, i.e., to determine how well each parameter
must be known to accurately estimate the two performance
metrics. The analysis method allows for investigation of the
relative impact of each parameter, i.e., the change in the two
performance metrics as a result of a change in a parameter.
Field tests were then performed to gather data for the validation
of both the three-zone modeling concept and the sensitivity
analysis method for determining parameter impact. To cover the
broad range of parameters and conditions needed for testing, most
of the tests were performed without duplicates. The field tests
consisted of a series of tracer experiments under varying HVAC
conditions in a test building. Visolite®, a fluorescent-tagged,
calcium carbonate aerosol, was chosen as the tracer material. The
aerosol was released using a custom-built eductor. The venue
for these tests was a three-story building, located in Anniston,
Alabama, that contained three separate air handling units
(AHUs), each possessing its own supply and return ductwork
and each servicing a different region of the building. After a few
initial building alterations, the test venue was able to represent
both a large and a small notional building, depending on which
zones were chosen as the zone of release and zone of interest.
An initial analysis using model simulations revealed that makeup
air and infiltration rates had little impact during an internal release
scenario and system parameters were of greater importance
than single-zone parameters (i.e., the recirculation rate of the
entire building was more important than the recirculation rate
of the zone of interest). The initial analysis work also identified
building size, interzonal leakage rate, recirculation rate, and
filter efficiency as the key parameters affecting the selected
performance metrics. The simulation results showed that as the
building size increased, the filtration efficiency went from being
a potentially dominant factor to a lesser factor compared to the
interzonal leakage rate. The system recirculation was found
to be of secondary importance but had a strong effect on the
importance of the filtration efficiency for smaller buildings (i.e.,
for building volumes less than five times the zone of interest).
On the whole, the excellent agreement observed between
experimental and model-predicted data validated the use of
the three-zone model to approximate contaminant spread in
a building. Excellent agreement between the lumped third
zone in the model and multiple rooms throughout the test
building validated the use of a three-zone model to predict
contaminant levels in real buildings with more than three
zones. Modeling, as supported by the experimental data,
provides a useful tool for estimating and assessing which
parameters significantly affect the spread of contaminant in a
building. Buildings of varying size and HVAC performance
can be assessed. It was demonstrated that changes in HVAC
filter efficiency, air exchange, and building "size" could be
varied with a predictable impact on contaminant spread.
Ancillary experiments suggest that the use of an in-room
air cleaner can greatly reduce the paniculate matter level in
the room. The magnitude of this reduction will vary greatly
depending on the volume of the room, as well as the throughput
and efficiency of the in-room air cleaner.
From this project it is clear that there is no one universal answer
regarding which building or HVAC parameter is most important
or how accurately it needs to be known. It depends on the release
location, HVAC characteristics, building characteristics, and
proximity of the zone of interest to the release location. This
implies that the usefulness of mitigation strategies to protect
buildings (or more precisely its occupants) must be considered on
a case-by-case basis. The sensitivity analysis method discussed
herein provides some indication of the general trends and
identifies the most important parameters impacting contaminant
spread for various combinations of HVAC parameters and
building volumes relative to the zone of interest. The modeling
approach developed here could be used to assess various
scenarios and buildings of specific interest, without the need for
extensive knowledge of HVAC and building parameters. Using
the modeling tool developed for initial analysis may be useful in
determining the merits of modifying the building to enhance the
protection of occupants or to mitigate the spread of contaminants.
-------
Recommendations for future study focus on two aspects of the
well-mixed model: usability and applicability. The usability
of the model could be improved by developing a user-friendly
graphical user interface (GUI), which would allow casual users
(e.g., building operators) to rapidly perform simple impact
analyses for a building of interest given limited building
information (e.g., HVAC settings, building and room volumes).
The development of a GUI would increase the utility and impact
of this effort by making the model available to more people. Also
recommended is an enhancement to expand the applicability of
the tool by developing and verifying an analogous model for a
building with a more complex HVAC ductwork scheme. The
present model is applicable only to a building with one common
return ductwork system. Developing an analogous model that
effectively represents buildings with multiple return ductwork
systems (i.e., multiple air handling units) and performing an
experimental verification, similar to this work, would aid in
making a model more applicable to large buildings with more
complex air handling schemes.
-------
1.0
Introduction
In general, the spread of a contaminant following the release
of a biological agent within a building is affected by the
building characteristics (e.g., building volume, air infiltration
rates, or room-to-room leakage) and parameters of the HVAC
system. However, a quantitative understanding of the relative
impacts that each of the building and HVAC parameters has
on contaminant dispersion is needed. A better understanding
of these parameters' impacts will improve the ability to
protect buildings from a biological attack. Therefore, a
project was undertaken in which theoretical and experimental
analyses were conducted to determine the impact that HVAC
and building modifications have on the spread of paniculate
agents within a building. The results of the theoretical
analyses would guide the design of experiments, and the
experiments would help to validate the model.
A building test bed inAnniston, Alabama, was selected
for the experimental program, as it allowed for access
and control of most of the parameters of interest.
The size of the building was also shown to influence
which controllable building or HVAC parameter was
important regarding the control of contaminant spread.
Although the size of the building is not a controllable
parameter in actual applications, the test bed facility did
allow manipulation of room sizes so that "large" and
"small" building configurations could be assessed.
2.0
Objective
The objective of this project was to evaluate the impact that
modifications to HVAC and building design and operation
have on the spread of biological (aerosolized) agents within
a building. In the course of conducting this project—
specifically the theoretical analysis of important parameters
for determining the impact of an attack on buildings —it
became apparent that another objective would also be to
experimentally demonstrate the application of the three-zone,
well-mixed model.
3.0
Scope
The approach and development of a three-zone, well-mixed
model is described. The experimental methods used during
this study are detailed. A series of experiments at a test bed
were used to experimentally confirm the usefulness of a
three-zone, well-mixed volume model in simulating a "real"
building. The results are then used to analyze what parameter
changes had the largest impact on potential building
performance, and conclusions that can be made from this
work are stated.
-------
-------
4.0
Model Development
4.1 General Approach
The modeling approach taken was to develop an algorithm
to estimate contaminant transport in a notional building as
a function of time. The model allowed building parameters
(e.g., HVAC airflow, filter efficiency) to be investigated. The
model was used to generate time-dependent contaminant
concentration profiles in zones of interest, which were then
used as performance metrics for estimating exposures. Model
parameters were then varied over a range of conditions so
that the changes in performance metrics could be assessed.
These changes in performance metrics were then used to
determine the most important parameters and how well
they need to be known. (In a separate but related project, a
more mechanistic model of contaminant transport based on
aerosol dynamics is being developed. That model considers
contaminant concentrations in zones of a notional building
and how HVAC flow, air exchange rates, infiltration rates,
and other parameters impact those concentrations. That
model has been leveraged for consistency with this simplified
approach as briefly discussed below. The simplified model
approach, input parameters of interest, performance metrics,
and data interpretation/analysis process are discussed in
Sections 4.2 through 4.5 with a few illustrative examples.)
The model approach used for this analysis of building and HVAC
parameters and their effect on hazard exposure is purposely
general to allow consideration for a range of building types and
sizes. The analysis is based on relative changes in exposure in a
zone of interest with changes in building or HVAC parameters.
The absolute determination of whether a change in HVAC or
building parameter is important (whether a resulting exposure
was reduced below a lethal concentration, for example) is not
determined by this analysis. Absolute determination of those
effects is dependent on scenario attack parameters such as agent
release, building occupant exposure durations, and agent type.
So, for example, in some scenarios it may have significantly
different impacts on occupants if a 90% efficient filter is used and
in other scenarios it may not make a difference. That is not to say
that the model presented here is not valid for estimating resulting
aerosol and other contaminant concentrations as a function
of time and zone for specific release and building operating
scenarios, just that the approach presented here for determining
the effects of various parameters is not an absolute approach. The
fact that the model developed here is generalized and based on
relative change in exposure does not diminish its usefulness. The
model allows for quick assessment and sensitivity analysis, with
little specific building information. Thus, it is very applicable for
building engineers and planners.
The HVAC Aerosol Dynamics Model, HVACADM, developed
under a separate project, was leveraged here for investigating the
effects of different contaminant behavior mechanisms. (Although
the primary focus of the modeling and experimental program is
for agents in paniculate form, the approach could also be adapted
for chemical agents as well.) To leverage the HVACADM model
to address the objectives of this task, a simplified model that
focuses on mass balances around three zones was developed. This
simplified model allows the parameters affecting contaminant
transport to be assessed without the use of mechanistic models.
The goal of the project is two-fold: (1) determine which
parameters are most important for predicting the impact of a CB
attack on a building and (2) determine the degree of accuracy to
which these parameters needs to be known for practical purposes
(i.e., at what point the level of exposure becomes insensitive to
a change in the parameters controlling the contaminant spread
and exposure).
The nomenclature used to define building and HVAC parameters
in this report follows as closely as possible that used by the
American Society of Heating, Refrigerating and Air-Conditioning
Engineers (ASHRAE). In this fashion, consistent use of terms is
intended to aid with an understanding of the model approach and
results. Definitions of the key terms are provided in the Glossary
section of this report.
4.2 Simplified Model Development
The simplified model is based on the well-mixed zone
methodology, which implies instantaneous, perfect mixing within
a zone and thus a uniform concentration throughout the zone
volume. A mass balance was applied to each well-mixed zone to
derive the governing equations for the model (see Equation 1 and
Figure 1).
[Accumulation] = [Input] - [Output] + [Generation] (1)
The accumulation term accounts for changes in the concentration
of agent within the zone volume, the input term accounts for
agent that is added to the zone via the input streams (makeup
air, recirculation, infiltration, and interzonal leakage entering the
zone), the generation term accounts for a release within the zone,
and the output term accounts for agent removed from the zone via
the output streams (exhaust, recirculation, and interzonal leakage
leaving the zone). In this model, a zone could be an individual
room or a group of rooms served by a common HVAC system.
-------
IniiliMituni
Figure 1. Well-Mixed Zone Model Diagram
The mass balance can be expressed in terms of particle
concentration by dividing both sides of the equation by the
zone volume; explicitly writing out the terms for a balance
around Zone 1 yields Equation 2.
CM.
(2)
Where
is the concentration of contaminant in both the
well-mixed Zone 1 volume and the streams exiting Zone
1 (an unknown function of time),
/ is the time,
fffliter is the fractional efficiency of the paniculate filter
used in the HVAC system for general makeup air and
recirculation,
rjj is the fractional efficiency of an additional paniculate
filter used for the makeup air and recirculation of Zone 1,
Q,-j is the makeup airflow rate of fresh air into Zone 1,
V i is the volume of Zone 1,
Cvent(t) is the concentration of the contaminant in the
makeup air stream prior to filtration,
Qirfi is the infiltration flow rate into Zone 1,
Cinj(t) is the concentration of the contaminant in the
infiltration stream,
Q2] is the interzonal leakage from Zone 2 to Zone 1,
C2(t) is the concentration of the contaminant in both the
well-mixed Zone 2 and the streams exiting Zone 2 (an
unknown function of time),
Q31 is the interzonal leakage from Zone 3 to Zone 1,
C3(t) is the concentration of the contaminant in both the
well-mixed Zone 3 and the streams exiting Zone 3 (an
unknown function of time),
QR1 is the recirculation flow rate for Zone 1,
Q12 is the interzonal leakage from Zone 1 to Zone 2,
Q13 is the interzonal leakage from Zone 1 to Zone 3, and
G, is the generation rate of agent per unit time in Zone 1.
Next, Euler's Method, a numerical solution approach, is
applied by converting the derivative to a simple difference
(dt -> At, dC, -> AC, = Clt+At-C1:t), and solving for Clt+At
yields Equation 3.
C,,iA, =(
-
vl vl vl
(3)
Where
Ci,t+At is the concentration in Zone 1 at the next time
step (t+At),
Cu is the concentration in Zone 1 at the current time
step,
C2>t is the concentration in Zone 2 at the current time
step,
C5tt is the concentration in Zone 3 at the current time
step, and
At is the size of the time increment.
-------
It is important to note that this is a simplified numeric
solution and thus requires a small time increment
(< 0. 10 min). Similar equations are then written for the
other zones.
The concentration of the contaminant in the H VAC stream
prior to filtration (Cvent) can easily be calculated as a
weighted average of its constituents (see Equation 4).
+ a,.,
•1+fil'2+filJ(4)
While this effort will not specifically consider the effects
of local exhaust, the effects of local exhaust could be
incorporated into the model through the adjustments of Cit,
which denotes the concentration in the makeup air (i.e., the
HVAC air intake). Using the model, it would be possible
to study local exhaust effects by quantifying the results of
varying the makeup air concentration as a function of the
exhaust concentration.
Once Euler's Method has been used to obtain the
concentration within each zone as a function of time, the
cumulative exposure within each zone can be calculated by
numerical integration; e.g., for Zone 1 (see Equation 5):
^1=1^(0* (5)
Where E; denotes the cumulative exposure in Zone 1.
The mass balance approach described above allows for
time- and location-dependent concentrations to be predicted
for each zone of the building. Those time-dependent
concentration estimates are then used in Section 7.0, when
the model results are compared to the measured aerosol
concentration for the various release scenarios, building
configurations, and HVAC operating conditions.
-------
-------
5.0
Experimental Methods
5.1 Test Design
Introductory note: A Quality Assurance Project Plan (QAPP)
was developed for the experimental program discussed in
this chapter and was approved by the EPA project lead. The
QAPP was not approved by the EPA QA officer; however,
this final report does address the key QA concerns of the
QA manager.
The test design for this experimental work was guided by the
effort in developing the simple three-zone, well-mixed model
described in Section 4.0. This effort identified the filtration
efficiency, leakage, system recirculation, and building
size relative to the room of release as the key parameters
in assessing the impact of an indoor CB attack. Other
parameters (e.g., the infiltration and makeup air) were found
to have a lesser impact on the resulting concentrations and
exposures in comparison to the aforementioned parameters
of filtration efficiency, leakage, and system recirculation.
This initial modeling analysis was used as a basis to design
the experimental field study. Filtration efficiencies ranging
from no filtration (i.e., the natural loss of aerosol in ductwork
and other AHU components) to a high-efficiency filter (i.e.,
a 95% dioctyl phthalate [DOP] rated filter) were selected
to cover a range of possible filtration efficiencies. Natural
losses of aerosol were treated as a filtration efficiency to
represent the lower possible range of filtration efficiencies.
Recirculation rates of 3, 5, and 7 air changes per hour
(ACH) were initially planned to cover the range of likely
system recirculation rates; however, due to limitations of the
HVAC system, it was not possible to achieve recirculation
rates above 6 ACH. Therefore, it was decided to encompass
only the low (3 ACH) and moderate (5 ACH) recirculation
conditions. Two leakage conditions were chosen as goals for
the study: low leakage (<0.4 ACH) and high leakage (>0.6
ACH). These selections were made due to the desire for a
natural leakage path as would occur in any common building
(as opposed to a forced leakage with a small fan) and the
resulting lack of control over the leakage rates. The high and
low leakage values were chosen to approximate "loose" and
"tight" building construction, respectively.
A tiered test matrix was initially designed to provide a
planned approach of study for numerous possible test
outcomes and contingencies. This matrix provided adequate
coverage of the range of parameters of interest to demonstrate
that the model findings were accurate with respect to
estimating the effects of building/HVAC parameters and
contaminant concentration. Rapid analysis and reduction
of results were then used to decide which tests in the tiered
matrix would be performed. The final test matrices are shown
in Tables 1 and 2 for moderate and low recirculation.
Table 1. Moderate Recirculation Test Matrices
"Moderate" (5 ACH)
8,
3
CD
CO
X
><
O IT)
—1 I--
i — i
C)
X
- o
J= <
.9?^
X C\J
= o
1 — 1
X
. o
O IT)
-1 C\J
- i — i
O
X
- o
J= <
.5?^
x i-v.
= i — i
i — i
"Low" (10%)
"Moderate-Low" (25%)
"Moderate-Low" (25%)
"Moderate" (50%)
"High" (90%)
"Low" (10%)
"Moderate" (50%)
"High" (90%)
"Low" (10%)
"Moderate" (50%)
"High" (90%)
"Low" (10%)
"Moderate" (50%)
"High" (90%)
"Large" notional building/ "Low" leakage denotes a closed door position
"Large" notional building / "High" leakage denotes a door ajar position
"Small" notional building / "Low" leakage denotes a door open 20 cm
"Small" notional building / "High" leakage denotes a door fully open
-------
Table 2. Low Recirculation Test Matrices
X
o
<
ro
1
8,
.3
03
GO
>§
o<
>, °
C1 LO
.0) 0
= o
- o
5 0 X
0 CXJ £
^ 0 <
X
- o
J= <
.5P0
X LO
= 01
o
"Moderate"
(0.725ACH)
"Moderate" (10%)
"Low" (10%)
"Moderate" (50%)
"High" (90%)
"Moderate" (50%)
"Low" (10%)
"Moderate" (50%)
"High" (90%)
"Large" notional building/ "Very Low" leakage denotes a closed door position
"Large" notional building / "Low" leakage denotes a door partially ajar
"Large" notional building / "High" leakage denotes a door open about 1 cm
"Small" notional building / "Moderate" leakage denotes a door open 70 cm
5.2 Test Building Description
The venue for these tests was a three-story building, located
in Anniston, Alabama. The building's HVAC system
consisted of three separate air handling units (AHUs), each
servicing a specific region of the building (see Figure 2).
The 1st floor of the building did not possess an active HVAC
system (shaded in white in Figure 2). Air handler 1 (AHU1)
serviced approximately 2/3 of the 3rd floor of the building
(shaded in blue in Figure 2). Air handler 2 (AHU2) serviced
approximately 2/3 of the 2nd floor of the building (shaded
in red in Figure 2). Air handler 3 (AHU3) serviced a multi-
floor region of the building consisting of approximately 1/3
of the 2nd and 3rd floors (shaded in green in Figure 2). Each
of the AHUs possessed its own supply and return ductwork.
Since the simple well-mixed model considers only a simple
HVAC system with a common supply and return ductwork, a
subsection of the test building, defined as the HVAC region
serviced by AHU3, was selected for subsequent study. The
HVAC region serviced by AHU3 was of particular interest
for this study because it spanned multiple floors and would
require the least number of building modifications to provide
a suitable test venue.
Each supply vent and return duct was equipped with a simple,
manually operated damper that controlled the fractioning
of supply and return flows using "path of least resistance"
principles. While this simple construction prevented the
creation of significant pressure differences between zones, it
is believed to be representative of typical HVAC design. The
fraction of total airflow recirculated (and thus the fraction
of fresh air) was adjusted via variable flow controllers and a
central computer control system. In addition, the total airflow
delivered by the air handling unit was also controlled via the
computer control system.
-------
3rd Floor
2nd Floor
1st Floor
Figure 2. Test Building HVAC Region Diagram
Legend
AHU1
AHU2
AHU3
NoAHU
Some initial building alterations were required prior to testing
to allow for the desired test conditions. These alterations
involved the extension of existing walls to create independent
zones, the addition of new walls to create independent zones,
and the creation of a number of return inlets within the newly
created zones. The alterations made to the building were
performed by professional contractors and are described in
detail below, as well as illustrated in Figures 3 and 4.
The test bed contains three independent air handling units
that service three independent HVAC regions. Throughout
this document the terms "test volume" and "building" will
be used interchangeably to refer to the HVAC region that
constitutes the easternmost portion of the top two floors of
the building (i.e., the portion shaded in green in Figure 2).
The building construction is a standard commercial design
in which a limited number of walls extend beyond the drop
ceiling to the slab. For the purposes of modeling the building,
it is important to note that only walls that extend to the slab
define a zone. Prior to alterations, the HVAC region contained
only four independent zones of approximately the same size
(i.e., two on each floor). As discussed previously, this effort
required both large and small notional building scenarios.
To achieve this, walls were either added or extended to
define new zones. Again, it is important to note that "large"
and "small" refer to the size of the building relative to
the zone of interest. For example, a 3,000 m3 building
composed of 100 rooms that are each 30 m3 in volume is
equivalent to a 50,000 m3 building composed of 100 rooms
that are each 500 m3 in volume. The building size relative
to the zone of interest for both of these examples would be
100, making each of them a "large" notional building.
On the upper floor (3rd floor), a portion of an existing wall
was extended to the slab and a new slab-to-slab wall was
constructed to split Room B209 into two independent
zones. The newly created zones were approximately equal
in volume with each representing approximately one-tenth
(1/10) of the test volume. These alterations created the
zone of interest and zone of release for the "small" notional
building and are illustrated in Figure 3. In addition, a door
was incorporated into the new wall to provide a variable leak
path between the zone of release and zone of interest.
On the lower floor (2nd floor), the existing walls that defined
a series of small rooms were extended to the slab to create
four independent zones. Each of the newly created zones was
approximately one-seventy-fifth (1/75) of the test volume.
Two of the zones, B113 and B114, were used as the zone of
interest and the zone of release, respectively, for the "large"
notional building. These alterations are illustrated in Figure 4.
Doors were also added to the walls between each of the
newly created zones to provide a variable leak path between
the zones.
While each of these newly created zones already possessed
a makeup air inlet (i.e., a fresh air supply vent), most lacked
a return inlet (i.e., a recirculated air intake). For this reason,
return inlets were added to each of the newly created zones
(see Figures 3 and 4). Each return inlet was equipped with
a damper to adjust the recirculated airflow taken from each
zone. These building alterations produced an experimental
test volume suitable for the planned experiments. The
test volume is considered to represent exceptionally tight
construction. All doors possessed high quality seals and
sweeps. All walls were well caulked and sealed. For this
reason, the leakage observed under various door positions
during this study should not necessarily be taken as standard
leakage rates for other buildings.
-------
•d
«
Ml
O
CD
X
s
130
_g
•*->
01
'*
W
o
Q
-3
oo
o
+->
f>
d
CD
o
&
o
Q
"
CD rc
CD
-------
•d
«
DC
«
hJ
w
bo
W
O
O
Q
W)
U
8
Q
I
I
-------
5.3 Release Methods
Visolite®, a fluorescent-tagged, calcium carbonate (chalk)
dust, was used as the innocuous aerosol tracer. A paniculate
aerosol was chosen as the tracer for this effort because of
the advantages of existing filtration systems and real-time
measurement systems, such as optical particle counters. It
is important to note that while a paniculate aerosol was
used for this experimental effort, the model and analysis
presented here is valid for any contaminant that does
not appreciably deposit or adsorb on surfaces during the
timeframe of interest.
The amount of Visolite® to be released was estimated based
on the results of previous indoor air quality field studies
performed by Battelle. These previous results, combined with
past experience, suggested that a release of 1 to 20 grams
of Visolite® would be sufficient to produce the desired test
aerosol. The main elements of the release mechanism are
an air supply, an eductor, a grooved turntable, and a rotary
motor (see Figure 5). Air flowing through the eductor creates
a suction that pulls the Visolite® powder from the grooved
turntable. Provided the powder is evenly distributed within
the groove of the turntable, the speed at which the motor
rotates the turntable defines the release rate of the aerosol.
In previous work this eductor type mechanism has proven
its effectiveness in dispersing similar quantities of a powder
aerosol. The particle size distribution of Visolite® aerosolized
via this eductor mechanism is shown in Figure 6. Since it
is possible that the solid aerosol dissemination will produce
nuisance dust levels that are temporarily above suggested
safety limits, the release will be triggered by an electronic
timer to avoid any potential exposure of test personnel.
Figure 5. Photograph of the Particulate Eductor Release Mechanism
-------
99.9-
99-
90-
70-
50-
30
0.1 -
0.01 -
n nn-i -
A Cumulative Number %
A Cumulative Mass %
I
A
A
j
A
A
A
A A
A
A
A
A
I A
A
A
A
A A
A A
A ^
A A
A
A
A
L . . ......
0.5
0.7
2 34
Particle Size {urn)
5 6 7 8 9 10
Figure 6. Particle Size Distribution of Visolite® Aerosolized with an Eductor Release
Mechanism Operating at a Gas Flow Rate of 100 Ipm as Measured by an Aerosizer®
5.4 Sampling Methods
During the testing period, five handheld particle counters
(MetOne model HHPC-6, manufactured by Hach Ultra
Analytics, see Figure 7) were used to monitor aerosol levels
at selected locations within the area of interest. The MetOne
particle counter uses optical particle counting techniques
(i.e., measuring the scattered light as particles pass through
a laser) to provide data on six particle size ranges from 0.3
to 20 urn (0.5 to 0.7 urn, 0.7 to 1.0 urn, 1.0 to 2.0 urn, 2.0
to 5.0 urn, 5.0 to 10.0 urn, and 10.0 to 20 urn). Since the
Visolite® particles are being used only as a tracer or indicator,
the particle size distributions were not of interest, only
particle concentration. For this effort, particle concentrations
from the MetOne channel ranging from 2.0 to 5.0 um were
used as the indicator of the Visolite® concentration. Figure
8 shows a typical example of MetOne data gathered in the
zone of interest during testing. Typical background levels
can be observed in the time period prior to release (i.e., time
less than zero). The Data Quality Objectives (DQOs) and an
electronic archive of the data gathered during this work can
be found in Appendices C and D, respectively.
The MetOne particle counter is lightweight and automated,
making it easy to deploy/recover, which facilitates rapid
turnaround. Pretest procedures involve turning the instrument
on and pressing the start button. MetOne particle counters
were factory calibrated prior to use in this study. In addition,
the outputs from the multiple MetOne units used in this
study were compared against each other to ensure sizing
consistency throughout the study. When not in use, MetOne
particle counters were run continuously on HEPA-filtered air
to flush and clean the devices.
Figure 7. Photograph of MetOne Handheld
Optical Particle Counter
-------
100,000
10JOOO
I
I
ro
Q.
1.000
100
10
0.5 to 0.7 m icron
0.7 to 1.0 micron
1.0 to 2.0 micron
2.0 to 5.0 micron
5.0 to 10,0 micron
10.0 to 20.0 micron
1
-10 0 10 20 30
Time [min]
Figure 8. Typical Particulate Data Gathered by a MetOne During Testing
40
50
The apparent discrepancy between the particle size
distribution in comparison to Figure 6 is attributed to a
difference in the operating principle of the particle sizer
(optical versus time of flight) and a potential error in the
sizing of a fluorescent particle using optical means. This
does not significantly impact this work since the 2.0 to
5.0 um channel is used as a tracer (i.e., the exact size range
of particles measured by a given channel is not critical to this
study as long as it is constant throughout experimentation).
-------
6.0
Preliminary Efforts
Prior to conducting the experimental study, a number of
preliminary efforts were required to ensure relevant, usable
data. These preliminary efforts consisted of initial building
alterations, airflow measurements, and initial leakage tests.
Since the main goal of the experimental work was
comparison with a model, the actual values of HVAC
parameters were of critical importance. To this end, a brief
series of HVAC parameter quantification tests was performed
to experimentally determine, or in some cases verify, the
actual values of key HVAC parameters. These tests can
be divided into airflow measurements and leakage tests.
Airflow measurements consisted of measuring the volumetric
flow rate or face velocity at all ductwork throughout the
test bed. Leakage tests were initially intended to provide a
direct measure of the interzonal leakage under operational
conditions. Unfortunately, due to leakage across a return
damper, the intended method was no longer viable and a
combinatorial approach was adopted.
6.1 Airflow Measurements
The first portion of the preliminary characterization
tests consisted of measuring the airflow using various
anemometers, adjusting the blower and Phoenix valve
settings that control the HVAC flows, and subsequently
remeasuring the airflows. A balometer (ALNOR model
number APM 150) and an anemometer (Davis model
number LCA30 VT) were used to measure flow (see
Figure 9 for photograph). These methods were sufficient
for experimentally measuring the system recirculation and
makeup air rates for each zone.
The low, moderate, and high recirculation rates initially
planned for this study were intended to cover a range of
recirculation rates from 3 to 7 ACH. However, due to
limitations of the HVAC system, it was not possible to
achieve recirculation rates greater than 6 ACH. Therefore, it
was decided to reduce the scope of the study to encompass
only the low and moderate recirculation conditions. In
addition, the use of four mixing fans each in both the zone of
interest and zone of release of the "small" notional building
were agreed to enhance mixing. Tables 3 and 4 contain the
results for moderate recirculation flow measurements made
on the lower and upper floors, respectively, while Table 5
contains summary information in the form of air exchange
rates for the moderate recirculation condition (nominally 5
ACH). Similarly, Tables 6 and 7 contain the results for low
recirculation flow measurements made on the lower and
upper floors, respectively, while Table 8 contains summary
information in the form of air exchange rates for the low
recirculation condition (nominally 3 ACH).
a.)
b.)
Figure 9. Photographs of Flow Measurement Using a.) a Balometer and b.) an Anemometer
-------
Table 3. Lower Floor Airflow Measurements Under Moderate Recirculation
Zone ID Room Number Room Volume [ft3] Vent ID Type Measured Flow [cfm]
LI
L2
L3
L4
L5
L6
B113
B114
B115
B116
B110E
B110D
B110
Bill
B112
B110B
B117
B118
561
561
561
557
7,417
381
7,709
484
421
1,092
1,790
1,826
SI
Rl
S2
R2
S3
R3
S4
R4
S5
S6
S7
S8
S9
S10
R5
Supply
Return
Supply
Return
Supply
Return
Supply
Return
Supply
Supply
Supply
Supply
Supply
Supply
Return
56
54
55
58
56
57
57
56
140
144
128
120
142
132
860
N/A
Sll
S12
S13
S14
S15
S16
S17
Supply
Supply
Supply
Supply
Supply
Supply
Supply
151
170
145
148
153
53
56
N/A
S18
R6
S19
Supply
Return
Supply
209
1,280
194
N/A is defined as not applicable.
Bold room numbers are those rooms that nominally represent the zone and are referred to as nominal room numbers in other tables.
-------
Table 4. Upper Floor Airflow Measurements Under Moderate Recirculation
Zone ID Room Number Room Volume [ft3] Vent ID Type Measured Flow [cfm]
Ul
U2
U3
B209A
B211
B209B
B210
B229A
B229
B203
B204
B205
B206
B207
B208
B213
B214
B215
B216
B217
3,989
579
3,420
512
379
3,125
513
380
99
477
745
3,175
252
621
490
1,030
508
S20
S21
S22
S24
R7
S23
S25
S26
R8
S27
Supply
Supply
Supply
Supply
Return
Supply
Supply
Supply
Return
Supply
114
110
101
110
499
66
171
187
445
106
N/A
S28
S29
Supply
Supply
190
137
N/A
N/A
S30
S31
S32
S33
S34
S3 5
R9
S36
S3 7
S38
Supply
Supply
Supply
Supply
Supply
Supply
Return
Supply
Supply
Supply
61
67
141
113
90
171
1,240
75
110
116
N/A is defined as not applicable.
Bold room numbers are those rooms that nominally represent the zone and are referred to as nominal room numbers in other tables.
Table 5. Moderate Recirculation Airflow Summary
Measured Makeup Air Measured Recirculation
Zone ID Nominal Room Number Zone Volume [ft3] [ACH] [ACH]
LI
L2
L3
L4
L5
L6
Ul
U2
U3
B113
B114
B115
B116
B110E
B110
B209A
B209B
B229
561
561
561
557
7,798
13,322
4,568
4,311
11,415
1.0
1.0
1.0
1.0
1.0
1.0
1.1
1.1
1.1
5.0
4.9
5.0
5.1
5.1
4.8
5.5
5.4
5.6
-------
Table 6. Lower Floor Airflow Measurements Under Low Recirculation
Zone ID Room Number Room Volume [ft3] Vent ID Type Measured Flow [cfm]
LI
L2
L3
L4
L5
L6
B113
B114
B115
B116
B110E
B110D
B110
Bill
B112
B110B
B117
B118
561
561
561
557
7,417
381
7,709
484
421
1,092
1,790
1,826
SI
Rl
S2
R2
S3
R3
S4
R4
S5
S6
S7
S8
S9
S10
R5
Supply
Return
Supply
Return
Supply
Return
Supply
Return
Supply
Supply
Supply
Supply
Supply
Supply
Return
37
32
38
39
38
39
36
38
86
93
71
85
84
88
510
N/A
Sll
S12
S13
S14
S15
S16
S17
Supply
Supply
Supply
Supply
Supply
Supply
Supply
108
130
116
114
126
38
36
N/A
S18
R6
S19
Supply
Return
Supply
170
998
160
N/A is defined as not applicable.
Bold room numbers are those rooms that nominally represent the zone and are referred to as nominal room numbers in other tables.
-------
Table 7. Upper Floor Airflow Measurements Under Low Recirculation
Zone ID Room Number Room Volume [ft3] Vent ID Type Measured Flow [cfm]
Ul
U2
U3
B209A
B211
B209B
B210
B229A
B229
B203
B204
B205
B206
B207
B208
B213
B214
B215
B216
B217
3,989
579
3420
512
379
3,125
513
380
99
477
745
3,175
252
621
490
1,030
508
S20
S21
S22
S24
R7
S23
S25
S26
R8
S27
Supply
Supply
Supply
Supply
Return
Supply
Supply
Supply
Return
Supply
63
80
68
73
341
58
97
115
270
63
N/A
S28
S29
Supply
Supply
133
80
N/A
N/A
S30
S31
S32
S33
S34
S3 5
R9
S36
S3 7
S38
Supply
Supply
Supply
Supply
Supply
Supply
Return
Supply
Supply
Supply
30
30
95
69
38
125
780
38
67
80
N/A is defined as not applicable.
Bold room numbers are those rooms that nominally represent the zone and are referred to as nominal room numbers in other tables.
Table 8. Low Recirculation Airflow Summary
Nominal Room Measured Makeup Air Measured Recirculation
Zone ID Number Zone Volume [ft3] [ACH] [ACH]
LI
L2
L3
L4
L5
L6
Ul
U2
U3
B113
B114
B115
B116
B110E
B110
B209A
B209B
B229
561
561
561
557
7,798
13,322
4,568
4,311
11,415
1.0
1.0
1.0
1.0
1.0
1.1
1.1
1.0
1.0
2.9
3.1
3.1
2.9
2.9
3.4
3.4
2.9
3.1
-------
6.2 Leakage Tests
The second portion of the preliminary characterization
tests consisted of tests designed to determine the interzonal
leakage under normal operating conditions. According to the
results from the theoretical simulations, the quantification
of the interzonal leakage rate represents a key measurement
for this study. The interzonal leakage was to be measured by
setting the HVAC system to run with 100% fresh air (i.e.,
no recirculation, with a greatly increased fresh air makeup
air rate) and performing tracer gas tests with SF6. Assuming
there was no cross-contamination between the building
exhaust and fresh air inlet, any SF6 that entered the zone of
interest must be due to interzonal leakage. By performing this
leakage test at the various HVAC conditions planned for this
study, an experimental measure of a key parameter could be
determined. Other leakage measurement methods involve the
overpressurization of the zone of interest and do not reflect
normal building operation (ASTM, 2003).
The results of the initial leakage test indicated that
significant recirculation was occurring (see Appendix
A). Upon further investigation, it became apparent that
significant leakage was occurring across the recirculation
damper and that reducing that leakage to zero was not
feasible given the materials and time constraints of
the study. While it was possible to estimate the actual
recirculation using the theoretical model, this represented
an indirect method of estimating the interzonal leakage,
which depended heavily on the model-predicted value of
the recirculated airflow. For this reason, a combinatorial
approach to determining the leakage was adopted.
Under a combinatorial approach, the leakage rates for a given
building configuration (i.e., door position and recirculation
rate) are estimated using the results of a single experiment.
A leakage estimate was obtained by varying the leakage
parameter within the model to find a value that resulted
in good agreement between experimentally measured
and model-predicted results for a single experiment (i.e.,
a minimum in the square of the error between model-
predicted and experimentally measured values). The resulting
estimated leakage rates are then used for all tests under that
building configuration (i.e., door position and recirculated air
rate). The resulting leakage estimate should be valid for any
filtration efficiency since the filtration efficiency impacts the
concentration in the zone, not the leakage rate (i.e., the flow
rate across the leak path). In this fashion, a combinatorial
approach effectively used a portion of the experimental
data to directly estimate the interzonal leakage rate. The
application of this combinatorial method is further clarified in
the Results section of this report.
-------
7.0
Experimental Results
The test approach, which was adopted on-site, required
a reorganization of the originally planned test matrices.
Furthermore, since manual adjustments were needed for
each recirculation condition, test matrices were organized by
recirculation rate for efficiency reasons. Initially, all possible
combinations of three filtration levels and two leakage levels
were performed for two notional buildings for the moderate
recirculation rate. This provided a solid basis for comparison
of the experimental and model-predicted results, as well
as the large body of data required for the combinatorial
approach. Based on the interim results, it was agreed that a
reduced test matrix was appropriate for the low recirculation
condition given the excellent agreement of experimental
and model results observed for the moderate recirculation
condition (Sparks, 2005). For this reason, a limited matrix
using three filtration levels and three leakage levels for the
two notional buildings was executed for the low recirculation
condition. Tables 9 and 10 contain the moderate and low
recirculation test matrices, respectively.
The "large" notional building can be simplified in terms of
three well-mixed zones and is shown in Figure 10. Room
B113, which had a volume of 560 ft3, was the zone of
release (shaded blue in Figure 10). Room B114, which had
a volume of 560 ft3, was the zone of interest (shaded in cyan
in Figure 10). Several rooms constituted the lumped zone,
which represented the rest of the building in the model. The
lumped zone had a volume of approximately 42,480 ft3.
RoomBllO, which had a volume of approximately 7,710 ft3,
was selected as representative of the lumped zone for the rest
of the building (shaded in green in Figure 10). With the entire
test volume constituting approximately 43,600 ft3, the zone of
interest and zone of release each represented approximately
one-seventy-fifth (1/75) of the test volume.
The "small" notional building can be simplified in terms of
three well-mixed zones and is shown in Figure 11. Room
B209B, which had a volume of 4,310 ft3, was the zone of
release (shaded blue in Figure 11). Room B209A, which had
a volume of 4,570 ft3, was the zone of interest (shaded in
cyan in Figure 11). Due to the size of both the zone of release
and the zone of interest, four standard house fans were used
in each of these zones to promote mixing. Room B110, which
had a volume of approximately 7,710 ft3, was selected as the
room that would represent the lumped zone for the rest of the
building (shaded in green in Figure 11). With the entire test
volume constituting approximately 43,600 ft3, the zone of
interest and zone of release each represented approximately
one-tenth (1/10 ) of the test volume.
In addition to the aforementioned zones, Room B115,
which had a volume of 560 ft3, was selected as an additional
sampling location for the "large" notional building tests
(shaded gold in Figure 10). By comparing the "large"
notional building experimental test data from Rooms
B115 and B110, the validity of using a lumped zone
to approximate the majority of the test volume can be
established. Experimental data gathered during this study
indicate that the concentrations measured in Room B115
were indistinguishable from those in Room B110, thus
showing the validity of the lumped zone approach (lumping
multiple rooms into a single zone for modeling purposes).
This agreement is shown graphically in Figure 12. After
establishing the validity of this assumption, data from Room
B110 and Room B115 were averaged to produce the lumped
zone concentration curves for subsequent "large" notional
building tests.
-------
Table 9. Moderate Recirculation Test Matrix Results Key
Particulate
Particulate
Particulate
Particulate
Particulate
Particulate
Particulate
Particulate
Particulate
Particulate
Particulate
Particulate
Particulate
Particulate
"Moderate" (5 ACH)
~0)
60
CD
_l
"ro
E
CO
X
. o
O IT)
—i i-v.
i — i
CO
X
- O
J= <
.5?^
X CM
- °.
1 — 1
X
o
O IT)
-1 C\J
- i — i
O
X
- O
i= <
.5?^
x r-~
= i — i
i — i
"Low" (10%)
"Moderate-Low" (25%)
"Moderate-Low" (25%)
"Moderate" (50%)
"High" (90%)
"Low" (10%)
"Moderate" (50%)
"High" (90%)
"Low" (10%)
"Moderate" (50%)
"High" (90%)
"Low" (10%)
"Moderate" (50%)
"High" (90%)
15
12
13
14
16
18
17
19
25
20
22
24
21
23
"Large" notional building / "Low" leakage denotes a closed door position
"Large" notional building / "High" leakage denotes a door ajar position
"Small" notional building / "Low" leakage denotes a door open 20 cm
"Small" notional building / "High" leakage denotes a door fully open
-------
Table 10. Low Recirculation Test Matrix Results Key
Tracer
Filtration
Results Shown in Figure #
Particulate
Particulate
Particulate
Particulate
Particulate
Particulate
Particulate
Particulate
!<
= ro
~CP
s?
03
03
GO
o <
£* IT)
-g O
= o
X
0 O
-1 0
o
X
- O
.5Po
X LO
= 01
o
"Moderate"
(0.725ACH)
"Moderate" (50%)
"Low" (10%)
"Moderate" (50%)
"High" (90%)
"Moderate" (50%)
"Low" (10%)
"Moderate" (50%)
"High" (90%)
27
30
28
31
29
33
32
34
"Large" notional building / "Very Low" leakage denotes a closed door position
"Large" notional building / "Low" leakage denotes a door partially ajar
"Large" notional building / "High" leakage denotes a door open about 1 cm
"Small" notional building / "Moderate" leakage denotes a door open 70 cm
-------
-
t
(T3
E
Q.
n
Location
00
"o.
E
-------
o
•42
, OJ n.
-------
The initial set of tests focused on the "large" notional
building under moderate recirculation (5 ACH, see Tables 3
through 5 for further details). Figures 12 through 16 contain
graphical comparisons of experimental and model-predicted
data for four filtration levels at conditions of "low" leakage
(i.e., the door between B113 and B114 closed). Both
experimental and model-predicted data are presented as
concentration in particles per liter (ppl) as a function of time
with a release occurring at time zero. Experimental data
presented are from the 2.0 to 5.0 um channel of MetOne
particle counters. Under the combinatorial approach, this
set of experimental curves was used to estimate the "low"
leakage rate, as well as the filtration efficiencies. Estimates
were obtained by varying a parameter within the model
to find a value that resulted in good agreement between
experimentally measured and model-predicted results for
a single experiment (i.e., a minimum in the square of the
error between model-predicted and experimentally measured
values). The resulting parameters were then used for
subsequent tests under the appropriate conditions. Figure 12
illustrates the validity of the lumped zone assumption by
comparing two of the constituent zones (B 110 in green and
B115 in gold). The good agreement exhibited in Figure
12 is indicative of all the "large" notional building results.
As a result of this agreement, the lumped zone curves in
other "large" notional building graphical comparisons will
represent an average of these two zones. This validates
the use of the lumped zone assumption and illustrates
how a three-zone model can be used to provide useful
information on real buildings with more than three zones.
In addition, good agreement was observed between the
two samplers placed in the zone of interest, supporting the
well-mixed assumption in the small, 6' xlO' zone of interest
(see Figure 13). As a result of this good agreement, data
from the two samplers located in the zone of interest were
numerically averaged to represent this zone. A comparison of
Figures 12 and 13 also provides a limited demonstration of
the repeatability of these results.
In general, all experimental data gathered over the course of
this study exhibit imperfect mixing in the form of a mixing
lag. The mixing lags observed seemed to correlate well
with zone volume, meaning that smaller zones exhibited
smaller mixing lags (see Figures 14 and 20). A mixing lag is
a delay in the rise of a zone's experimental concentration in
comparison to the perfect mixing exhibited by a theoretical
model-predicted curve (see Figure 12). In addition to this, the
experimentally observed release zone data collected during
"large" notional building tests exhibit a plateau during the
initial peak. This plateau was due to the saturation limit of the
MetOne handheld particle counters used in this study. While
a reduction in release mass could eliminate this sampling
artifact, the release mass was nominally held constant to
provide sufficient paniculate levels in the lumped zone (i.e.,
the rest of the building) during high filtration tests.
The estimates of the effective filtration efficiency gleaned
from this series of tests provide valuable insight. The results
of this study indicate that even with no filter installed, the
basic HVAC system used in the test bed is 10% efficient in
removing paniculate in the 2.0 to 5.0 um range (see Figure
15). The data indicate that the MERV 7 and MERV 8 filters
used during this test performed below the rated efficiency
of their respective MERV ratings (25% estimated during
this study versus a specification of roughly 50 to 70% for
the MERV 7, 50% estimated during this study versus a
specification of 70% for the MERV 8). The MERV 7 results
are consistent with those of a related project (referred to
here as Task 2; Hecker, 2006), while no data from Task 2
were available for the MERV 8 filter. The 95% OOP filter
used during testing performed slightly below the rated
efficiencies (90% estimated during this study versus a
specification of 95%+ for the 95% OOP). Task 2 data on
the 95% DOP filter indicate a filtration efficiency greater
than 99% over the applicable particle size range (Hecker,
2006). It is possible that the discrepancy between the
results of this study and the rated efficiencies for the MERV
8 and 95% DOP filter are due to these estimates being
made in an operating building rather than in a specially
designed test fixture as used to obtain a MERV rating
(i.e., leakage around the filters or leaks in ducting could
be responsible for the reduced filtration efficiencies).
Figure 17 contains a graphical comparison of experimental
and model-predicted data for the "high" leakage condition
at moderate filtration (i.e., MERV 8 filter). Under the
combinatorial approach these data were used to estimate the
leakage rate associated with the "high" leakage condition
(i.e., door between B113 andB114 ajar). Having successfully
estimated both leakage rates (Figures 12 and 17) as well as
effective filtration efficiencies (Figures 12 through 16), the
remaining tests performed on the "large" notional building
under moderate recirculation illustrate direct comparisons
between experimental data and model predictions. Figures
18 and 19 graphically illustrate comparisons between
experimental data and model predictions in which the
value of all the HVAC parameters have previously
been determined. Although the agreement between the
experimental and model-predicted lumped zone (i.e., the
rest of the building) suggests a slight error in the filtration
efficiency estimates, on the whole, the agreement between
experimental and model-predicted data for the "large"
notional building under moderate recirculation is excellent.
-------
Given the promising results of the initial test set, a second
set of tests was undertaken focusing on the "small" notional
building under moderate recirculation conditions (5 ACH, see
Tables 3 through 5 for further HVAC flow details). Figures 21
and 22 contain graphical comparisons of the experimental
and model-predicted data for two leakage conditions with
a moderate filtration level (i.e., a MERV 8 filter). Under
the combinatorial approach previously adopted, these two
comparisons were used to estimate the leakage rates for
the "small" notional building under moderate recirculation.
Again, leakage estimates were obtained by varying the
leakage parameter within the model to find a value that
resulted in good agreement between experimentally measured
and model-predicted results for a single experiment (i.e., a
minimum in the square of the error between model-predicted
and experimentally measured values). The resulting leakage
parameter was then used for all tests performed under the
given combination of door position and recirculated air rate.
MetOne HHPC-6 Saturation Limit
1,000.000
100000
Q.
0.
10000
g 1,000
o
o
Lumped Zone
Assumption Verified
—Experimental Zone of Release
— Experimental Zone of Interest
— Experimental Resl of Building
- Theoretical Zone of Release
Theoretical Zone of Interest
Theoretical Resl of Building
Experimental Rest of Building II
> Mixing Lag
(Imperfect Mixing)
15
Time [Win]
25
35
Figure 12. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Moderate
Recirculation (5 ACH), Low Leakage (0.175 ACH), Moderate-Low Filtration (25% / MERV 7 Filter),
Standard Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that the Standard Infiltration
(0.5 ACH), Low Leakage (0.175 ACH), and Moderate-Low Filtration (25%) parameter was fit in this
comparison by minimizing the sum of the square of the residuals between the experimentally observed
and model-predicted data.
-------
1.00Q000
MetOne HHPC-6 Saturation Limit
Good Mixing in Small
Rooms
Mixing Lag
(Imperfect Mixing)
—Experimental Zone of Release
— Experimental Zone of Interest
— Experimental Rest of BuiWing Zone
- Theoretical Zone of Release
Theoretical Zone of Interest
Theoretical Rest of Building Zone
— Experimental Zone of Interest (Remote)
15
Time [Mm]
25
35
45
Figure 13. Demonstration of Repeatability by Comparison of Experimental and Model-Predicted Data for "Large"
Notional Building with Moderate Recirculation (5 ACH), Low Leakage (0.175 ACH), Moderate-Low
Filtration (25% / MERV 7 Filter), Standard Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH).
Note that all HVAC parameters for the model-predicted data were taken from previous comparisons with
experimental data.
1,000,000
MetOne HHPC-6 Saturation Limit
Mixing Lag
(Imperfect Mixing)
— Experimental Zone of Release
— Experimental Zone of Interest
— Experimental Rest of Building
—Theoretical Zone of Release
-Theoretical Zone of Interest
Theoretical Rest of Building
15
Time [Mm]
25
35
45
Figure 14. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Moderate
Recirculation (5 ACH), Low Leakage (0.175 ACH), Moderate Filtration (50% / MERV 8 Filter), Standard
Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that the Moderate Filtration (50%)
parameter was fit in this comparison by minimizing the sum of the square of the residuals between the
experimentally observed and model-predicted data.
-------
1.001000
100000
1QOOO
1,000
100
MetOne HHPC-6 Saturation Limit
10
Mixing Lag
(Imperfect Mixing)
-1S -5 5 15
Time [Win)
25
35
45
Experimental Zone of Release
Experimental Zone of Interest
— Experimental Rest of Building
— Theoretical Zone of Release
Theoretical Zone of interest
—Theoretical Rest of Building
Figure 15. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Moderate
Recirculation (5 ACH), Low Leakage (0.175 ACH), Low Filtration (10% / No Filter), Standard Makeup Air
(1 ACH), and Standard Infiltration (0.5 ACH). Note that the Low Filtration (10%) parameter was fit in this
comparison by minimizing the sum of the square of the residuals between the experimentally observed
and model-predicted data.
MetOne HHPC-6 Saturation Limit
1,000000
Mixing Lag
(Imperfect Mixing)
— Experimental Zone of Release
— Experimental Zone of Interest
Experimental Rest of Building
—Theoretical Zone of Release
-Theoretical Zone of Interest
—Theoretical Rest of Building
15
Time [Min]
Figure 16. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Moderate
Recirculation (5 ACH), Low Leakage (0.175 ACH), High Filtration (90% / 95% OOP), Standard Makeup
Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that the High Filtration (90%) parameter was
fit in this comparison by minimizing the sum of the square of the residuals between the experimentally
observed and model-predicted data.
-------
1,000.000
MetOne HHPC-6 Saturation Limit
Mixing Lag
(Imperfect Mixing)
10
-15
-5
15
Time [Win]
25
35
- Experimental Zone of Release
- Experimental Zone of Interest
- Experimental Rest of Building
-Theoretical Zone of Release
Theoretical Zone of Interest
-Theoretical Rest of Building
45
Figure 17. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Moderate
Recirculation (5 ACH), High Leakage (1.025 ACH), Moderate Filtration (50% / MERV 8), Standard
Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that the High Leakage (1.025 ACH)
parameter was fit in this comparison by minimizing the sum of the square of the residuals between the
experimentally observed and model-predicted data.
MetOne HHPC-6 Saturation Limit
1,000,000
10QOOO
Q.
Q.
O
1
10,000
1.000
100
Mixing Lag
(Imperfect Mixing)
15
Time [Win]
25
35
Experimental Zone of Release
— Experimental Zone of Interest
—Experimental Rest of Building
—Theoretical Zone of Release
- Theoretical Zone of Interest
—Theoretical Rest of Building
Figure 18. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Moderate
Recirculation (5 ACH), High Leakage (1.025 ACH), Low Filtration (10% / No filter), Standard Makeup Air
(1 ACH), and Standard Infiltration (0.5 ACH). Note that no HVAC parameters were fit in this comparison.
This comparison is a direct indication of agreement between experimental data and model predictions.
-------
MetQne HHPC-6 Saturation Limit
100000
I
o
u
Possible Error in Estimated.
Filtration Efficiency
Mixing Lag
(Imperfect Mixing)
15
Time [Win)
25
35
45
Experimental Zone of Release
Experimental Zone of Interest
Experimental Resl of Building
—Theoretical Zone of Release
Theoretical Zone of Interest
- Theoretical Rest of Building
Figure 19. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Moderate
Recirculation (5 ACH), High Leakage (1.025 ACH), High Filtration (90% / 95% OOP), Standard
Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that no HVAC parameters were fit in this
comparison. This comparison is a direct indication of agreement between experimental data and model
predictions.
Mixing Lag (Imperfect Mixing)
UDDQflOO
— Experimental Zone of Release
— Experimental Zone of Interest
— Experimental Resl of Building Zone
— Theoretical Zone of Release
— Theoretical Zone of Interest
—Theoretical Rest of Building Zone
— Experimental Zone of Release {Remote)
— Experimental Zone of Interest (Remote)
-15
Figure 20. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with Moderate
Recirculation (5 ACH), Low Leakage (0.125 ACH), Moderate Filtration (50% / MERV8), Standard
Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that the Low Leakage (0.125 ACH)
parameter was fit in this comparison by minimizing the sum of the square of the residuals between the
experimentally observed and model-predicted data.
-------
Throughout the tests for the "small" notional building, a
mixing lag was observed within both the zone of release
and the zone of interest. This mixing lag is evident when
comparing samplers located at different areas of the two
zones (for the zone of release, compare the dark blue and
light blue lines in Figures 21 and 22; for the zone of interest,
compare the cyan and red lines in Figures 21 and 22). Note
that locations labeled "remote" were located farther from
the source of particles (e.g., the release source in the zone of
release or the primary leakage path in the zone of interest).
From Figures 21 and 22, it is clear that, even with the
assisted mixing of the house fans, there was a finite lag in the
transport and mixing of Visolite® across the zones. However,
it is also clear that, after that finite mixing lag, the volume
appears to have been well mixed and subsequently agrees
very well with model predictions.
Since the filtration efficiencies have already been estimated
from the initial "large" notional building tests, the remaining
four tests illustrated direct comparisons between experimental
and model-predicted data. Figures 22 through 25 graphically
illustrate the excellent agreement between experimental data
and model predictions. The slight discrepancies observed can
easily be attributed to an initial mixing lag and a combination
of experimental error in measured and estimated parameters.
This excellent agreement was a clear indication that the
three-zone, well-mixed model is accurate in simulating the
effects of changes in filtration efficiency and interzonal
leakage rates for a notional building. Thus, after a cursory
analysis of the data, a subsequent series of tests was focused
on the effects of changing the recirculation.
Mixing Lag (Imperfect Mixing)
1,000000
—Experimental Zone of Release
—Experimental Zone of Interest
— Experimental Rest of Building Zone
—Theoretical Zone of Release
— Theoretical Zone of Interest
— Theoretical Rest of Building Zone
— Experimental Zone of Release {Remote)
— Experimental Zone of Interest (Remote)
15
Time [Min]
Figure 21. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with Moderate
Recirculation (5 ACH), High Leakage (1.175 ACH), Moderate Filtration (50% / MERV 8), Standard
Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that the High Leakage (1.175 ACH)
parameter was fit in this comparison by minimizing the sum of the square of the residuals between the
experimentally observed and model-predicted data.
-------
Mixing Lag (Imperfect Mixing)
15
Time [Win]
25
35
—Experimental Zone of Release
— Experimental Zone of interest
—Experimental Rest of Building Zone
— Theoretical Zone of Release
Theoretical Zone of Interest
— Tneonetical Real of Buildirtg Zone
—Experimental Zone of Release (Remote)
— Experimental Zone of Interest (Remote)
45
Figure 22. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with Moderate
Recirculation (5 ACH), Low Leakage (0.125 ACH), High Filtration (90% / 95% OOP), Standard Makeup
Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that no HVAC parameters were fit in this
comparison. This comparison is a direct indication of agreement between experimental data and model
predictions.
Mixing Lag (Imperfect Mixing)
15
Time [Win]
—Experimental Zone of Release
— Experimental Zone of Interest
— Experimental Rest of Building Zone
—Theoretical Zone of Release
Theoretical Zone of Interest
— Theoretical Rest of Building Zone
— Experimental Zone of Release (Remote)
— Experimental Zone of Interest (Remote)
45
Figure 23. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with Moderate
Recirculation (5 ACH), High Leakage (1.175 ACH), High Filtration (90% / 95% OOP), Standard
Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that no HVAC parameters were fit in this
comparison. This comparison is a direct indication of agreement between experimental data and model
predictions.
-------
Mixing Lag (Imperfect Mixing)
1,000000
100,000
Q.
CL
§
10,000
1.000
100
-15
25
—Experimental Zone of Release
— Experimental Zone of Interest
— Experimental Resl of Building Zone
—Theoretical Zone of Release
— Theoretical Zone of (merest
— Theoretical Rest of Building Zone
—Experimental Zone of Release (Remote)
—Experimental Zone of Interest (Remote)
-5 5 15
Time [Min]
Figure 24. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with Moderate
Recirculation (5 ACH), High Leakage (1.175 ACH), Low Filtration (10% / No filter), Standard Makeup Air
(1 ACH), and Standard Infiltration (0.5 ACH). Note that no HVAC parameters were fit in this comparison.
This comparison is a direct indication of agreement between experimental data and model predictions.
1,000000
100000
10.000
1,000
Mixing Lag (Imperfect Mixing)
IW
1A
-15 -5 5 15 26
Time [Min]
36 4
—Experimental Zone of Release
— Experimental Zone of Interest
— Experimental Rest of Building Zone
—Theoretical Zone of Release
Theoretical Zone of Interest
— Theoretical Rest of Building Zone
— Experimental Zone of Release (Remote)
—Experimental Zone of interest (Remote)
Figure 25. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with Moderate
Recirculation (5 ACH), Low Leakage (0.125 ACH), Low Filtration (10% / No filter), Standard Makeup Air
(1 ACH), and Standard Infiltration (0.5 ACH). Note that no HVAC parameters were fit in this comparison.
This comparison is a direct indication of agreement between experimental data and model predictions.
-------
After establishing the validity of the model in terms of
changing the filtration efficiency and interzonal leakage
for both a "large" and "small" notional building in the first
two test sets, the third set of tests focused on verifying the
effects of changing the recirculation rate. This process was
complicated by a lack of precise control over the interzonal
leakage rate. That is, a change in the recirculation rate or the
door position can and will result in a significantly altered
leakage value. Figure 26 illustrates this through photographs
of the various door positions that produce various interzonal
leakage rates at moderate and low recirculation. It is of
particular interest that after altering the air handling unit,
Phoenix valves, and dampers to produce the low recirculation
condition (see Tables 6 through 8 for details), the interzonal
leakage rate changed from approximately 0.175 ACH to
0.05 ACH for an identical door position (i.e., a closed door).
This variation indicates that the interzonal leakage rate can
be extremely sensitive to the HVAC operating conditions.
In addition, it is important to note the large variation in
interzonal leakage that results from a slight change in the
door position. This sensitivity of the interzonal leakage rate
to changes in HVAC operating conditions and door position
strongly support the decision to use a combinatorial approach
to estimate the leakage under normal operating conditions
rather than use a more standard blower door test method
that estimates the leakage under increased pressures and
extrapolates to the standard operating pressure.
Thus, due to the lack of precise control over the interzonal
leakage rate, an initial set of tests at low recirculation was
necessary to estimate the door position that would produce
a leakage level within an acceptable range. Figures 27
through 29 contain graphical comparisons of experimental
and model-predicted data for three leakage levels (i.e.,
varying door positions) for the "large" notional building
under low recirculation (3 ACH, see Tables 6 through 8 for
further details) and moderate filtration (i.e., MERV 8 filter).
Under the combinatorial approach, these experiments were
used to estimate the "very low," "low," and "high" leakage
rates under varying conditions (i.e., door position between
B113 and B114). Experimental data and model predictions
based on previously estimated HVAC parameters are
graphically displayed in Figures 30 and 31.
Door Closed
Door Partial ly Ajar
Door Ajar
Moderate (5 ACH)
Low (3 ACH)
0.175 ACH
0.050 ACH
Moderate (5 ACH)
Low (3 ACH)
Not Run
0.200 ACH
Moderate (5 ACH)
Low (3 ACH)
1.025 ACH
Not Run
Figure 26. Photographs of Various Door Positions for the "Large" Notional Building (i.e., the door between Bl 13 and
B114) and Recirculation Rates (i.e., low or moderate) with Corresponding Interzonal Leakage Rates
-------
1.00Q000
100000
10,000
1.000
100
MetOne HHPC-6 Saturation Limit
— Experimental Zone of Release
— Experimental Zone of Interest
— Experimental Rest of BuiWing
—Theoretical Zone of Release
- Theoretical Zone of Interest
-Theoretical Rest of Building
Mixing Lag
(Imperfect Mixing)
15
Time [Min]
25
35
45
Figure 27. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Low
Recirculation (3 ACH), Very Low Leakage (0.05 ACH), Moderate Filtration (50% / MERV 8 filter),
Standard Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that the Very Low Leakage (0.05
ACH) parameter was fit in this comparison by minimizing the sum of the square of the residuals between
the experimentally observed and model-predicted data.
MetOne HHPC-6 Saturation Limit
1,000,000
100,000
Mixing Lag
(Imperfect Mixing)
15
Time [Mln]
25
35
45
Experimental Zone of Release
— Experimental Zone of Interest
— Experimental Rest of Building
—Theoretical Zone of Release
Theoretical Zone of Interest
-Theoretical Rest of Building
Figure 28. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Low
Recirculation (3 ACH), Low Leakage (0.200 ACH), Moderate Filtration (50% / MERV 8 filter), Standard
Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that the Low Leakage (0.200 ACH)
parameter was fit in this comparison by minimizing the sum of the square of the residuals between the
experimentally observed and model-predicted data.
-------
MetOne HHPC-6 Saturation Limit
1,000,000
100,000
10,000
1,000
100
10
Mixing Lag
(Imperfect Mixing)
-15
15
Tims [MIn]
25
35
—Experimental Zone of Release
— Experimental Zone of Interest
—Experimental Rest of Building
—Theoretical Zone of Release
Theoretical Zone of Interest
—Theoretical Rest of Building
Figure 29. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Low
Recirculation (3 ACH), High Leakage (0.950 ACH), Moderate Filtration (50% / MERV 8 filter), Standard
Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that the High Leakage (0.950 ACH)
parameter was fit in this comparison by minimizing the sum of the square of the residuals between the
experimentally observed and model-predicted data.
MetOne HHPC-6 Saturation Limit
1,000,000
100,000
a.
a.
1
10000
1,000
100
10
Mixing Lag
{Imperfect Mixing)
— Experimental Zone of Release
— Experimental Zone of Interest
— Experimental Rest of Building
—Theoretical Zone of Release
Theoretical Zone of Interest
—Theoretical Rest of Building
-15
-5
25
35
45
5 15
Time [Min]
Figure 30. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Low
Recirculation (3 ACH), Low Leakage (0.200 ACH), Low Filtration (10% / No filter), Standard Makeup Air
(1 ACH), and Standard Infiltration (0.5 ACH). Note that no HVAC parameters were fit in this comparison.
This comparison is a direct indication of agreement between experimental data and model predictions.
-------
MetOne HHPC-6 Saturation Limit
1.000,000
10
1
-15 -5 5 15 25 35
Time [Win]
4
—Experimental Zone of Release
Experimental Zone of Interest
—Experimental Rest of Building
— Theoretical Zone of Release
Theoretical Zone of Interest
— Theorelical Rest of Building
Figure 31. Comparison of Experimental and Model-Predicted Data for "Large" Notional Building with Low
Recirculation (3 ACH), Low Leakage (0.200 ACH), High Filtration (90% / 95% OOP filter), Standard
Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that no HVAC parameters were fit in this
comparison. This comparison is a direct indication of agreement between experimental data and model
predictions.
As with previous comparisons, the agreement between
experiment and model for the "large" notional building under
low recirculation was generally quite good.
Given the excellent agreement between model and
experimental data, a reduced number of runs, consisting of
three levels of filtration efficiency at one leakage condition,
were executed for the "small" notional building under low
recirculation. The leakage rate was estimated from the
moderate filtration test (see Figure 32), leaving the remaining
two tests as a direct comparison of model-predicted and
experimentally measured data (see Figures 33 and 34).
Again, the agreement observed between experimental and
model curves for a "small" notional building under "low"
recirculation is rather favorable.
The excellent agreement observed between model-predicted
and experimental data for changing recirculation rates
indicates that the three-zone, well-mixed model is accurate
in simulating the effects of changes in the recirculation rate
for a notional building. The consistent nature and excellent
quality of the model agreement with experimental data
indicates that, in addition to accurately simulating the effects
of changes in filtration efficiency and interzonal leakage
rates, the model can accurately predict the effects of changing
recirculation rates for a notional building.
-------
1.000000
10
-15
Mixing Lag
(Imperfect Mixing)
15
Time [Win]
25
35
— Experimental Zorve of Release
—Experimental Zone of Interest
— Experimental Rest of Building Zone
—Theoretical Zone of Release
— Theoretical Zone of Interest
- Theoretical Resl of Building Zone
— Experimental Zone of Release (Remote)
— Experimental Zone of Interest (Remote)
45
Figure 32. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with Low
Recirculation (3 ACH), Moderate Leakage (0.725 ACH), Moderate Filtration (50% / MERV 8 filter),
Standard Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that the Moderate Leakage
(0.725 ACH) parameter was fit in this comparison by minimizing the sum of the square of the residuals
between the experimentally observed and model-predicted data.
1.000.000
Mixing Lag
(Imperfect Mixing)
—Experimental Zone of Release
—Experimental Zone of Interest
—Experimental Rest of Building Zone
—Theoretical Zone of Release
— Theoretical Zone of Interest
— Theoretical Rest of Building Zone
—Experimental Zone of Release (Remote)
—Experimental Zone of Interest (Remote)
15 25 35 45
Time [Min]
Figure 33. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with Low
Recirculation (3 ACH), Moderate Leakage (0.725 ACH), Low Filtration (10% / No filter), Standard
Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that no HVAC parameters were fit in this
comparison. This comparison is a direct indication of agreement between experimental data and model
predictions.
-------
1,000,000
10
Mixing Lag
(Imperfect Mixing)
-15
15
Time [Win]
25
35
—Experimental Zone of Release
—Experimental Zone of Interest
—Experimental Rest of Building Zone
—Theoretical Zone of Release
— Theoretical Zone of Interest
— Theoretical Rest of Building Zone
— Experimental Zone of Release (Remote)
—Experimental Zone of Interest (Remote)
45
Figure 34. Comparison of Experimental and Model-Predicted Data for "Small" Notional Building with Low
Recirculation (3 ACH), Moderate Leakage (0.725 ACH), High Filtration (90% / 95% OOP filter), Standard
Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Note that no HVAC parameters were fit in this
comparison. This comparison is a direct indication of agreement between experimental data and model
predictions.
-------
8.0
Discussion of Experimental Results
There was excellent agreement between experimentally
observed and model-predicted data. For this comparison,
the time to reach a critical exposure (Ct) of 50,000 particles
per liter (ppl)* minutes, and the cumulative exposure at
30 minutes, were selected as the performance metrics
to compare model estimates with measured data. These
performance metrics were selected to allow comparison
of the different operating conditions and hence the effect
of varying building/HVAC operating variables of interest.
Tables 11 and 12 contain the tabulated performance
metrics for experimental and model-predicted data for
moderate and low recirculation conditions, respectively.
Given that the average percentage errors are 9% for
the time to reach the Ct and 13% for the exposure at 30
minutes, it is clear that the excellent agreement, which
has been displayed graphically in Figures 12 through 34,
is also present in the calculated performance metrics.
That is to say that the model-predicted performance
metrics agree well with the experimentally determined
performance metrics within the experimental variability.
Ideally, a stringent analysis of parameter impact would be
conducted using the experimental data. This type of analysis
would compare two experiments in which only a single
parameter of interest varied and would quantify the change
in performance metrics due to the change in the parameter
of interest. A slight variation in the mass released and a lack
of precise control over the leakage preclude this type of
stringent analysis. Due to the formation of small deposits
within the eductor nozzle, the mass released during each test
varied by approximately +/- 0.6 grams from the mass loaded
into the release mechanism. For this reason, comparisons
between experimental results convolute the effects of
changing both a release variable (i.e., the release mass) and
key system variables (i.e., anHVAC parameter). In addition,
it was not possible to "dial" in the leakage value when
considering different notional buildings (i.e., different rooms)
or differing recirculation flow rates. This lack of precise
control over the leakage is due to the realistic nature of the
leakage path and driving force used in this study. The natural
leakage path of a door and the natural driving force of slight
HVAC flow imbalances resulted in realistic results but did not
provide precise control of the leakage itself. Because of the
excellent agreement of model estimates with experimental
results, the model allows for this type of stringent analysis to
be made with confidence.
While the variable release mass and a lack of precise control
over the leakage preclude a stringent analysis of parameter
impact from the experimental results, some generalizations
concerning parameter impact can be made from the
experimental data gathered during this study. For "large"
notional buildings, the significant changes in performance
metrics between high and low leakage rates suggest that
the leakage is the dominant parameter (see Table 11).
In comparison, it appears that changes in the filtration
efficiency and system recirculation rate have lesser impacts
on the performance metrics for a "large" notional building
scenario (see Tables 11 and 12). In contrast to the "large"
notional building results, the results for a "small" notional
building imply that the filtration efficiency is the dominant
parameter while the leakage may play a secondary role (See
Table 11). These generalities represent the trends observed
in the experimental data. Given the excellent agreement
between modeled and experimental data, the following
section (Section 9.0) develops an approach to determining the
relative impact of key parameters and to finding out how well
a parameter must be known to assess an attack.
-------
Table 11. Comparison of Experimental and Predicted Performance Metrics for the Zone of Interest for the "Moderate"
Recirculation Condition
Experimentally Measured
Model-Predicted
Filtration
Time to Ct
of 50,000
ppl*min [min]
Exposure @ 30
min [ppl*min]
Time to Ct
of 50,000
ppl*min [min]
Exposure @
30 min
[ppl*min]
"Moderate" (5 ACH)
~CP
op
CD
_l
"ro
E
CO
X
. o
£<
O IT)
—i i-v.
CJ
X
- o
J= <
.9?^
X C\J
= o
1 — 1
X
o
O IT)
-1 C\J
- i — i
O
X
: O
-^
-------
9.0
Model Impact Analysis
While it is neither feasible nor realistic to conduct
experiments under all possible conditions and configurations,
the model represents a tool that can easily and reasonably
be used to analyze a tremendous number of possible
conditions and building configurations. This section
focuses on developing an approach to determining
which parameter has the largest impact and subsequently
applying that approach to a number of combinations
of system parameters and building configurations.
9.1 Impact Analysis Approach
In evaluating the impacts of changing various system
variables, it is necessary to choose some type of performance
metrics for quantifying the protection of the building
occupants. Two performance metrics were considered in
this study: (1) the level of exposure in the zone of interest
(Zone 2) over the first 30 minutes (E30i2), and (2) the time to
reach an effective dosage Ct of interest (tct). These metrics
represent the severity and speed of the exposure, respectively,
and provide an accurate description of the impact of HVAC
parameters on exposure in the zone of interest. The exposure
metric (E30i2) alone can yield valuable information about
the performance of the system to reduce total exposure. The
examination of only the time to reach an effective Ct (tct)
gives an indication useful for response time analysis. While
the specific details of the chosen performance metrics (e.g.,
30 minutes as the timeframe for the exposure metric) relate
to acute hazards, which likely occur in the unsteady state
immediately following a release, alternative details could be
chosen to apply this approach to longer timeframes. In this
respect, this impact analysis approach is very flexible and
could be applied to numerous different scenarios through the
choice of appropriate performance metrics.
The combination of these exposure metrics summarizes
valuable information about the performance of the system,
while conclusions based on only one metric may ignore
critical performance details. For example, while time-
based metrics are important for evacuation and protection
strategies, they do not directly correlate to the total exposure
for a given scenario. In a similar fashion, exposure-based
metrics do not provide any indication of how quickly a
critical exposure value is realized within a zone.
-------
20
18
1 16
J
fc_ 14
I 12
| 10
HI
£ 8
6
4
2
0
u
Filtration Efficiency = 80%
Filtration Efficiency = 90%
Filtration Efficiency = 99,99%
Arbitrary Critical Exposure
0
10 15
Time (min)
20
25
30
Figure 35. Illustration of Exposure-Based Metric Inadequacies for a Hypothetical Case of Various
Filtration Efficiencies (80, 90, and 99.99), Equal Zone Volumes (Vi=V2=V3), Makeup Air
1 ACH), Moderate Recirculation (5 ACH), Infiltration (0.5 ACH), and High Leakage (1 ACH)
For illustration, Figure 35 contains plots of cumulative
exposure for Zone 2, for a hypothetical case similar to some
of the experimental conditions used during the field study
except that all three-zone volumes are equal. By plotting an
arbitrary critical exposure value (e.g., 10 CFU/m3 *min), the
importance of both types of metrics are graphically illustrated
in Figure 35. (Note that CPU stands for colony-forming
unit.) The arbitrary critical exposure value is reached at 6.1
and 8.8 minutes for filtration efficiencies of 80% and 90%,
respectively, while the cumulative exposure for a filtration
efficiency of 99.99% never reaches the arbitrary critical
value. Thus, considering a time-based metric, one might
conclude that changing the filtration efficiency from 80% to
90% has relatively little impact on the system performance.
However, the total cumulative exposure for these two cases
varies by 50% (17.2 CFU/m3 *min for 80% versus 11.8 CPU/
m3 *min for 90%), which may be significant to occupant
survival. For these reasons, both a time-based metric and an
exposure-based metric are needed to assess performance.
The goal of this impact analysis is to determine the relative
impact of the various HVAC and building parameters on
the performance metrics (i.e., the parameter that has the
largest impact on the protection offered by the building).
To accomplish this, a series of runs was performed varying
parameters one at a time over their particular ranges of
variability. To graphically display the sensitivity of the
performance metrics to changes in various parameters, a
normalization procedure was applied to eliminate any bias
towards parameters with inherently small magnitudes and/
or limited ranges. For example, filter collection efficiency
may realistically range over five logs (i.e., ~0 to 99.99%),
but infiltration rates, only one (i.e., 0.1 to 1). To this end,
both the performance metrics and the parameters were
normalized by baseline values and the results plotted
(for examples, see Figures 36 and 37). In Figure 36, the
normalized performance metric (E30/E30 base) is plotted versus
the normalized study parameter (P/Pbase)- By comparing
the gradient of the plot at the analysis baseline condition
(i.e., the slope of the curves at P/Pbase=l), the effects of a
relative change in the model parameter on the performance
metric can be compared. As one would intuitively expect,
a larger gradient (i.e., a steeper slope) indicates that a
small relative change in the model parameter will produce
a large change in the performance metric. For example,
a smaller slope for the filter efficiency curve (shown in
dark blue in Figures 36 and 37) and a larger slope for the
leakage rate (shown in bright red in Figures 36 and 37)
-------
indicate that the leakage rate between rooms has a larger
relative impact on the exposure metric for the hypothetical
building than the filter efficiency. Please note that the
filtration efficiencies were subjected to a log transform
according to Equation 6. This log transform frames filtration
efficiencies such that they may vary from zero to infinity.
p
1 Filtration Efficiency
1
(6)
100-[Filtration Efficiency]
Analysis baseline condition / Point of equation.
0129466
Normalized Parameter [P/PgjuJ
Figure 36. Normalized Exposure at 30 Minutes (E302) Versus Model Input Parameters for a "Large"
Notional Building (V3=100V1) Under 1 ACH Makeup Air, 5 ACH Recirculation, 0.3 ACH
Infiltration, 30% Filtration, and 1 ACH Interzonal Leakage Note that P/PbaSe=l signifies the
analysis baseline.
Analysis baseline condition.' Point of evaluation
o i 2 3 4 a e
Normalized Parameter [P/P^jJ
Figure 37. Normalized Time to Critical Exposure (tct) Versus Model Input Parameters for a "Large"
Notional Building (V3=100V!) Under 1 ACH Makeup Air, 5 ACH Recirculation, 0.3 ACH
Infiltration, 30% Filtration, and 1 ACH Interzonal Leakage Note that P/Pbase=l signifies the
analysis baseline.
-------
Table 13. Typical Change in Parameters for Use in Calculating Scale Factors
Model Parameter Set Change(a) Baseline Value(b) Scale Factor(c)
Filter Efficiency [%]
Makeup Air Rate [ACH]
Recirculation Rate [ACH]
Infiltration Rate [ACH]
Room Leakage [ACH]
^ FVilter Efficiency W)
0.25 ACH
0.5 ACH
0.1 ACH
0.25 ACH
30%
1 ACH
5 ACH
0.3 ACH
1 ACH
0.25
0.25
0.10
0.33
0.25
(a) denoted APget Change 'n Equations 8 and 9
(b) denoted Pbase 'n Equations 8 and 9
(c) denoted as the quantity (APget Change/Phase) 'n Equations 8 and 9
(d) Va, Ppjiter Efficiency f°r a f''ter change depends on the filter efficiency. For a 10% efficient filter, 1A of Ppiiter Efficiency 's equal to 18.0%, while for a
99% efficient filter, Vt of Ppiiter Efficiency is eclual to 0.2%.
The sum of the absolute value of the gradients of the two
normalized performance metric curves associated with
a model parameter (i.e., the quantity E'JftP | + t'ctP |),
evaluated at P/Pbase=l, are an indication of how much impact
a parameter has. Equation 7 describes this mathematically
where E'JOP is the slope of the normalized exposure metric
curve at P/Pbase=l for parameter P (with P denoting any of the
five parameters from the legend of Figures 36 or 37), t'ctP is
the slope of the normalized time to critical exposure metric
curve at P/Pbase=l for parameter P (with P denoting any of the
five parameters from the legend of Figures 36 or 37), Pbase
is the base case parameter value (i.e., the notional building
being considered), AP is the change in parameter P being
considered (i.e., a reasonable change in the parameter), and
the quantity (AE3(/E3abase + Atc/tctibase) is taken to be the
cumulative change in performance metrics possible for a
potential change in parameter P of AP.
Ct,P
A£3o
E
-
**<*
tajase
AP
p
A Base
(7)
Equation 7 provides an indication of the impact of a
reasonable change in a parameter; however, it must be
adjusted so that (1) any dependence on the baseline parameter
is eliminated and (2) once the baseline parameter dependence
has been eliminated, a factor to account for the magnitude
of each parameter is included to allow for comparisons
between parameters (a detail that normalization by the
baseline parameter was previously used to accomplish). To
accomplish both of these items, a scale factor defined as the
ratio of a set change in the parameter (e.g., changes in filter
efficiencies are likely on the order of a quarter of a log while
changes in leakage rates are likely on the order of 0.25 ACH)
to the baseline parameter was introduced. An illustration is
shown in Table 13, which lists the assumed values for typical
changes in a parameter, the baseline parameter values, and
the calculated scale factors.
The absolute value of the gradients of the two curves
associated with a model parameter, that is the quantity
E'so.p | + t'CiP | where E'30 P and t'ct P denote the slopes for
each of the performance metrics of the curve for a given
parameter (P) evaluated at P/Pbase=l, can then be multiplied
by the resulting scale factors to produce an impact score for
each parameter (see Equations 8 and 9). In Equations 8 and
9, IP is the impact score, APSet change is the set change in a
parameter, and the quantity (AE3(/E30ibase + Atc/tct:base) is
taken to be the cumulative change in performance metrics
possible for a potential change in parameter P of AP. In this
way, the relative impact of each parameter on the model can
be quantified.
Ip=-
Set Change _
At
ct
AP
1 Base
AP (8)
AP
LU
Set Change
Ct,P
?t Change
(9)
*
For the sake of simplicity, the impact scores are then
renormalized by the maximum value. This sets the range
of possible impact scores to between zero and unity, which
provides a set scale of relative impact (i.e., an impact score
of unity indicates the most important model parameter). In
this way, the resulting normalized impact score indicates
the relative impact of simulation parameters on model
performance (i.e., the higher the normalized impact score,
the stronger the effect on model performance).
Table 14 illustrates the results from this analysis. The
normalized impact scores indicate that leakage is the
dominant factor in determining model performance with
recirculation, filtration, makeup air, and infiltration all
playing lesser roles. By noting the sign (i.e., positive
or negative) of the slope (see Figures 36 and 37 for the
illustrative case), the nature of the dominant mechanism
associated with the parameter (i.e., delivery or removal
-------
mechanism) can be ascertained. Thus, for the illustrative
case, it can be concluded that the delivery of contaminant
is limited by the leakage rate while removal of contaminant
is limited by the combination of the recirculation, filtration,
and makeup air that dilute and filter the air in the simplified
HVAC system. Using this approach, it is possible to conclude
that reducing the interzonal leakage is easily identifiable as
the key parameter in improving the protection offered by this
building. This is not to say that other parameters do not affect
the exposure metrics, but rather that the interzonal leakage
has the largest impact for parameters scrutinized for the set
change used in the analysis. The impact analysis does not
identify the magnitude of the impact; it only identifies the
relative impact of each parameter.
9.2 Impact Analysis Limitations
As with any analysis, it is important to understand the
limitations of the method outlined above as well as to
understand the applicability of the results. To illustrate these
two concepts, an alternative set of baseline parameters with
a greatly reduced leakage (i.e., all parameters are the same as
the previously detailed baseline case except for an interzonal
leakage of 0.01 ACH) was used for comparison with the
illustrative results of the previous two sections. Comparisons
between the alternative baseline results and the illustrative
results from Sections 9.1 are intended to demonstrate
that: (1) the results and discussion stated in this work are
applicable only to cases described by the parameter set used
in the analysis; (2) inferences made based on the gradients
of the normalized plots (e.g., Figures 36 and 37) are valid
only in the parameter space represented by the baseline case
(i.e., at or near P/PBase=l); and that (3) while this analysis
was designed for typical buildings under standard daytime
operation (e.g., the baseline parameters outlined in Table 13),
it is valid for atypical buildings under nonstandard operating
conditions (e.g., an alternative baseline with a greatly
reduced leakage rate of 0.01 ACH). Thus, while the specific
analysis results presented in this document may be valid only
for notional buildings under standard daytime operation, the
analysis approach used here is valid for virtually all buildings
under any operating conditions.
Figure 38 is the analog to Figure 36 for the alternative
baseline case (i.e., original baseline case with a greatly
reduced leakage rate of 0.01 ACH). Note that inFigure 38,
the gradient of the infiltration rate curve (shown in green)
seems small in comparison to that of the filter efficiency
curve (shown in blue), while in Figure 36, the two curves
are very similar (i.e., they exhibit approximately the same
slope atP/PBase=l inFigure 36). This difference indicates
that the relative importance of both the filtration efficiency
and the infiltration rate depend on other parameters. This
illustrates that the results of an analysis are valid only when
the parameters used in the analysis describe the situation of
interest. No set of parameters is universally applicable. Some
situations are not represented by the parameter set used in
this illustration (e.g., a building equipped with a HEPA filter).
It is also essential to note that comparing the slopes of the
curves at P/PBaSe ^ 1 is not valid since those parameter spaces
represent the simultaneous variation of multiple parameters.
For example, comparing the slopes of the curves from
Figure 38 for filter efficiency and leakage rate at P/PBase= 2
is comparing the impact of a relative change in room leakage
for a leakage of 2 ACH and 30% filter efficiency to the
impact of a relative change in filter efficiency for a leakage
of 1 ACH and 65% filter efficiency. While this information
is interesting, no conclusions can be drawn without further
exploring and defining the parameter space surrounding those
points (i.e., repeat the analysis of the previous two sections
for an alternative set of baseline conditions corresponding to
the parameter space represented by each of those points).
It is also important to note that the impact analysis outlined
here is focused on zones directly linked via leakage to the
zone of release. An analogous approach could be used to
determine the impact of various parameters on the rest of the
building (i.e., zones not directly linked to the zone of release
via leakage). This could be done by treating the exposure and
time to peak within the third, lumped zone in an identical
manner as the exposure and time to peak for the zone of
interest were treated in the impact analysis (see Section 9.1).
Table 14. Normalized Impact Scores for the Base Case for a "Large" Notional Building (V3=100V1) Under 1 ACH
Makeup Air, 5 ACH Recirculation, 0.3 ACH Infiltration, 30% Filtration, and 1 ACH Interzonal Leakage
Sum of Normalized Renormalized
Zone 2 Parameter Gradients13' Scale Factor(b) Impact Score(c) Impact Score
Filter Efficiency
Makeup Air Rate
Interzonal Leakage
Infiltration Rate
Recirculation Rate
0.075
0.245
1.32
0.060
1.16
0.25
0.25
0.25
0.33
0.10
0.019
0.061
0.33
0.020
0.116
0.057
0.18
1.00
0.061
0.35
(a) denoted E'j^p + t'ct:P \ in Equations 8 and 9
(b) denoted as the quantity (APSet Change/Phase) in Equations 8 and '
(c) denoted IP in Equations 8 and 9
-------
£
u?
Hi
(A
25
- Filter Efficiency
- Makeup Air
- Infiftration
- Leakage
- Recirculation
Leakage 0.01 AC'I I, Filler Efficiency 30%
0123
Normalized Parameter
Figure 38. Normalized Exposure at 30 Minutes (E302) Versus Model Input Parameters
for a "Large" Notional Building (V3=100Vi) Under 1 ACH Makeup Air, 5 ACH
Recirculation, 0.3 ACH Infiltration, 30% Filtration, and 0.01 ACH Interzonal
Leakage. Note that P/Pbase=l signifies condition is identical to the previously
described base case with the exception of a greatly reduced leakage rate (0.01 ACH).
9.3 Preliminary Impact Analysis Results
In order to identify the key parameters of interest, a
preliminary set of model simulations were performed.
These simulations examined the impact of the makeup air,
recirculation, infiltration, leakage, and filtration efficiency
for notional buildings of varying sizes for a typical set
of building parameters (i.e., 1 ACH Makeup Air, 5 ACH
Recirculation, 0.3 ACH Infiltration, 1 ACH Leakage, and
30% Filtration).
To examine the effects of building size, the volume of Zone
3 was set to 1, 5, 10, 25, 50, and 100 times the volume of
a standard 142-m3 room, yielding Zone 3 volumes of 142,
710, 1,420; 3,550; 7,100; and 14,200 m3, respectively. Next,
each of the HVAC parameters was individually varied from
the typical set of building parameters. Table 15 contains
the impact analysis results for the typical set of building
parameters for a Zone 1 release for various Zone 3 volumes.
These results can also be viewed graphically as displayed in
Figure 39. Note that the curves connecting the data points
are merely to help illustrate the dependency. For reference,
Table 16 contains the typical set of building parameters used
and the ranges used in the analysis. The detailed results, in
the form of graphs of normalized performance metrics versus
normalized parameters, can be found in Appendix B.
The low normalized impact scores for the makeup air
and infiltration indicate that over the range of building
volumes modeled, for the typical building parameters and
a release scenario within a zone (Zone 1), the makeup
air and infiltration have little effect on the performance
metrics. Thus, the makeup air and infiltration rates would
not be parameters of interest in attempting to improve the
performance of a building described by the typical building
parameters against a similar release scenario. From a
theoretical standpoint, it is interesting to note the shape of the
system makeup air impact curve, which exhibits a maximum
near a building volume of -1,000 m3 (which corresponds to
a Zone 3 volume of five times that of Zone 2). This region
indicates that for low volumes (i.e., V3 < 5V2), described
by the base case conditions, the dilution provided by the
system makeup airflow rate of 1 ACH is of slightly greater
importance than the system recirculation rate of 5 ACH.
As building volume (i.e., the Zone 3 volume) is increased
beyond -1,000 m3, the dilution effects of the system
recirculation surpass those of the system makeup air.
-------
Table 15. Impact Analysis Results for Various Zone 3 Volumes Under 1 ACH Makeup Air, 5 ACH Recirculation, 0.3
ACH Infiltration, 30% Filtration, and 1 ACH Interzonal Leakage
Zone 2 Parameter
Normalized
Impact Score
Filtration Efficiency
Leakage
System Makeup Air
System Infiltration
System Recirculation
Filtration Efficiency
Leakage
System Makeup Air
System Infiltration
System Recirculation
Filtration Efficiency
Leakage
System Makeup Air
System Infiltration
System Recirculation
1.00
0.21
0.16
0.06
0.07
1.00
0.87
0.26
0.10
0.20
0.76
1.00
0.33
0.10
0.32
Zone 2 Parameter
Normalized
Impact Score
Filtration Efficiency
Leakage
System Makeup Air
System Infiltration
System Recirculation
Filtration Efficiency
Leakage
System Makeup Air
System Infiltration
System Recirculation
Filtration Efficiency
Leakage
System Makeup Air
System Infiltration
System Recirculation
0.42
1.00
0.28
0.17
0.41
0.12
1.00
0.21
0.07
0.37
0.06
1.00
0.21
0.07
0.38
Table 16. Typical Parameter Values and Model Parameter Ranges
Variable Min Base case Max
Filter Efficiency [%]
Makeup Air Rate [ACH]
Recirculation Rate [ACH]
Infiltration Rate [ACH]
Room Leakage [ACH]
Zone 1 Volume [m3]
Zone 2 Volume [m3]
Zone 3 Volume [m3]
0
0
0
0
0
N/A
N/A
143
30
1
5
0.3
1
143
143
14,300
99.9999
15
15
2
2
N/A
N/A
14,300
N/A is defined as not applicable.
-------
1 • *
+ Filtration Efficiency
* Interzonal Leakage
— — System Makeup Air
System Reciroolafen
— — System Infiltration
50
Zone 3 Volume [xV,]
75
100
Figure 39. Impact Analysis Results for Various Zone 3 Volumes Under 1 ACH Makeup Air, 5 ACH
Recirculation, 0.3 ACH Infiltration, 30% Filtration, and 1 ACH Interzonal Leakage
The results of the nitration efficiency, recirculation rate,
and leakage rate are more interesting. The analysis shows
that the most important parameter affecting building
performance (with respect to changing the exposure or time
for a critical exposure to occur) changes as the building
size —Zone 3— changes. For building volumes less than
—3,800 m3 (which corresponds to a Zone 3 volume of 25
times that of Zone 2), the dominant factor is clearly the
removal of agent from the combined recirculation stream
via filtration. Figure 39 indicates that as the building
volume increases beyond -1,000 m3 (i.e., V3=5V2), the
dominant factors are the leakage rate, the recirculation
rate, and, to a lesser extent, the filtration efficiency. This
trend suggests that as the building size increases, or more
specifically as the volume of recirculated air increases,
the main mitigation mechanism becomes dilution, while
the filter efficiency becomes of lesser importance. As this
occurs, the interzonal leakage, which acts as a delivery
mechanism, becomes of increased importance and assumes
the role of the dominant factor determining performance.
The above discussion regarding the use of impact scores to
assess which parameters are most important and how well
the parameters need to be known illustrates that a single,
global generalization cannot necessarily be made. That is, the
effect of parameters on the spread of contaminant depends
on numerous other parameters and on the system HVAC
configuration/operation. Consequently, in assessing factors
that affect building performance, a good understanding of the
building operation is needed. It would then be best to assess
the building of interest using the simplified model approach
presented here. The usefulness of the simplified model
approach is that detailed modeling and input data sets would
not be required for an initial assessment.
9.4 Parameter Space Map Impact Analysis
Based on preliminary modeling results, a map of the
parameter space defined by varying interzonal leakage rates,
system recirculation rates, and general filtration efficiencies
was constructed. For "large" and "small" notional building
volumes, simulations exploring these three key variables
were performed. Interzonal leakage rates of 1.0, 0.7, 0.4,
and 0.1 ACH were used to examine the effects of various
leakage rates. These values were chosen to approximate the
range over which the interzonal leakage can realistically be
controlled. General filtration efficiencies of 10, 30, 60, 90,
and 99% were used to further map out the parameter space of
interest. Again, these values were chosen to approximate the
range over which the filter efficiency could realistically vary.
For each combination of filtration efficiencies and leakage
rates, further simulations were performed at system
recirculation rates of 3, 5, and 7 ACH. These values were
chosen based on rough estimates of the maximum system
-------
recirculation rates achievable by conventional HVAC
systems. Generally speaking, there are two main types of air
handling systems used in typical commercial office buildings:
(1) fixed volume systems and (2) variable air volume systems
(Bell, 2000). Fixed volume systems typically operate at full
capacity (100%), and thus no further increase in the system
recirculation rate is possible (Bell, 2000). Variable air volume
systems are used in higher-end office buildings and operate
at 50 to 75% of full capacity (Bell, 2000). Based on this
information, a maximum system recirculation rate of 7 ACH
was determined to be appropriate for this work. In order to
provide insight into the impact of the system recirculation
rate, a recirculation rate of 3 ACH was also added to the
simulation matrix.
Also based on preliminary results, building volumes of
5 and 100 times that of the zone of interest were chosen
to represent "small" and "large" notional buildings,
respectively. For "large" and "small" notional building
volumes, each combination of these three parameters (i.e.,
filtration efficiency, leakage rate, and recirculation rate) was
used as an analysis baseline case. Thus, for each possible
combination of the aforementioned values, the procedure
described in Section 9.1 (illustrated in Figures 36 and 37, as
well as Table 14) was performed.
9.4.1 "Large" Notional Building Parameter Space Map
Results
A "large" notional building was defined for this effort as
having a volume of 14,600 m3. An example of the "large"
building, with a nominal ceiling height of 3 m, might include
a 1,200-m3 lobby (20 m x 20 m), ten 75-m3 reception areas
(5 m x 5 m), twelve 150-m3 conference rooms (10 m x
5 m), one hundred twenty 75-m3 offices (5 m x 5 m), and
approximately 1,850 m3 of hallways, stairwells, elevators,
and restrooms. This "large" notional building would have a
standard occupancy of 134 to 148, based on estimates of one
person per office, four to eight persons in the lobby, and one
to two persons in the reception areas.
The impact analysis results for the parameter space of
interest for a Zone 1 release in a "large" building are listed
in Table 17. The parameter space map is also illustrated
by a series of three-dimensional surface plots in Figures
40 through 42. Note that the three-dimensional plots use a
surface spline to connect the actual data points contained in
Table 17 and are intended as an illustrative tool to graphically
present the results.
-------
Q_
03
O>
o
03
Q_
CO ^
-2-1
"5 o3
1 =
03 0>
O £
g.E
o
o>
=! 03
m •>
O)
"
03 03
r-1 <=
o>
O> O>
.£= -Q
14-1 t/)
^ 03
t/) C
±i o
££
-------
System Recirculation
Leakage
Fillnititm
oo
» » « * eif nTio-«
•^ton Effi«*ncy J%] Fill*t™ E'teancyfis.j mcy ,„
Figure 40. "Large" Building (14,600 m3) Parameter Mapping Results for a Recirculation Rate of 7 ACH
20»
'"trailon Efficiency [%i
System Recirculalian
Leakage
nitration
FHlratiw, Effidwicy f%j F^«EW Effictaney (nj
Figure 41. "Large" Building (14,600 m3) Parameter Mapping Results for a Recirculation Rate of 5 ACH
System Recirculation
Leakage
Filtration
1 D
i Efficiency (%j
Figure 42. "Large" Building (14,600 m3) Parameter Mapping Results for a Recirculation Rate of 3 ACH
-------
The parameter space map data for a notional "large" building
indicate that under most conditions the leakage rate between
zones has the largest impact on building performance for an
interior release. Even for extreme conditions not sampled in
this study where another parameter may have a larger impact,
the leakage still would have a large impact (i.e., greater than
0.6 on a scale normalized to unity). This strongly suggests
that for a "large" building, the interzonal leakage rate is the
key parameter in improving the performance of the building
against an internal release.
In understanding these results, it is paramount to consider
that the impact analysis indicates the parameter for which
a reasonable change (i.e., the Set Changes in Table 13) will
produce the largest change in building performance (i.e., the
sum of the performance metrics described in Section 9.1).
To further elucidate this trend, let us consider our three key
parameters (recirculation, leakage, and filtration) in terms of
a rough mass balance around the zone of interest. Leakage
from the release zone and the common HVAC recirculation
represent the two potential routes by which contaminant may
enter the zone of interest at any given instant in time. In the
parameter space studied, leakage from the release zone to
the zone of interest represents a flow path of 0.1-1 ACH with
a concentration equal to that of the release zone. Similarly,
in the parameter space studied, the system recirculation
represents a flow path of 3-7 ACH with a concentration equal
to roughly 1/100 of the concentration in the release zone,
based on a building volume of 100 times that of the zone of
interest. Considering these estimates, it is clear that leakage
from the release zone is the dominant delivery mechanism.
Similarly, leakage from the zone of interest, system
recirculation, and filtration represent the potential routes
by which contamination may be removed from the zone of
interest at any given instant in time. In the parameter space
studied, the leakage from the zone of interest represents a
removal rate of 0.1-1 ACH (for the purposes of this crude
analysis, the makeup air and infiltration will be neglected).
The system recirculation represents a removal rate of
3-7 ACH in the parameter space of this study. Filtration
represents a reduction in the concentration delivered by the
system recirculation. At first glance, filtration may appear
to be the dominant removal mechanism; however, it is
important to consider that the reduction in the recirculated
concentration due to filtration is in addition to the reduction
in the recirculated concentration due to dilution. For the
"large" building, the concentration in the system recirculation
has already been diluted by a factor of 100 due to the size
of the building. Thus, the impact of a 90% efficient filter on
the performance metrics can be approximated as the impact
of changing the concentration of the recirculated flow from
roughly 1/100 of the release zone concentration to roughly
1/1,000 of the release zone concentration. Clearly, the impact
of this change in the recirculated concentration is much less
than the impact of the leakage, which can be approximated
as the impact of changing the interzonal leakage rate, which
possesses a contaminant concentration equal to that in the
zone of release. In this manner, the impact of filtration on the
performance metrics is marginalized by dilution in the system
recirculation in a "large" building under the conditions
studied here.
These results imply an interesting trend in addition to the
importance of the interzonal leakage. As illustrated in
Figures 40 through 42, the filtration impact tends to decrease
as filtration efficiency increases. The slope of this trend is
increased at higher recirculation rates and minimal at lower
recirculation rates, confirming the coupled nature of filtration
and recirculation. Thus, although the recirculation rate has
a minimal impact score in and of itself, it has a significant
effect on the filtration impact.
9.4.2 "Small" Notional Building Parameter Space
Map Results
A "small" notional building was defined for this effort as
having a volume of 1,000 m3. A notional example of the
"small" building, with a nominal ceiling height of 3 m,
might include a 300-m3 lobby (10 m x 10 m), a 150-m3
conference room (10 m x 5 m), six 75-m3 offices (5 m x 5 m),
and approximately 100 m3 of hallways and restrooms. This
"small" notional building would have a standard occupancy
of approximately seven to eight persons based on estimates
of one person per office and one to two people in the lobby.
The impact analysis results for the parameter space of interest
for a Zone 1 release in a "small" building are listed in Table
18. The parameter space map is also illustrated by a series
of three-dimensional surface plots in Figures 43 through 45.
Note that the three-dimensional plots use a surface spline
to connect the actual data points contained in Table 18 and
are intended as an illustrative tool to graphically present the
results. In order to further reduce the data, a parameter space
map indicating the approximate parameter with the highest
impact is given in Figure 46.
The impact analysis results for the notional "small" building
are more complex and "feature rich" than their "large"
building counterparts. Similar to the "large" building results,
the parameter space maps for the "small" building indicate
that at high filtration efficiencies the interzonal leakage has
the largest impact on building performance. However, at
lower filtration efficiencies the filtration efficiency has the
largest impact. The division between these two regimes
appears to be linked to recirculation, varying from roughly
75% at a recirculation rate of 7 ACH to approximately 30%
at a recirculation rate of 3 ACH. Thus, while recirculation
is not a dominant parameter in terms of its direct impact, it
does play a determining role in what is the dominant impact
parameter. These results indicate that, generally speaking, for
the notional "small" building, removal of contaminant via
filtration and delivery of contaminant via interzonal leakage
have the largest impacts on building performance for an
interior release.
-------
Which of these parameters dominates appears to depend
mainly on the filtration efficiency with interzonal leakage
dominating at high filtration efficiencies and filtration
efficiency dominating at low filtration efficiencies. As
described above, recirculation rate plays a secondary role
by altering the division between the filtration-dominant
region and the leakage-dominant region. Thus, the critical
filtration efficiency value will depend on both the building
size (relative to the zone of interest) and the system
recirculation rate. For example, the critical filtration
efficiency is approximately 50% for a building volume of
1,000 m3 (relative to a zone of interest volume of 143 m3)
and a recirculation rate of 5 ACH, while the critical filtration
efficiency is roughly 75% for a building volume of 1,000 m3
(relative to a zone of interest volume of 143 m3) and a
recirculation rate of 7 ACH. Again, as in previous sections,
the estimated accuracy requirements follow the impact
analysis results in that high impact parameters have tight
estimated accuracy requirements.
The marked differences between the "small" and "large"
building results imply that for a "small" building the delivery
of contaminant via interzonal leakage and the removal
of contaminant via filtration of the recirculated air are of
comparable magnitudes. The reason for this is the large
reduction in the dilution of contaminant in the recirculated
stream (a factor of ~5 versus a factor of -100). With the
reduction in dilution, contaminant removal via filtration of
the recirculated air becomes the dominant mechanism for the
removal of contaminant from the zone of interest. Thus, the
reduction in dilution has greatly increased the importance
(i.e., the impact) of filtration and created regions in the
parameter space where a change in the filtration efficiency
will produce the largest change in building performance (i.e.,
the performance metrics).
-------
-------
Recife illation
Leakage
I i!l> LI 11MM
»»*»«» «ns"
"*•*>" E"ta-««y»j _.,«„
Figure 43. "Small" Building (1,000 m3) Parameter Mapping Results for a Recirculation Rate of 7 ACH
P«"»ton Efficiency (%J
Rccircutation
Filtration
Figure 44. "Small" Building (1,000 m3) Parameter Mapping Results for a Recirculation Rate of 5 ACH
Recirculation
Leakage
Filtration
F*r«wn Efficiency |*j
F«ra
-------
Recirculation7ACH
Legend
Filtration Efficiency
Leakage
0.1 ACH LOACH
Leakage
Recirculation 5 ACH
0.1 ACH 1.0 ACH
Leakage
Recirculation 3 ACH
0.1 ACH LOACH
Leakage
Figure 46. Parameter Space Map of the Dominant Parameter, or Parameter with the Highest Impact Score,
for a "Small" Building (1,000 m3) (Note that grey regions are intended to indicate a region where
two or more parameters are dominant.)
-------
9.5 Functional Analysis Guidelines
Although it is not possible to make blanket statements about
dominant parameters, a simple functional analysis can be
used to develop some general guidelines regarding which
HVAC parameters are important for a range of situations.
This simple functional analysis will generalize the key
delivery and removal mechanisms for the zone of interest
(i.e., Zone 2) in terms of simple functional forms.
The simplified analysis begins with combining the makeup
airflow rate and the recirculation flow rate into a lumped
HVAC flow rate (see Equation 10).
expression (i.e., Equation 13) is an approximation and
assumes that the lumped HVAC flow rate for any given zone
is much larger than the interzonal leakage entering or exiting
that zone. Assuming that the concentration within the zone of
release is the highest of the three zones, this approximation
represents an overestimation of the contributions of the
HVAC flow to contaminant delivery. While this approach
is not exact, it is more than sufficient for a simple, rough
analysis such as this.
HVAC.
* C2(t) + Qmc? * C3(Q
(13)
Q.
HVACj
(10)
Where Qti, QR>;, and QHVAC.I denote the makeup airflow rate,
the recirculated airflow rate, and the lumped HVAC flow rate
for Zone i.
Furthermore, since both the makeup air and the recirculated
airflow rate are nominally functions of the zone volume,
the lumped HVAC flow rate can be rewritten as an integer
function of the zone volume (see Equation 11).
Assuming that the recirculation and makeup air are system-
wide variables (i.e., each zone has the same recirculation
and makeup air when expressed in units of air changes
per hour), this expression can be further reduced using the
previously derived expression for the lumped HVAC flows
as a function of zone volume (i.e., Equation 11) to a volume
weighted average of the concentrations within each zone (see
Equation 14). The resulting expression is quite helpful in
estimating the contributions of common HVAC system to the
delivery of contaminant.
Q,
HVAC,i
~ ™ '
(11)
(14)
Where Vi denotes the volume of Zone i and N is an integer
frequency, nominally on the order of 3-8 hr1.
Neglecting infiltration (i.e., assuming there is no infiltration
or that the magnitude of infiltration is so small that it can be
ignored), there are two delivery mechanisms for the zone of
interest (i.e., Zone 2). One delivery mechanism is interzonal
leakage from the zone of release (i.e., Zone 1), which is
functionally dependent on the product of the leakage flow
rate, denoted QLeak, and the concentration within the zone of
release, denoted C](t). The other delivery mechanism is the
simple HVAC system, where the delivery of contaminant
is functionally dependent on the product of the lumped
HVAC flow (QHVACU), the concentration in the common
ductwork (CHvAc(t)), and a filtration efficiency term (l-r\).
It is important to note that both the concentration within the
zone of release (Q (t)) and the concentration in the common
HVAC ductwork (CHvAc(t)) are functions of time under
non-steady-state conditions. Despite the time dependency of
the concentrations, a general functional form for the delivery
term is obtained (see Equation 12).
[Delivery] « Q^ * C,(f) + (1 -r\)
(12)
This delivery mechanism functional form can be further
reduced by approximating the concentration in the common
HVAC ductwork (CHVAc(t)) in terms of the HVAC flows
and concentrations of each of the contributing zones
(Equation 13). It is important to note that the resulting
Similarly, when infiltration is neglected, there are two
mechanisms for the removal of contaminant from the zone of
interest (i.e., Zone 2). One removal mechanism is interzonal
leakage from the zone of interest to the bulk of the building
(i.e., Zone 3), which is functionally dependent on the product
of the leakage flow rate, denoted QLeak, and the concentration
within the zone of interest, denoted C2(t). The other removal
pathway is the simple HVAC system, where the removal
of contaminant is functionally dependent on the product of
the lumped HVAC flow (QHvAc,2) and the concentration in
the zone of interest (C2(t)). Again, the concentration within
the zone of interest (C2(t)) is a function of time under non-
steady-state conditions. Following this approach, a general
functional form for the removal of contaminant is obtained
(see Equation 15).
[Removal] « QLeak * C2(f) +
C2(f) (15)
Through consideration of the functional forms of the
delivery mechanisms (Equation 12), removal mechanisms
(Equation 15), and the contaminant concentration in the
common HVAC ductwork (Equation 14) a rough, back-of-
the-envelope analysis of a given situation can be performed
to provide an initial indication as to which parameters are
most important and which parameters, if any, can potentially
be neglected or ignored.
-------
9.5.1 Contaminant Transport Dominated by HVAC
Mechanisms
One potential condition is that the delivery and removal
of contaminant is dominated by the HVAC mechanisms.
This would mean that the HVAC terms in the delivery and
removal functional forms dominate (i.e., are much larger
than their leakage counterparts). This dominance can be
mathematically expressed using the individual terms from
Equations 12 and 15 (see Equations 16 and 17 for the
delivery and removal, respectively).
(1 -Tl )
(0 » QLeak * Q (0 (16)
Dividing the equation obtained from manipulating
the removal functional form (i.e., Equation 17) by the
concentration within Zone 2 yields a mathematical statement
which, intuitively, must be true for HVAC mechanisms to
dominate (see Equation 18).
Q.
HVAC,1
Leak
(18)
Assuming that the lumped HVAC flow for the zone of
interest is much larger than the interzonal leakage flow rate,
the derived expression for the concentration in the common
HVAC ductwork as a volume weighted average of the zone
concentrations can be substituted into the delivery functional
form (see Equation 19).
Assuming that the mass of contaminant in the zone of release
is significantly larger than the mass of contaminant in the
rest of the building (i.e., VjC^t) » V2C2(t)+ V3C3(t)), and
dividing both sides of the equation by the concentration in the
release zone and the lumped HVAC flow for Zone 2, yields a
time-independent statement (see Equation 20).
(1-r,)'
ZLeak
Qt
(20)
HTAC,2
By comparing the reduced delivery and removal forms
(Equations 20 and 18, respectively), it is apparent that
the delivery condition is more severe (i.e., if the delivery
condition stated in Equation 20 is met, the removal
condition in Equation 18 will be satisfied). In this fashion,
the conditions whereby the common HVAC system will
dominate can be identified (i.e., Equation 20).
For example, under conditions of little or no filtration (i.e.,
(1-T|) ~ 0), Zone 1 must represent a volume fraction of the
building smaller than the ratio of the interzonal leakage flow
rate to the lumped HVAC flow for Zone 2 (see Equation 21).
V
V\
iLeak
'HVAC,2
(21)
Furthermore, if the interzonal leakage was equivalent
to 1 ACH and the lumped HVAC flow for Zone 2 was
equal to 6 ACH (e.g., 1 ACH of makeup air and 5 ACH of
recirculation), Zone 1 would need to be significantly larger
than 1/6 of the building volume (see Equation 22).
(22)
Thus, for the HVAC flows to dominate the delivery of
contaminant, Zone 1 must represent a very large fraction of
the building. Effectively, this will happen when the building,
or rather volume served by the common HVAC ductwork,
is small. In addition, any filtration will further increase the
required size of Zone 1 in relation to the rest of the building.
9.5.2 Contaminant Transport Dominated by Interzonal
Leakage
Another potential condition is that the delivery and removal
of contaminant is dominated by the interzonal leakage. In
this case, the leakage terms in the delivery and removal
functional forms would dominate (i.e., would be much larger
than their HVAC counterparts). Again, this dominance can
be mathematically expressed using the individual terms
from Equations 12 and 15 (see Equations 23 and 24 for the
delivery and removal, respectively).
1HVAC,2
lHVAC,2
c2(o
(24)
Dividing the equation obtained from manipulating
the removal functional form (i.e., Equation 24) by the
concentration within Zone 2 yields a mathematical statement
which, intuitively, must be true for leakage mechanisms to
dominate (see Equation 25).
» Q.
HVAC,2
(25)
-------
Given that the nitration term is, by definition, less than unity
(i.e., (1-T|) < 1) and the concentration in the zone of release is
larger than the concentration in the common HVAC ductwork
under all but trivial conditions (i.e., C](t) > CHvAc(t)X the
removal condition is more severe than the delivery condition
(i.e., if the removal condition stated in Equation 25 is met,
the delivery condition in Equation 23 will be true). Thus, a
very intuitive statement regarding the conditions under which
the interzonal leakages dominate the removal and delivery of
contaminant is derived (i.e., Equation 25).
9.5.3 Perfect Filtration
Using this approach, it is also possible to examine the
effects of perfect filtration (i.e., a filtration efficiency of
100%). Under perfect filtration, the filtration term in the
delivery mechanism goes to zero (i.e., (!-T|) —> 0) and the
leakage clearly dominates the delivery of contaminant (see
Equation 26). However, the filtration has no impact on the
removal functional form (see Equation 27).
[Delivery] »
[Removal] « QLeak * C2 (f) +
(26)
;C2(0 (27)
Thus, while perfect filtration eliminates the delivery of
contaminant via the common HVAC system, it does not
affect the removal functional form. Under the nominally
"common" conditions of 1 ACH of leakage and 6 ACH of
lumped HVAC flow, perfect filtration would result in leakage
dominating the delivery of contaminant and HVAC flows
dominating the removal of contaminant.
-------
-------
10.0
In-Room Air Cleaners
The effects of an in-room air cleaner were also briefly
investigated. During the in-room air cleaner test, a
commercially available air cleaner (Whirlpool Model
AP4503HO, see Figure 47) was operated in the zone of
interest (B114), as well as in one of the small rooms, which
were lumped together with the rest of the building (B115).
Table 19 contains a comparison of the performance metrics
for the zone of interest for tests performed under moderate
recirculation (5 ACH) and moderate filtration (50%) with
and without the in-room air cleaner operating. The results
indicated that the in-room air cleaner successfully reduced
the concentration at 30 minutes by roughly one order of
magnitude and prevented the Ct from reaching the critical
Ct value selected for this study. These results are shown
graphically in Figure 48, where the impact of an in-room
air cleaner on the zone of interest (B 114) can be viewed as
the difference between the purple and magenta data plots
and the impact of an in-room air cleaner on a small room
not adjacent to the zone of release (B115) can be viewed as
the difference between the neon green and turquoise data
plots. In contrast to the significant reduction in concentration
observed in rooms with in-room air cleaners, rooms without
an in-room air cleaner displayed a negligible reduction in
concentration during the in-room air cleaner tests (denoted
by the difference between the neon green and dark green data
plots in Figure 48).
These plots clearly illustrate that whether the in-room air
cleaner is operating in a room near the release or far from the
release, it can have a large impact on the concentration within
that particular zone/room. This result is not surprising given
the high throughput (-440 cfm) and HEPA-grade filtration
efficiency of the in-room air cleaner used.
Figure 47. Photograph of In-Room Air Cleaner Used During
Testing (Whirlpool Model AP4503HO)
-------
Table 19. Comparison of Experimental Performance Metrics for the Zone of Interest for the "Moderate" Recirculation
Condition With and Without an In-Room Air Cleaner
"Moderate" (50%)
with IAC
N/R denotes that specified cumulative Ct was not reached.
N/A is defined as not applicable.
— IAC1 Experimental Zone of Release
IAC1 Experimental Zone of I merest (IAC)
— IAC1 Experimental Rest of the Building (No IAC)
IAC1 Experimental Rest of Building (IAC)
— B1 Experimental Zone of Release
B1 Experimental Zone of Interest (No IAC)
— B1 Experimental Rest of Building (Mo IAC)
o
I
o
o
10000
1000
-5,00
5.00
15.00
Time [Min]
25.00
35,00
45.00
Figure 48. Comparison of Experimental Data for an In-Room Air Cleaner for a "Large" Notional Building with
Moderate Recirculation (5 ACH), Low Leakage (0.175 ACH), Moderate Filtration (50% MERV 8 Filter),
Standard Makeup Air (1 ACH), and Standard Infiltration (0.5 ACH). Test IAC1 was conducted with in-
room air cleaners in selected zones denoted as (IAC). Zones without in-room air cleaners during test IAC1
are denoted as (No IAC). Test Bl was conducted with no in-room air cleaners.
-------
11.0
Conclusions and Recommendations
A three-zone model of a building was developed to determine
which HVAC and building operating parameters are most
important, and how accurately they need to be known, to
determine the impact of an indoor bio-agent attack. The
three-zone model consists of a zone of release and zone
of interest that are adjacent and of equal size, as well as a
"lumped" zone that consists of the remainder of the building.
All three zones are serviced by a single HVAC system with
common recirculation.
An experimental study was then conducted to verify the
effectiveness and validity of using the three-zone model
to approximate contaminant spread in a building. In the
experimental study in the test building, modifications were
made to the building and HVAC system to assess the impact
those changes had on the spread of the contaminant. The
building was modified through the addition of new slab-to-
slab walls, the extension of existing partial walls, and the
addition of several return vents to create a test volume that
resembled the model.
Two performance metrics were used to compare the model
and experimental results: the cumulative exposure to the
paniculate agent for the first 30 minutes following a release
and the time to reach a critical exposure value (referred to as
Ct). On the whole, the excellent agreement observed between
experimental and model-predicted data validated the use of
the three-zone model to approximate contaminant spread
in a representative building. Excellent agreement between
the lumped third zone in the model and multiple rooms
throughout the test building validated the use of a three-zone
model to provide useful information on real buildings with
more than three zones.
A sensitivity analysis method was then developed for the
three-zone model to provide an estimate of how well a
parameter must be known to assess the impact of an attack, as
well as to determine which parameter has the largest impact.
An initial analysis using the model simulation revealed that
building makeup air and infiltration rates were noncritical
parameters for an internal release scenario and that system
parameters were of greater importance than single-zone
parameters (i.e., the recirculation rate of the entire building
was more important than the recirculation rate of the zone
of interest). The initial analysis also identified the building
size, interzonal leakage rate, recirculation rate, and filter
efficiency as the key parameters affecting the two selected
performance metrics. The simulation results also showed
that the most important parameter to consider depended on
building size relative to the fixed volume zone of interest
(e.g., a 3,000-m3 building composed of 100 rooms that are
each 30 m3 in volume is equivalent to a 50,000-m3 building
composed of 100 rooms that are each 500 m3 in volume).
As the building size increased, the filtration efficiency went
from being a potentially dominant parameter to a lesser
factor compared to the interzonal leakage. The system
recirculation was found to be of secondary importance in
and of itself but had a strong effect on the importance of the
filtration efficiency for smaller buildings (i.e., for building
volumes less than five times the zone of interest). With
all parameters studied, as the importance of the parameter
increased, so did the accuracy with which it needs to be
known for reliable estimates of performance metrics.
The results of the experimental study corroborated the
findings and trends identified by the model simulations.
The most important conclusion to be made from the
model simulation results, and the supporting experimental
measurements of the spread of a contaminant, is that
there is not a dominant parameter, nor a single value of
how accurately it needs to be known, to accurately assess
the impact of an indoor release in a building. The model
approach, as supported by the experimental data, provides
a useful and easy tool for estimating and assessing which
parameters most impact the spread of contaminant in a
building. Buildings of varying size and HVAC performance
can be assessed by varying the corresponding model
parameters accordingly.
The analysis method provided does indicate the general
trends and identifies key parameters for specific combinations
of HVAC parameters and building volumes relative to
the zone of interest as discussed above. The simplified
modeling approach developed here could be used to assess
various scenarios and buildings of specific interest, without
the need for extensive knowledge of HVAC and building
parameters. Using the modeling tool may have merit for
rapidly identifying the value of modifying the building to
enhance the protection of occupants or to mitigate the spread
of contaminants.
Ancillary experiments suggest that the use of an in-room
air cleaner can greatly reduce the paniculate matter level
in the room. The magnitude of this reduction will vary
greatly depending on the volume of the room, as well as the
throughput and efficiency of the in-room air cleaner.
Recommendations for future study focus on two aspects
of the well-mixed model: usability and applicability. The
usability of the model could be improved by developing
a user-friendly graphical user interface (GUI) that would
allow casual users (e.g., building operators) to rapidly
perform simple impact analyses for a building of interest
given limited building information (e.g., HVAC settings,
building and room volumes). The development of a GUI
-------
would increase the utility and impact of this effort by making return ductwork system. Developing an analogous model
the model available to more people. Also recommended is that effectively represents buildings with multiple-return
an enhancement to expand the applicability of the tool by ductwork systems (i.e., multiple air handling units) and
developing and verifying an analogous model for a building performing an experimental verification, similar to this
with a more complex HVAC ductwork scheme. The present work, would aid in making a model more applicable to large
model is only applicable to a building with one common- buildings with more complex air handling schemes.
-------
12.0
References
American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE). ASHRAE
Terminology of Heating, Ventilation, Air Conditioning, & Refrigeration. (2nd edition). Atlanta, Georgia:
ASHRAE, 1991.
American Society for Testing and Materials (ASTM). ASTMStandard E 779-03: Standard Test Method for
Determining Air Leakage Rate by Fan Pressurization. West Conshohocken, Pennsylvania: ASTM, 2003.
Bell, A.A. HVAC Equations, Data and Rules of Thumb. New York: McGraw Hill, 2000.
Hawkins and Hofacre, 2006.
Hecker, R.T "Development of Performance Information for Common Ventilation Filters." Battelle Report
to U.S. Environmental Protection Agency, Draft Final Report GS-10F-0275K, Task Order 1105, Task 2
(in progress).
Sparks, Leslie, Ph.D. U.S. Environmental Protection Agency, Personal Communication, 2005.
-------
-------
Appendix A
SFfi Experimental Methods and Results
SF6 Experimental Methods
Release Methods. The tracer gas used was SF6. SF6 is an
inert, nonhazardous tracer gas and thus yields an accurate
indication of the air distribution without being affected
by surface interactions (e.g., reactions or adsorption). It
is also safe to use in indoor applications with occupants
present and can be detected at trace (tens of parts per
trillion) concentrations. The amount of SF6 released was
calculated by considering the maximum safe concentration
of SF6 according to established safety limits. The maximum
concentration in any zone should not exceed the time
weighted average (TWA) of the permissible exposure limit
(PEL) for the tracer gas used, which for SF6, is equivalent
to 1,000 ppm (MSDS, 2004). For the purposes of this
calculation, a release zone volume of 125 m3 was assumed
(roughly 22 x 20 x 10). Since the maximum observed
concentration will be in the release zone, the maximum
release mass will be 0.30 kg of SF6 (50 liters at STP).
This amount of SF6 (0.30 kg) will produce a maximum
concentration of 400 ppm in a release zone with a volume of
125 m3. If this release amount of SF6 is safe for a small room,
it will be safe for release in a larger, better ventilated hallway,
and will certainly be safe for release into an HVAC system
with even larger makeup air rates, serving an even larger
area. Based on a rough calculation of the entire building as a
single volume with one air change per hour, this release mass
should also produce measurable levels of SF6 throughout
large portions of the building and, thus, will be more than
adequate for this field study.
To facilitate model comparisons, the release duration
of the SF6 will be adjusted to correspond to release
characteristics used in model simulations (i.e., a
10-minute release duration). This will be achieved
through the use of an orifice (or other flow restriction)
to effectively meter the flow of SF6 from a cylinder.
Sampling Methods. Operation of all air samplers will be
conducted per standard operating procedures (SOPs). All
personnel that will operate the samplers were trained on their
operation and demonstrated proficiency of use during dry
runs. Operators signed the SOP acknowledgement sheet, and
the test leader confirmed that the personnel were proficient
with their operation.
During the testing period, an automated air sampler
developed by Battelle will be used to collect air samples at
selected locations within the area of interest. The functional
components of the automated sampler are composed of
one diaphragm pump, 12 three-way solenoid valves, 12
Tedlar® sampling bags, a flow restriction, and a custom-
designed electronics control board. Up to 100 samplers
will be used. A pyramidal sampling control concept will
be used to centralize control of the sampling parameters.
Thus, a single computer will dictate the sampling parameters
for each serialized sampler in a text file. The text file
containing the sampling parameters for each of the 100
serialized samplers will then be distributed via ten handheld
PODs (handheld program transmitter), which also serve
to retrieve the data log from the samplers and record the
barcodes of sample bags (see Figure A-l). The PODs
are roughly the size and shape of a standard handheld
transmitter (e.g., a palm pilot). This sampling concept will
allow for centralized control of the sampling parameters,
while retaining the convenience of having multiple staff
members distribute the sampling parameters. In addition,
staff members will not be required to be present to collect
samples, and thus, will not interfere with the local airflow
and transport of tracer. In this approach, the test leader
will direct the sample coordinator regarding the final input
and verification of the sampling event to be executed.
The automated sampler will use a serial configuration of
three-way valves to minimize any carryover concerns and
will offer flexibility in dictating various sample parameters
such as purge time, sample time, and delay time between
samples. The samples collected will then be analyzed using
a GC/ECD to determine the SF6 concentration of each
sample. A schematic diagram of the sampler's pneumatic
configuration when used to collect air samples for SF6
analysis is illustrated in Figure A-2. Photos depicting
the internal and external view of the samplers in this
configuration are provided in Figure A-3.
-------
1 Computer
(Master Program)
A
lOPODs
A
100 Battelle Air Samplers
Figure A-l. Pyramidal Sampling Control Concept
-------
Purge exhaust to ambient
Bypass/Purge
(not actuated)
Sample
(actuated)
Legend
pump
Three-way Solenoid
TedlarBag
Ambient Air
Figure A-2. Automated Sampler Pneumatic Configuration
-------
Figure A-3. Internal and External Pictures of the Automated Sample
-------
Each of the automated samplers operate at a fixed flow,
ranging from 20 to 500 cc/min. (within ± 10%), through
the use of a flow restriction (e.g., a capillary tube). Exact
flow rates for each test will be determined by the task leader
so that the desired concentration versus time profile for
the specific tests can be obtained. Operating with 500 cc
Tedlar® sampling bags, sampling durations of 1 to 25
minutes are possible. The delay between samples can be
controlled for each sample to allow for any desired sampling
schedule. Sampling duration can be controlled in 10-second
increments. The sampler is designed so that the sampling
pump operates for at least 1 minute to flush the sampling
lines prior to sample collection. (This occurs only when there
is a delay between samples of at least 1 minute; otherwise,
the sampler pump operates continuously and does not
require a flush period.) This operating feature will ensure
that there is no artifact associated with carryover from the
previous sample, which would be most significant if the room
concentration decreased rapidly. The sampler will be operated
so that a minimum of 250 cc will be collected, which will
allow for repeat analyses to be performed, if needed.
Sampler dimensions are 30 x 30 x 65 cm (w x h x 1) and the
sampler weighs 12 kg. The samplers are intended to sit on
the floor or desktop. The sample inlet tube can be secured
with a support rod and may be placed as high as 3 m above
the sampler. Indicator lights (LEDs) are visible on the top
of the control box. The LEDs indicate whether the pump
is operating and which sampling bag is being filled. These
provide the operator assurance of proper operation without
disturbing operation.
Since the tests may rely heavily on the automated sampler, a
rigorous battery of pretest checks will be performed. These
pretest checks are designed to verify that everything is in
working order prior to execution of the tracer gas test. Each
serialized automated sampler will execute a preprogrammed
sampling sequence without any sample bags attached. The
samplers will be monitored during the pretest to verify that
the pump is running and the valves are operating according
to the program. This monitoring will occur in the form of
checking the status of the LEDs, which indicate whether the
pump is running and which valves are actuated. In addition,
the flow rate for all sampling lines will be verified.
After pretest checks are performed on all of the automated
samplers, the samplers will be loaded with sample cartridges
containing twelve 500-cc sampling bags and then deployed to
the sampling locations.
SF6 Leakage Test Results
The results of the initial leakage tests suggested that
imperfect mixing would be one major discrepancy
between the experimental results and the model-predicted
concentration curves (see Figures A-4 and A-5). The results
of the initial leakage tests also indicated that significant
recirculation was occurring (see Figure A-4). Upon further
investigation it became apparent that significant leakage was
occurring across the recirculation damper and that reducing
that leakage to zero was not feasible given the materials
and time constraints of the project. While it was possible to
estimate the actual recirculation using the theoretical model
(see Figure A-5), this represented an indirect method of
estimating the interzonal leakage, which depended heavily
on the model-predicted value of the recirculated airflow.
For example, estimating the interzonal leakage while
neglecting the recirculation (as in Figure A-4) leads to an
interzonal leakage estimate of 0.4 ACH, while estimating
the recirculation (as in Figure A-5) leads to an interzonal
leakage estimate of 0.2 ACH. Given this strong dependence
of the estimated leakage rate on the estimated recirculation
rate, it was clear that this did not represent a direct measure
of the leakage, as planned. For this reason, a combinatorial
approach to determining the leakage was adopted.
-------
100000
10000
5- 1000
s
'I
100
0.01
Mixing Lag (Imperfect Mixing)
Lumped zone indicates recirculation is non-zero.
Experimental Zone of Release
Experimental Zone of Interest
Experimental Rest of Building
Theoretical Zone of Release
Theoretical Zone of Interest
—Theoretical Rest of Building
20 25
Time [Min]
Figure A-4. Comparison of Model-Predicted and Experimental Data for an Initial Leakage Test Assuming
No Recirculation. Experimental data were gathered for the "small" notional building under
a 6 ACH of makeup air. Model-predicted data are for the "small" notional building under 6
ACH of makeup air with 0.4 ACH interzonal leakage.
Mixing Lag (Imperfect Mixing)
100000
10000 -
51 1000 -
-•-Experimental Zone of Release
-•-Experimental Zone of Interest
-^-Experimental Rest of Building
— Theoretical Zone of Release
— Theoretical Zone of Interest
— Theoretical Rest of Building
Model-estimated recirculation is 1.5 ACH
0.01
20 25
Time [Mlii]
Figure A-5. Comparison of Model-Predicted and Experimental Data for an Initial Leakage Test Assuming
Limited Recirculation. Experimental data were gathered for the "small" notional building
under a 6 ACH of makeup air. Model-predicted data are for the "small" notional building
under 5 ACH of makeup air and 1.5 ACH recirculation with 0.2 ACH interzonal leakage.
-------
Appendix B
Preliminary Simulation Results
Filter Efficiency
Makeup Air
nfiltration
Leakage
Recirculation
234
Normalized Parameter [P/PBilse]
Figure B-l. Normalized Exposure at 30 Minutes (E30,2) Versus Model Input Parameters for a Zone 1
Release with a Zone 2 Parameter Scheme for a Building (V3=1V!). Note that P/Pbase=l
signifies the analysis baseline.
2.5 -
w
o
0.
X
111
2-
£
o
o
*-
«
F
•o
1.5 -
0.5 -
- Filter Efficiency
-Makeup Air
-Infiltration
-Leakage
-Recirculation
Normalized Parameter [P/Pe^se]
Figure B-2. Normalized Time to Critical Exposure (tct) Versus Model Input Parameters for a Zone 1
Release with a Zone 2 Parameter Scheme for a Building (V3=lVi). Note that P/Pbase=l
signifies the analysis baseline.
-------
Filter Efficiency
Makeup Air
Infiltration
Leakage
234
Normalized Parameter [P/PBi1se]
Figure B-3. Normalized Exposure at 30 Minutes (E30,2) Versus Model Input Parameters for a Zone 1
Release with a System Parameter Scheme for a Building (V3=1V!). Note that P/Pbase=l
signifies the analysis baseline.
Filter Efficiency
Makeup Air
Infiltration
Leakage
Recirculation
Normalized Parameter [P PB ,;e]
Figure B-4. Normalized Time to Critical Exposure (tct) Versus Model Input Parameters for a Zone 1
Release with a System Parameter Scheme for a Building (V3=1V!). Note that P/Pbase=l
signifies the analysis baseline.
-------
-Filter Efficiency
-Makeup Air
-Infiltration
-Leakage
-Recirculation
123456
Normalized Parameter [P/PBilse]
Figure B-5. Normalized Exposure at 30 Minutes (E30,2) Versus Model Input Parameters for a Zone I
Release with a Zone 2 Parameter Scheme for a Building (V3=5Vi). Note that P/Pbase=l
signifies the analysis baseline.
Filter Efficiency
Makeup Air
Infiltration
Leakage
Recirculation
234
Normalized Parameter [P/PB,ise]
Figure B-6. Normalized Time to Critical Exposure (tct) Versus Model Input Parameters for a Zone 1
Release with a Zone 2 Parameter Scheme for a Building (V3=5V1). Note that P/Pbase=l
signifies the analysis baseline.
-------
Filter Efficiency
Makeup Air
Infiltration
Leakage
^^Recirculation
234
Normalized Parameter [P/PB-1se]
Figure B-7. Normalized Exposure at 30 Minutes (E30?2) Versus Model Input Parameters for a Zone 1
Release with a System Parameter Scheme for a Building (V3=5Vi). Note that P/Pbase=l
signifies the analysis baseline.
*[ 2.5 -\
I
I 2H
a-
o
o
0.5 -
- Filter Efficiency
-Makeup Air
-Infiltration
-Leakage
-Recirc illation
0123456
Normalized Parameter [P/PB,ise]
Figure B-8. Normalized Time to Critical Exposure (tct) Versus Model Input Parameters for a Zone 1
Release with a System Parameter Scheme for a Building (V3=5V1). Note that P/PbaSe=l
signifies the analysis baseline.
-------
-Filter Efficiency
-Makeup Air
-Infiltration
-Leakage
-Recirculation
234
Normalized Parameter [P/PBase]
Figure B-9. Normalized Exposure at 30 Minutes (E30,2) Versus Model Input Parameters for a Zone 1
Release with a Zone 2 Parameter Scheme for a Building (V3=10V1). Note that P/Pbase=l
signifies the analysis baseline.
Filter Efficiency
Makeup Air
Infiltration
Leakage
Recirculation
234
Normalized Parameter [P/PB^]
Figure B-10. Normalized Time to Critical Exposure (tct) Versus Model Input Parameters for a Zone 1
Release with a Zone 2 Parameter Scheme for a Building (V3=10V1). Note that P/PbaSe=l
signifies the analysis baseline.
-------
-*-Filter Efficiency
-•-Makeup Air
-*- Infiltration
-•-Leakage
^t-Recirculation
234
Normalized Parameter [P/PBase]
Figure B-ll. Normalized Exposure at 30 Minutes (E30 2) Versus Model Input Parameters for a Zone 1
Release with a System Parameter Scheme for a Building (V3=10V1). Note that P/PbaSe=l
signifies the analysis baseline.
Filter Efficiency
Makeup Air
Infiltration
Leakage
Recirculation
234
Normalized Parameter [P/PB,ise]
Figure B-12. Normalized Time to Critical Exposure (tct) Versus Model Input Parameters for a Zone 1
Release with a System Parameter Scheme for a Building (V3=10V1). Note that P/PbaSe=l
signifies the analysis baseline.
-------
Filter Efficiency
Makeup Air
Infiltration
Leakage
Recirculaflon
234
Normalized Parameter [P/PBase]
Figure B-13. Normalized Exposure at 30 Minutes (E30 2) Versus Model Input Parameters for a Zone 1
Release with a Zone 2 Parameter Scheme for a Building (V3=25V1). Note that P/PbaSe=l
signifies the analysis baseline.
2.5 -
x
ill
"55
u
£
O
-------
Filter Efficiency
Makeup Air
Infiltration
Leakage
Recirculation
234
Normalized Parameter [P/PBlse]
Figure B-15. Normalized Exposure at 30 Minutes (E30 2) Versus Model Input Parameters for a Zone 1
Release with a System Parameter Scheme for a Building (V3=25V1). Note that P/PbaSe=l
signifies the analysis baseline.
Filter Efficiency
Makeup Air
Infiltration
Leakage
Recirculation
234
Normalized Parameter [P/PB,ise]
Figure B-16. Normalized Time to Critical Exposure (tct) Versus Model Input Parameters for a Zone 1
Release with a System Parameter Scheme for a Building (V3=25Vi). Note that P/Pbase=l
signifies the analysis baseline.
-------
-Filter Efficiency
-Makeup Air
-Infiltration
-Leakage
-Recirculalon
0123456
Normalized Parameter [P/PBflse]
Figure B-17. Normalized Exposure at 30 Minutes (E30 2) Versus Model Input Parameters for a Zone 1
Release with a Zone 2 Parameter Scheme for a Building (V3=50V1). Note that P/PbaSe=l
signifies the analysis baseline.
-•- Filter Efficiency
-•-Makeup Air
-*-Infiltration
-•-Leakage
^^Recirculation
234
Normalized Parameter [P/PB;ise]
Figure B-18. Normalized Time to Critical Exposure (tct) Versus Model Input Parameters for a Zone 1
Release with a Zone 2 Parameter Scheme for a Building (V3=50Vi). Note that P/Pbase=l
signifies the analysis baseline.
-------
Filter Efficiency
Makeup Air
Infiltration
Leakage
Redrculaton
234
Normalized Parameter [P/PBlse]
Figure B-19. Normalized Exposure at 30 Minutes (E30,2) Versus Model Input Parameters for a
Zone 1 Release with a System Parameter Scheme for a Building
Note that P/Pbase=l signifies the analysis baseline.
- Filter Efficiency
-Makeup Air
-Infiltration
-Leakage
-Recirculation
123456
Normalized Parameter [P/PF,, P]
Figure B-20. Normalized Time to Critical Exposure (tct) Versus Model Input Parameters for a
Zone 1 Release with a System Parameter Scheme for a Building (V3=50V!). Note
that P/Pbase=l signifies the analysis baseline.
-------
-Filter Efficiency
-Makeup Air
-Infiltration
-Leakage
-Redrculation
123456
Normalized Parameter[P/PBilse]
Figure B-21. Normalized Exposure at 30 Minutes (E30,2) Versus Model Input Parameters for a Zone 1
Release with a Zone 2 Parameter Scheme for a Building (V3=lVi). Note that P/Pbase=l
signifies the analysis baseline.
2.5 -
Q.
X
111
o
o
N
W 0.5 ^
- Filter Efficiency
-Makeup Air
-Infiltration
-Leakage
-Redrculation
0123456
Normalized Parameter [P/PB.IS*]
Figure B-22. Normalized Time to Critical Exposure (tct) Versus Model Input Parameters for a Zone 1
Release with a Zone 2 Parameter Scheme for a Building (V3=lVi). Note that P/Pbase=l
signifies the analysis baseline.
-------
-Filter Efficiency
-Makeup Air
-Infiltration
-Leakage
-Re circulation
1 23456
Normalized Parameter [P/PBl1Se]
Figure B-23. Normalized Exposure at 30 Minutes (E30 2) Versus Model Input Parameters for a Zone 1
Release with a System Parameter Scheme for a Building (V^IV^. Note that P/PbaSe=l
signifies the analysis baseline.
s
-£ 2.5 -
2
1 2^
w
o
O
O
1.5 -
i_
O
0.5 -
- Filter Efficiency
-Makeup Air
-Infiltration
-Leakage
-Recirculation
0123456
Normalized Parameter [P/Pt,, *]
Figure B-24. Normalized Time to Critical Exposure (tct) Versus Model Input Parameters for a Zone 1
Release with a System Parameter Scheme for a Building (V3=1V!). Note that P/Pbase=l
signifies the analysis baseline.
-------
Appendix C
Data Quality
Data Quality
Data Quality Objectives (DQOs) are qualitative and
quantitative statements designed to ensure that the type,
quality, and quantity of data used are appropriate for the
intended application. As discussed in the body of the report
(Sections 1, 2, 3, and 5), the experiments performed had
two purposes. One purpose of the experiments was to
evaluate the effect of HVAC, architectural, and operating
procedure modifications on the spread of gases and aerosols
through a building. The other purpose was to experimentally
demonstrate the validity of the three-zone, well-mixed model
developed in Task 7. In an effort to collect a wide range of
data, replicate testing was not performed under this project;
therefore, it was not possible to quantitatively state the
required agreement between replicates.
The type of data being collected during these tests, paniculate
concentrations, are appropriate since they can be compared
to one another to determine the effect of the modification
on the spread of gases and aerosols. A qualitative analysis
of the collected data was performed by comparing these
data to those collected on similar projects and to the results
of the Task 7 theoretical analyses to confirm the trends
noted in the Task 7 results. For example, a short-term or
instantaneous release in a well-mixed volume zone serviced
by a typical HVAC system would be expected to produce
a rapid increase in contaminant concentration, followed by
a logarithmic decay in the concentration. The height of the
peak is indicative of the ratio of the mass released to the
zone volume, while the decay is indicative of the removal
constant. The removal constant is an aggregation of many
potential factors, including contaminant decay (if applicable),
particle deposition (if applicable), filtration (if applicable),
and exhaust (if applicable).
Engineering logic was also used to compare the data sets to
each other to determine whether or not they made qualitative
sense. (For example, one would expect a lower paniculate
concentration when a higher efficiency filter was used in the
HVAC system.) Although some variation is expected, gross
deviations from past data and logically expected results were
immediately investigated as they were indicative of problems
that arose. An excellent example of this was observed in the
SF6 leakage test results discussed in Appendix A. The results
of the SF6 leakage tests showed that the building HVAC
system was unable to operate with 100% makeup air (i.e., no
recirculated air). This prompted a physical inspection of air
handling dampers in which a test technician climbed inside
of the air handling unit, revealing that significant leakage
was occurring across the recirculation damper. Another
example of the use of engineering logic in achieving data
quality objectives is apparent in the discussions of MetOne
particle counter saturation in Section 7.0. In this example,
experimental measurements indicated that the release mass
used was sufficiently large to cause saturation of the MetOne
optical particle counter in the zone of release. In this case,
an engineering judgment was made not to reduce the release
mass in order to maintain a sufficient particle concentration
in zones far from the release. This judgment was deemed
appropriate for two reasons. The first reason was that the
concentration within the release zone exhibited the expected
logarithmic decay from the saturation concentration within
the test timeframe, indicating a high likelihood that the
particle concentration within the release zone was behaving
as expected. The second reason was that the concentration
in zones far from the release was sufficiently above baseline
noise to prove useful to model comparisons. Furthermore,
given that subsequent planned tests would use higher
efficiency filtration, it was the judgment of the test leader
that reducing the release mass would result in concentrations
below background in zones far from the release. In this
manner, engineering logic was used in conjunction with
qualitative analyses to scrutinize the quality of data sets,
make data-based test decisions, and achieve the data quality
objectives of this task.
Sample calculations are detailed throughout the body of the
report, particular in Sections 7, 8, and 9.
-------
$EPA
United States
Environmental Protection
Agency
PRESORTED STANDARD
POSTAGES FEES PAID
EPA
PERMIT NO. G-35
Office of Research and Development
National Homeland Security Research Center
Cincinnati, OH 45268
Official Business
Penalty for Private Use
$300
Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free
------- |