Summary of paperONE (Michael et al., 2006)

Understanding the dynamic behavior of indoor aerosols is essential for accurately predicting their concentrations and fates within a building, and for estimating human exposures. (Michael et el., 2006 (this paper))
Processes such as coagulation, deposition, and removal by indoor filtration can depend strongly on the particle size distribution of the species, and these processes can affect the overall airborne concentration in buildings.
The particle size distribution is also an important element in estimating the quantity and location of particle deposition in the lung.
Mathematical Eq(s). to express these process in building, and computer software to solve them have been developed and applied successfully [for predicting aerosol concentrations in various indoor systems]

1. Nazaroff and Cass, 1989:
The aerosol dynamics model (ADM):
*simulates the evolution of particle size distributions, using a multi-component sectional representation
*utilizes a multi-zone representation of a building. [each zone is considered independently well-mixed]]
*equations are written based on the principle of mass conservation (solve numerically)

An important component of ADM is the explicit incorporation of particle size distribution information.

Particle size governs almost all particle behavior (Hinds, 1982), including:
# the removal efficiency of an air filter,
# the rate of deposition onto surfaces,
# the coagulation of small particles with larger ones, and
# the retention efficiency of the lung upon inhalation.

The ADM accounts for many factors, including:
@. Direct ETS particle emissions,
@. Inter-zonal mixing,
@. Ventilation,
@. Filtration,
@. Coagulation, and
@. Deposition onto surfaces.

But The ADM not included accounts:
¨ Evaporation,
¨ Condensation, and
¨ Homogenous nucleation.



COMIS:

CONTAM:

CONTAM (Dols, W. Stuart. Walton, G. N. (BUILDING ENVIRONMENT DIVISION - 863)
CONTAMW 2.0 User Manual (NISTIR 6921) - November 01, 2002 )

This manual describes the computer program CONTAMW version 2.0 developed by NIST. CONTAMW is a multizone indoor air quality and ventilation analysis program designed to help you determine: airflows and pressures – infiltration, exfiltration, and room-to-room airflows and pressure differences in building systems driven by mechanical means, wind pressures acting on the exterior of the building, and buoyancy effects induced by temperature differences between the building and the outside; contaminant concentrations – the dispersal of airborne contaminants transported by these airflows and transformed by a variety of processes including chemical and radio-chemical transformation, adsorption and desorption to building materials, filtration, and deposition to building surfaces; and/or personal exposure – the prediction of exposure of building occupants to airborne contaminants for eventual risk assessment. CONTAMW can be useful in a variety of applications. Its ability to calculate building airflows and relative pressures between zones of the building is useful for assessing the adequacy of ventilation rates in a building, to determine the variation in ventilation rates over time, to determine the distribution of ventilation air within a building, and to estimate the impact of envelope airtightening efforts on infiltration rates. The program has been used extensively for the design and analysis of smoke management systems. The prediction of contaminant concentrations can be used to determine the indoor air quality performance of buildings before they are constructed and occupied, to investigate the impacts of various design decisions related to ventilation system design and building material selection, to evaluate indoor air quality control technologies, and to assess the indoor air quality performance of existing buildings. Predicted contaminant concentrations can also be used to estimate personal exposure based on occupancy patterns. Version 2.0 contains several new features including: non-grace contaminants, unlimited number of contaminants, contaminant-related libraries, separate weather and ambient contaminant files, building controls, scheduled zone temperatures, improved solver to reduce simulation times and several user interface related features to improve usability.

Keywords: airflow analysis , building technology , computer program , contaminant dispersal , controls , indoor air quality , multizone analysis , smoke control , smoke management

Point2: Coupled airflow and aerosol transport

We use the COMIS airflow model (Feustel, 1999)
to predict the airflows between rooms, and between indoors and outdoors.
COMIS predicts the steady-state flow of air (las: airflow) induced by:
> wind,
> thermal buoyancy,and
> mechanical ventilation
by representing a building:
$ as a collection of zones,
$ connected by flow paths (such as cracks, doors and windows, and ductwork).

Assumptions:
1. Air=incompressible fluid flow,
2. Airflow (through these pathways) is calculated by
balancing pressure differences between the zones.

Feustel (1999) and Lorenzetti (2002) describe the mathematical foundations of the COMIS model.

COMIS has been applied to predict airflow and gas transport:
@ in residences (Feustel et al., 1985; Sextro et al., 1999),
@ in small office buildings (Feustel, 1990),
@ in controlled experimental test houses (Haghighat and Megri, 1996), and
@ in single-family houses (Haghighat and Megri, 1996, Zhao et al., 1998).

OUTPUT from COMIS: airflows. (Michael et al., 2006)this paperONE)
Air flows between rooms and across the building envelope for:
& every user-defined building operating mode, and
& meteorological condition.

Then Michael et al. (2006) use the MIAQ4 aerosol dynamics software based-on
(Nazaroff and Cass, 1989):
@ to predict the size-resolved transport, and
@ to predict the evolution of particle concentrations
prompted by the COMIS-calculated airflows,
and directed by particle dynamics behavior such as
>gravitational settling,
>coagulation, and
>thermal diffusion.

MIAQ4 simulates a size- and chemically-resolved particle size distribution.
It does not take into account:
* evaporation,
* condensation, or
* homogeneous nucleation.

The aerosol model was originally developed for and applied to predicting the behavior of:
1. particles from cigarette smoke in a chamber(Nazaroff and Cass, 1989) and
2. particulate matter (PM) in museums (Nazaroff et al., 1990).

The COMIS&MIAQ4 are linked in a feed-forward manner.(Las: in--->COMIS---(Perl)--->MIAQ4--->out (mungkin dapat di umpanbaikkan ke sisi in dari COMIS))
The airflows predicted by COMIS serve as inputs to MIAQ4.
Feedback from MIAQ4 to COMIS is unnecessary since:
<> the total airflow mass is much greater than the pollutant mass, and
Feedback from MIAQ4 to COMIS can thus be ignored in the airflow mass balance equations.

We linked the COMIS&MIAQ4 by writing a computer program that transforms output from COMIS into MIAQ4.
We wrote the linking software in the Perl scripting language:
? because PERL contains several built-in functions for formatting text and numbers, and
? because PERL is available for most computer systems.

Since linking is in a feed-forward manner, the Perl script first runs COMIS for an entire simulation for:
# all HVAC operations, and
# meteorological conditions of interest.
Perl then runs MIAQ4:
% in intermediate steps,
% halting the simulation (to readjust the flows when changes in airflow or temperature conditions occur), and
% restarting the simulation with the new state of the pollutant mass transport or loss.

3. Application COMIS&MIAQ4:
!. predict ETS particle transport in a three-room experimental chamber, and
!. compared predictions to data.

Apte et al. (2004) conducted tracer and ETS experiments in a full-scale, three-room, laboratory chamber. (as in Fig. 1)

Side-stream smoke was produced by machine-smoked cigarettes in Room 2 for approximately 8 min, while mainstream smoke was vented outside.

Sulfur hexafluoride (SF6) was injected simultaneously into Room 2 as a tracer gas.

Small mixing fans were running in each room at all times to increase well-mixed conditions.
Air sampling tubes were installed in each room to draw air from the chamber to external analytical equipment.

For our purposes here, we chose an experiment where:
@ the door between Rooms 1 and 3 was open fully and
@ the door between Rooms 2 and 3 was partially open (0.0254 m).
Both doors were standard height (2.12 m).
The temperature difference between rooms varied from 0 to 1 degree of Celcius during the experiment.

>Gas- and particle-phase ETS tracer concentrations,
>ETS particle mass, and
>particle size distributions in each room were measured as a
function of time.

3.1. Tracer gas and ETS particle characterization

Data consisted of time-series and point measurements of:
(1) tracer gas concentrations,
(2) total ETS mass concentration
(3) size-resolved particle concentrations and
(4) room air temperatures.

SF6 tracer gas was measured using a gas chromatograph (GC) with an electron capture detector (Hewlett Packard Model 5890).
Air was continuously drawn from each room via 3-mm-ID copper tubing (~1.8m from the floor) at 1 L/min.
To avoid lag time in the sample line, the gas sampling ports were sampled continuously, and vented out of the building when not being measured by the GC (every 4 min).

Total ETS particle mass was measured gravimetrically
(Cahn Model 21 Automatic Electrobalance, with 0.1-microgram resolution) by particle collection on precleaned (Sohxlet extration) and pre-weighed 47-mm-diameter Teflon-coated glass fiber filters (Fiberfilm T60A20, Gelman/Pallflex).

Filters in each room were connected to sampling pumps located outside the chamber.
Due to the low air exchange rate of the environmental chamber and to the very infrequent opening of the space, the infiltration of
ambient PM into the chamber was negligible.
Thus, the dominant source of PM in the chamber was the ETS generated during experiment.
For this reason, it was not necessary to use a size selective inlet for particle sampling in the chamber experiments.
The filter samples contained only respirable suspended PM (RSP) with a maximum particle aerodynamic diameter less than 1.5 mm.

Filter samples were each taken for 30 min at 3, 6 and 24 h during the experiment.
As discussed below, the intermittent operation of the particle-sampling pumps led to enhanced ventilation air flows in the
chamber for which we needed to explicitly account for in the modeling.

In addition, size- and time-resolved particle concentration measurements (pengukuran konsentrasi partikel dalam waktu [konsentrasi partikel sebagai fungsi waktu] dan ukuran partikel dalam waktu [ukuran partikel sebagai fungsi waktu]) were provided by:
#1. a Differential Mobility Particle Sizer (TSI/Classifier 3071 Ultrafine Condensation Particle Counter),
-->measure particle size diameters from 0.01 to 0.45 micrometer in diameter]
-->provide data from each room every hour, and
#2. an optical particle counter (LAS-X OPC),
-->measured particle diameters from 0.09 to 43.5 micrometer.
-->provide data from each room every 3 min.

These instruments (DMPS & OPC) all sampled from a continuously flowing sampling manifold connected to the center of each room.

3.2. Model-measurement comparisons

We developed a COMIS model of:
* the three-room chamber with room dimensions,
* the size of door openings, and
* room temperatures (as a function of time) as model inputs.

Because the experiments were run with the doors between the rooms open, we did not incorporate any added room-to-room leakages
(e.g. cracks) in the model.
We did include the air leakage between the chamber and the outside in the model, which we determined to be approximately 0.01 air changes per hour from tracer gas decay rate measurements, when the SF6 and DMPS/OPC sampler pumps were operating, but not the pumps for the open face filter samples.

We distributed this leakage uniformly across the outer walls of the
rooms.

Fig. 1 shows the predicted airflows at one instant in time during the experiment when the filter pumps were not running.
The airflows change moderately over time due to changing room temperatures and, as we describe later, due to the intermittent operation of the pumps for the aerosol filter samples. The airflow between Rooms 2 and 3 is significantly lower then the flow between Rooms 1 and 3 because the door between Rooms 2 and 3 is only partially open.

We used the airflow calculations to predict the dispersion of a puff release of 0.01 g of SF6 in Room 2 and compared the COMIS predictions
to measurements (Fig. 2). The COMIS model was not calibrated to the data; the input parameters, including size of the door opening
between Rooms 2 and 3, were measured independently.
Fig. 2 shows model-to-data comparisons with and without including the intermittent operation of the pumps to collect the total mass ETS particle samples at 3, 6 and 24 h. The figure shows that the
sampling pumps, though the amount of air removed is small (4.5–4.8m3 h1 for 30 min), increase the ventilation rate of the chamber, causing an appreciable reduction in the observed concentration of SF6
in the air. Because the overall leakage in the chamber is low, the pumping is a significant driver for air leakage when the pumps are operating. The inclusion of the air sampling pumps reduced the
root-mean-squared error (RMSE) when comparing the predictions to the experimental data by 76% in zone 1 and 60% in zone 2. We also tested the sensitivity of the model to the size of the door opening between Rooms 2 and 3 in other model runs, and found openings larger or smaller than the actual measured opening size produced inferior
matches to the data.
The model-to-data comparison for Room 3 is
similar to the comparisons for Room 1 because the
door between them is completely open, and
temperature differences between the rooms, though
small, are predicted to generate large inter-room
airflow, and thus mixing, between them. These
model predictions are, therefore, also consistent
with the data.
We next predicted the dispersion of ETS particles
from smoking one cigarette in Room 2. Inputs to
the MIAQ4 model were the airflow conditions predicted by COMIS, with inclusion of the intermittent
use of the filter pumps, the chamber
dimensions, measured temperatures as a function
of time, and an emission rate profile for side-stream
ETS particles, which we specified independently of
the data from this experiment (Nazaroff et al., 1993)
(Fig. 3). We also specified the turbulence intensity
factor for the chamber, which describes the streamwise
velocity gradient at the vicinity of the chamber
wall (Crump and Seinfeld, 1981). In this application,
we adjusted the turbulence intensity factor to
calibrate the model predictions to the overall
airborne ETS concentration data. We selected an
intensity factor of 1.2 s1, which is consistent with values reported by Furtaw et al. (1996) and Lai and
Nazaroff (2000).
Fig. 4 shows the measured and modeled concentration
of total ETS particles in Rooms 1 and 2, as
measured by the filter samples and by the DMPS.
The slopes of the decay curves are fairly consistent
with the measured data, suggesting that, in general,
MIAQ4 is properly predicting the transport and
losses due to deposition and surface diffusion.
However, Fig. 4 shows predicted particle concentrations,
in general, lower than were measured
(R2 ¼ 0.93, RMSE ¼ 52 mgm3). Since the SF6
dispersion predictions agree well with measurements,
it is unlikely that the MIAQ4 is incorrectly
predicting airflows. It is more likely that the actual
ETS released during the experiment was larger than
we assumed for input into the model (Fig. 3) since
the total side-stream particle mass emitted from a
cigarette can vary from one experiment to the next
(e.g., see Fig. 12 in Apte et al.). Because of this
variability, however, estimating an experimentspecific
mass emission entails back estimating the
mass based on model-to-data comparisons. We
chose not to do this calibration since it diminishes
the persuasiveness of the model-data comparisons.
It is also possible that the turbulence intensity factor
chosen for the model was not correct, thus underpredicting
particle deposition onto surfaces, and, therefore, underestimating these losses. Further
experiments and modeling will be needed to
determine the primary causes of the differences,
but such analyses were beyond the scope of this
demonstration.
Figs. 5 and 6 shows MIAQ4-predicted concentrations
of ETS in Room 2 resolved by particle size
compared to DMPS measurements at 40, 160, 460,
and 640 min after the cigarette was smoked. Figs. 7
and 8 shows model-to-data comparisons for Room
1. In general, the overall shapes of the model
predictions agree well with the data. The difference
between modeled and measured binned particle
concentrations is consistent with the discrepancies
in total concentrations shown in Fig. 4. The
size-resolved plots also show that relative differences
between the measured and modeled are
small and that across particle size and concentration
no consistent biases exist in the model’s
predictions.

4. Discussion and concluding remarks

We have linked the COMIS-multizone airflow model with the MIAQ4-aerosol dynamics model:
@ to predict size-resolved aerosol transport in multiroom buildings.
-->Though both models have been reported and demonstrated individually in the
literature, they have not been applied together to predict size-resolved aerosol transport in multiroom buildings, or compared to real data.
-->The linked model will be a useful tool for examining the behavior of aerosols in multizone buildings and predicting concentrations and exposures as a function of particle size. The latter may be especially important for evaluating different strategies for reducing aerosol exposures within buildings.
@can also be used to aid in the design and interpretation of field experiments.
As proof of concept, we applied the models to predict the transport and behavior of tracer gas
and ETS particles measured in a three-room chamber. We obtained excellent agreement between the predicted and observed tracer gas concentrations in all three rooms. The predictions of both ETS particle mass and size-resolved particle
concentrations also agreed well with the measurements.

COMIS was particularly helpful for estimating the inter-room airflows caused by operating the
filter-sampling pumps, and therefore reduced the number of experiments needed to characterize the airflows.


From Literatures Study More:
Apte et al. (2004)
Test space description
Chamber layout and construction
A 50 m3 multizone environmental chamber was constructed within a temperature controlled single-story building at Lawrence Berkeley National Laboratory (LBNL). The chamber was designed to mimic conditions of a multi-room residential or office building where ETS might be generated in one room and transported to others. The layout of the chamber, shown in Figure 1, consisted of three rooms, a smoking room (SR), a connecting corridor (COR), and a non-smoking room (NSR). The chamber was built using wood frame construction with taped and painted gypsum wallboard walls and ceiling, and a high-quality nylon carpet laid over plywood sub-flooring in all three
chamber rooms. The entry and interconnecting doors were standard solid core wood design; however, magnetic refrigerator door seals (and accompanying steel flanges for the door openings) were added to the entire door perimeters to ensure near airtight sealing when the doors were fully closed. Low volatile organic compound (VOC) emitting
paints and sealants were used throughout the chamber in order to minimize the buildup of unwanted VOCs in the chambers. A plastic (PVC) membrane vapor barrier was placed behind the gypsum wallboard on the interior walls between the SR and the other two
chambers to retard any diffusion of ETS components through the wall materials between rooms. Fully closed, the baseline air exchange rate (λv) of the chamber and its composite sub-rooms was approximately 0.01 h-1.

Four 10- cm-diameter axial mixing fans were placed in each room, mounted at a height of approximately 1.5m above the floor, at about 1.5m along the diagonals between corners. The fan axes were horizontal and were oriented in opposing directions in order to
enhance the mixing within each room. The fan speed was controlled using a Variac transformer at a speed just high enough to achieve uniform mixing of gases, based upon previous chamber experiments.
Each chamber room was equipped with a set of small- and large-diameter sampling ports, each with a plug to close the ports not in use. The large sampling ports were just large enough to permit the insertion of a 47mm particle sampling filter holder, while the small
ports allowed for the insertion of nicotine sampling sorbent tubes into the chamber.

http://continuousimprovement.blogsome.com