Content uploaded by Jessica Few
Author content
All content in this area was uploaded by Jessica Few on Mar 31, 2020
Content may be subject to copyright.
Measuring the ventilation rate in occupied buildings and
adapting the CO2 tracer gas technique
Jessica Few*1 and Clifford Elwell1
UCL Energy Institute
London
WC1H 0NN, UK
* Corresponding author: jessica.few.16@ucl.ac.uk
ABSTRACT
Measuring ventilation rates in occupied dwellings is challenging but represents the conditions that occupants
experience. This paper explores the constraints of existing methods when measuring the ventilation rate of
occupied buildings and proposes a new method addressing some of them.
Ventilation rates in occupied buildings can change over short time scales due to changes in weather or window
and door positions. PFT based methods measure average ventilation over an extended period. Similarly, an
annualised ventilation rate based on the ‘20th rule of thumb’ can be approximated using pressurisation tests. It is
unclear how the results relate to the ventilation rate under any specific conditions. Conversely, measurement
periods of less than an hour are possible with tracer gas techniques.
The spatial scale over which ventilation is measured is important. Whole dwelling estimates are obtained from
pressurisation tests and tracer gas methods if the whole dwelling is dosed. In both cases internal doors must be
open, affecting the airflow, which may or may not reflect the occupied configuration. Differences between whole
dwelling and in-use ventilation could be important for IAQ and heat loss. Single zone tracer gas techniques have
been applied to one room and assumed to represent the whole dwelling. Multi-zone methods using several different
gases are not usually applied in occupied buildings because of the complexity of the experimental set-up.
Measurement techniques must be acceptable to occupants. Pressurisation tests are minimally disruptive, taking
less than an hour. PFT measurements are unobtrusive, only requiring small sources and sorption tubes. Injection
of tracer gases, such as SF6, and concentration measurement can require pumped gas sampling which is noisy and
bulky, causing inconvenience to the occupants. Metabolic CO2 based methods do not require the injection of gas
or pumped sampling and may be more acceptable to occupants.
CO2 decay techniques require knowledge of occupancy; historically this has relied on occupant reports, assumed
hours of occupancy, or hand-picked sections of data. Similarly, CO2 equilibrium or accumulation techniques
require knowledge of the CO2 generation rate and occupant activities.
A new approach has been developed to address some of these issues, based on metabolic CO2 tracer gas decay.
An algorithm for identifying when a dwelling is occupied has been developed, agreeing with reported occupancy
in 87% of cases, meaning that large volumes of data can be analysed automatically. CO2 is measured in each room
every 5 minutes, meaning that sub-hourly variations in ventilation rate, and variation between rooms, can be
explored. Proximity sensors are used to monitor windows and doors, so that ventilation can be calculated during
periods with constant conditions, and variation due to different configurations can be investigated. The proposed
method represents a step towards appropriate measurement of ventilation and its variation in occupied buildings.
KEYWORDS
Ventilation, ventilation measurement, tracer gas, occupancy, occupied buildings
1 INTRODUCTION
Through the Paris Agreement, 185 countries have agreed to make efforts to tackle climate
change (UNFCCC, 2015). Limiting excess ventilation improves the thermal efficiency of
buildings; ventilation can account for 50% of the primary energy used in a UK dwelling
(CIBSE, 2015). However, sufficient ventilation is required to maintain adequate indoor air
quality (IAQ) in occupied buildings (Persily, 2006). Poor IAQ has been associated with various
health implications, including: asthma, allergies, cancers, ‘sick building syndrome’ and
respiratory tract infections (Sundell et al., 2011). Fisk (2018) reviewed the literature and
concluded that there is likely to be an association between low ventilation rates and poor health
outcomes. Additionally, the ventilation rate that the occupant experiences is likely to affect their
thermal comfort and use of air conditioners (Iwashita and Akasaka, 1997).
Occupants can significantly influence the ventilation rate in their homes through their practices
of window and door opening (external and internal) and use of mechanical ventilation
(Kvisgaard and Collet, 1990). However, characterising the ventilation occupants experience is
challenging due to the spatial and temporal variability of ventilation rate and the acceptability
of the technique to the occupants. This can mean that it is difficult to understand the ventilation
rates that the occupants are likely to experience.
This paper explores some of the challenges of measuring ventilation in occupied buildings, and
suggests a method which addresses some of them. The following section reviews some of the
issues with current ventilation measurement methods when used in occupied buildings, Section
3 introduces the proposed method, and describes the algorithm that has been developed to
determine when a building is occupied, and the results of combining the measurement method
and the occupancy algorithm. Section 4 discusses the strengths and weaknesses of the proposed
method. Finally, Section 5 provides the conclusions and implications.
2 VENTILATION MEASUREMENT IN OCCUPIED BUILDINGS
Measurement of ventilation in occupied buildings can be complicated by factors including the
weather, its spatial variability and the influence of occupants. These aspects are reviewed in
relation to the most common methods used to characterise ventilation in this section.
2.1 Time Scale of Measurement
The ventilation rate estimated from a measurement is specific to the time for which the
measurement took place. However, the weather conditions and building configuration will vary
in occupied dwellings, thus the estimated ventilation rate only holds for the conditions during
which the measurement took place (Persily, 2016). For example, changes in window or internal
door position may alter the ventilation rate significantly, weather conditions particularly affect
naturally ventilated buildings. Measurements taken in a specific configuration cannot easily be
extrapolated to different conditions.
Methods to estimate ventilation may be broadly categorised as those which provide an average
rate over a long duration (days to months), and those that provide a shorter-term ‘snapshot’ of
the ventilation rate. Those methods estimating a long term ventilation rate are valuable in
estimating the overall conditions in the space, but cannot distinguish the impact of factors such
as door and window opening. This combined with the intermittent occupation of specific spaces
by occupants means it can be challenging to interpret occupant exposure to pollutants using
these methods. The physical interpretation of the calculated rate is also unclear in using these
methods (Sherman, 1990). Perflourocarbon tracer gas methods (PFT) are commonly used to
measure the average tracer gas concentration over varying time scales, examples range from
two days (Bornehag et al., 2005) to a month (Bekö et al., 2016). In analysis of this data a
constant ventilation rate over the period of measurement is assumed.
Another common method for estimating the average ventilation of a building is the ‘rule of
thumb’ that the air change rate measured at 50 Pa divided by 20 is a rough estimate of the
annualised air change rate under ‘normal’ conditions. Sherman (1998) stresses the limitations
of this given the influence of variations in weather conditions and dwelling characteristics.
Liddament (1996) suggests this can be particularly problematic for naturally ventilated
buildings as the instantaneous ventilation rate can deviate significantly from the average.
Other methods provide a ‘snapshot’ of the ventilation rate in a space and specific configuration
over a limited duration (minutes or hours). These may be useful to identify exposure to
pollutants under specific conditions, but give relatively little information about general
ventilation in a building. They are challenging to interpret if information about the weather,
doors and windows is missing (Persily, 2006). Tracer gas methods may be used to estimate the
ventilation rate from the rate of release required to produce constant concentration, from its rate
of accumulation given a known rate of release, or from its rate of decay. Tracer gas
concentration is usually monitored over approximately 1-2 hours, assuming the ventilation rate
to be constant over this period. Compared to the long term techniques above, the ventilation
rate is much more likely to be stable on this shorter time scale. However, changes in doors,
windows or operation of a ventilation system could still alter the ventilation rate significantly
during the measurement period and the contextual information required to interpret such
ventilation rates is considerable. Additionally, the relation of such estimated ventilation rates
to those in different building configurations and weather conditions is not clear and may require
extensive measurement and analysis to understand.
The number of measurement repeats carried out under different conditions varies considerably
in the literature. Wallace, Emmerich and Howard-Reed (2002) carried out ventilation
measurements every 2 to 4 hours in an occupied house over a year. They recorded weather
conditions and fan use, but window use was inconsistently recorded and internal doors were
not mentioned. A measurement period this long is extremely unusual, most measurements of
ventilation rates using non-PFT tracer gas methods last less than a week (Guo and Lewis, 2007;
Sharpe et al., 2015; Keig, Hyde and McGill, 2016). Apart from Wallace, Emmerich and
Howard-Reed’s extremely extended study (whose method is unlikely to be acceptable to many
building occupants, see section 2.3), little research has been published addressing the variation
in ventilation rates of buildings, which may only be investigated using techniques requiring a
short measurement duration.
2.2 Spatial Scale of Measurement
Air flows within buildings may be complex; dependent on the configuration of walls, the
opening of internal doors and furniture. In addition, the air exchange with outdoors may be
through both planned and unplanned ventilation paths. As a result, ventilation rates of different
spaces within the building can be variable, due to the changes in door and window opening,
amongst other factors. The IAQ may therefore vary substantially across the spaces that are
occupied, resulting in an exposure to pollutants in individual spaces that may not reflect the
whole house ventilation, or that in different rooms (Persily, 2006). It can also impact the thermal
comfort of occupants and their consequent actions, for example in particularly still parts of a
building occupants may increase their use of cooling in the summer (Iwashita and Akasaka,
1997), or in particularly draughty parts of the building may increase their use of space heating
in the winter.
Some ventilation measurement methods are applied to the whole dwelling, such as the n/20 rule
of thumb, which assumes that the building can be adequately described as a single zone
(Sherman, 1998). Multiple blower doors can be used to characterise different parts of buildings,
but this is uncommon and air leakage between internal spaces is challenging to characterise.
Single-zone tracer gas experiments can also be used to estimate whole building ventilation rates.
A uniform concentration of tracer gas throughout the building is required (ASTM, 2011),
meaning that internal doors are opened, which may alter the conditions compared to those
experienced when the building is in use (Keig, Hyde and McGill, 2016). Fans are often used to
ensure that uniform concentrations are achieved. However, Liddament (1996) suggests that fans
should not be used if the aim of the measurement is to understand air quality, since areas of
poor mixing are important in this context.
Multi-zonal tracer gas analysis can be used to investigate the effect of interzonal flows (e.g.
Penman and Rashid, 1982; Smith, 1988; Harrje et al., 1990). However, the analysis and
experimental set-up is much more complex than for single zone measurements and this method
is rarely carried out (Persily, 2006).
The ventilation rate of single rooms is often used to provide insight into IAQ and the exposure
of occupants to pollutants. For example, Guo and Lewis (2007) and Sharpe et al. (2015)
measured a single room and assumed this to be representative of the whole building. However,
in such cases the airflow between internal spaces has not been accounted for and the resultant
estimate of the ventilation rate does not represent the indoor-outdoor ventilation (Persily, 2006).
The complexity of airflow through and between spaces in buildings, and the resultant
limitations in estimating and interpreting a ventilation rate to provide insight into the conditions
experienced by occupants, is challenging and depends on the desired insights of the study.
Averaging the ventilation over an entire building means that the ventilation the occupant
experiences is unlikely to be understood, given that occupants will tend to move around rooms
and close doors. However, it is technically challenging to adequately account for interzonal
flows such that measurements can be taken in the building configuration that the occupant
experiences.
2.3 Invasiveness of Equipment
The inconvenience associated with a measurement technique will likely influence how long
people will tolerate its presence in their building. This section briefly reviews the invasiveness
of different methods to estimate the ventilation rates of properties.
Pressurization tests can usually be completed in less than an hour and do not require any
equipment to be left in a building. During testing occupants cannot use an external door and the
test is noisy; however, they have been used in many studies of occupied buildings (e.g.
Oreszczyn et al., 2005).
Tracer gas methods are likely to vary in their acceptability to occupants. PFT equipment is
small and silent so may be acceptable to occupants. Use of a safe tracer gas is essential, and
using CO2 (particularly metabolically generated) is likely to be more acceptable than other
gases as it is naturally present in the air and does not involve the release of any gas for the
purpose of measurements – a key motivation for the development of this method by Penman
and Rashid (1982). CO2 can be measured using NIR sensors, these are not excessively large
and are silent so may be acceptable to occupants. By contrast pumped gas sampling requires
tubes to be distributed around the building, increasing the spatial and visual burden to the
occupant, as well as being noisy. Wallace, Emmerich and Howard-Reed (2002) used pumped
gas sampling for a year, but this research took place in the home of one of the authors;
recruitment of non-researcher participants may be challenging for an extended campaign with
this method.
In order to ethically conduct ventilation measurements in occupied buildings, participants must
be aware of any disruption likely to be caused, and must find this acceptable for the duration of
the research. Less invasive techniques may be acceptable to a greater proportion of people, for
example pressure testing may be more widely accepted than tracer gas experiments, but the
insights gained may be reduced.
2.4 Knowledge of Periods of Occupancy
It is often important for the interpretation and analysis of measurements to know the times a
building is occupied, this is essential for CO2 based methods. However, methods to determine
the occupancy status of dwellings during ventilation measurements have not been widely
published. Guo and Lewis (2007) suggest that the difficulty in accurately determining when a
dwelling is occupied is one of the reasons that there are few examples in the literature of the
use of metabolic CO2 as a tracer gas. In Guo and Lewis’ study decay periods identified using
an occupant-reported daily log-sheet; occupant diaries may not always be accurate and they
increase the burden of the occupant participating in the research (Bryman, 2004). Roulet and
Foradini (2002) monitored a single office-room and manually identified periods of decaying
CO2, implying that prolonged periods of decreasing CO2 can be interpreted as indicating that
there were no occupants present. However, this does not account for the possibility of leakage
between zones and manual
identification of unoccupied
periods is a laborious process for
long monitoring campaigns.
3 DEVELOPING A NEW
TECHNIQUE TO ESTIMATE
VENTILATION RATE
A new method has been
developed, which refines the
application of the CO2 tracer gas
decay method using metabolically
generated CO2. It is essential that
the periods when the building is
unoccupied are reliably known with this method; to achieve this state sensing of doors and
windows is combined with an algorithm to determine the occupancy state.
The state of doors and windows was measured using binary magnetic contact sensors which
record when a door or window is open or closed. Not only are these data required to determine
occupancy, they also aid understanding of the configuration of the building during the
unoccupied measurement. Additionally, measurements of the weather conditions and internal
temperatures support the interpretation of the results.
3.1 Developing an algorithm for determining building occupancy
As discussed in Section 2.4, the ventilation literature has relatively few examples of methods
for determining when a building is occupied. Where building occupancy has been recorded this
has tended to be through occupant diaries, or by hand-picking sections of data for analysis.
Chen, Jiang and Xie (2018) reviewed the literature on determining occupancy in buildings,
finding that different sensors have different results are often obtained when different sensors
are combined. Dedesko et al. (2015) used beam-break sensors in an attempt to count the number
of people passing in or out of the room. They also measured CO2 concentrations and used
estimated CO2 generation rates to determine whether a beam-break event was associated with
someone entering or leaving the room.
The algorithm developed in this research can be used to filter the data collected from the
occupied buildings so that only those times which are identified as unoccupied are used in
ventilation rate calculations; it is based on similar principles to the Dedesko method. The
algorithm is based on the logic that if any of the internal doors or windows change state between
the front door closing and the next time it opens, then the building must be occupied during that
period. If none of the doors or windows change position, then the CO2 concentration gradient
is tested on the basis that if the CO2 rises significantly then the building is highly likely to be
occupied (or another significant source of CO2 is present, which precludes use of the decay
method). The first 30 minutes of data after the front door closes are disregarded if the period
under investigation is sufficiently long to allow some stabilisation of the airflow. This ensures
that if the door to a room with a high concentration of CO2 is opened shortly before the building
becomes unoccupied and causes the concentration of CO2 to rise in adjacent rooms, this period
isn’t falsely identified as occupied. The flow chart in Figure 1 shows the decision making
process used by the algorithm.
3.2 Testing the occupancy algorithm
A case study monitoring campaign was set-up in an occupied flat to provide data for developing
the occupancy detection algorithm. The flat was monitored between February and July 2018.
Three adults lived in the flat and it was unoccupied for several hours most days as all of the
occupants worked full-time. CO2 sensors were placed in every room except the bathroom. Door
Figure 1. Flow chart of the decision making of the occupancy
algorithm.
sensors were placed on all doors. The internal door sensors recorded the state (open or closed)
every 5 minutes (state monitoring), whereas the front door sensor recorded every time the door
opened or closed (event monitoring). The occupants were asked to record when the last person
left the dwelling (occupancy ends) and when the first person entered the dwelling (occupancy
begins). There were 62 reported start or end of occupancy times. Figure 2 shows an example of
a reported occupancy start and end time with the door opening data and CO2 concentrations in
the dwelling.
The occupant records were compared to the results from the algorithm and these were in
agreement in 87% of cases. To calculate this agreement the following logic was used: when the
period before the occupant reported a start of occupancy event was identified by the algorithm
as unoccupied this counted as one event of the algorithm correctly estimating the occupancy.
Similarly, when the period after the occupant reported start of occupancy was identified by the
algorithm as occupied this counted as another event of the algorithm correctly estimating the
occupancy (and vice versa for end of occupancy data). The percentage of agreement was then
calculated.
Disagreements between the occupant records and algorithm were investigated through detailed
study of all the measured parameters. In some cases it is likely that the disagreement was due
to window opening – the temperature rapidly dropped but the CO2 did not rise. This highlights
the need for window sensors to improve the algorithm performance. In other cases, the front
door was in frequent use (likely because the occupants were arriving or leaving for work at
similar times), in 10 of the 16 cases of disagreement the front door was opened with a frequency
of more than 30 minutes. Since the door and window state was recorded every five minutes, the
algorithm was sometimes unable to identify these periods as occupied; use of event logging
equipment will resolve this issue. By recording windows and using event logging, the
agreement would likely have been at or above 95%.
3.3 Measuring the ventilation rate
Calculation of the ventilation rate is based on the single zone approximation of the continuity
equation in which no sources are present (Sherman, 1990):
CO2,Diff (t) = CO2,Diff (t = 0) exp(−A.t) (1)
Figure 2. Example of the occupancy algorithm identifying the time at which the dwelling became
occupied and unoccupied. The top part of the graph shows the measured CO2 concentrations in each of
the rooms, with the black sections of the CO2 data indicate that the space has been identified as occupied.
The grey vertical lines indicate the front door opening and closing. The blue (red) dashed vertical line
indicates that the occupant reported the beginning (end) of the occupancy. The bottom part of the graph
shows the internal doors changing between open and closed.
Figure 3. Examples of data collected in the test dwelling. The top part of each figure shows the CO2 concentration
against time of day in each room: occupied periods are shown in black, data suitable for decay analysis are shown
in colour. The grey vertical lines show when the front door opened and closed. The bottom part of each graph
shows the internal door states. In part a) all the internal doors are open and in figure b) some doors are closed.
Where CO2,Diff(t) = CO2,int(t) – CO2,ext(t), CO2,int(t) is the indoor CO2 concentration, CO2,ext(t) is
the outdoor CO2 concentration, CO2,Diff(t = 0) is the concentration difference at the start of the
decay period, A is the air change rate, and t is the time since the start of the decay.
Figure 3a shows an example of the data collected to test the occupancy algorithm: the transition
from occupied to unoccupied and subsequent decay in CO2 concentration can be seen. In this
case, all internal doors were open and the CO2 concentrations are closely matched in all rooms.
In this case, CO2,int(t) would be the mean of the internal concentrations, as dwelling is behaving
as a single zone.
Figure 3b shows a second example in which two of the rooms have their door closed. The CO2
concentration decay is clearly different in different rooms. In this case, the dwelling does not
behave as a single zone, so a single ventilation rate does not adequately describe the internal
conditions, and the mean internal concentration should not be used in estimating the ventilation
rate. However, the CO2 concentration decay in different rooms can be used to estimate the
ventilation rate in those rooms, with a systematic bias in a known direction. This is a significant
advantage over measurements in which only a single room is measured, or in which the whole
house is treated as homogenous. In this example, one room has higher CO2 concentration
throughout the decay (shown in purple). This room has a lower ventilation rate than the other
rooms. To estimate the ventilation rate in this room CO2,int(t) would be the CO2 concentration
in this room only. The single room ventilation rate calculated would systematically larger than
the ‘true’ indoor-outdoor ventilation rate of this room, because some of the reduction in CO2 is
likely due to air exchange with the other areas of the building.
Conversely, any ventilation rate calculated for the rest of the building using the concentration
recorded in other rooms would be systematically higher than the ‘true’ indoor-outdoor
ventilation rate of this space. This is because the rate of CO2 decay would be reduced by any
leakage of CO2 from the first room. However, it is possible to see that the rest of the building
eventually becomes well mixed, and then decays at the same rate throughout.
Figure 3b highlights the importance of the internal doors, in addition to ventilation to the
outside, in determining the airflow in a dwelling. The implications of such issues on the data
analysis and interpretation are discussed in the following section.
4 DISCUSSION
The method developed in this work may be used to estimate the ventilation rate in intermittently
occupied properties. It does not account for interzonal flows as it uses the single zone
approximation. However, collecting CO2 and door opening data from each room supports
appropriate analysis and interpretation of the results. If the whole dwelling behaves as a single
zone (Figure 3a) this technique gives the outdoor ventilation rate of the space in that
configuration. When the building does not behave as a single zone (as in Figure 3b), the
calculated ventilation rate of any particular room (using the concentration from that room only
as CO2,int(t)) is that assuming the decay in CO2 is entirely due to exchange of air with outdoors
(Mumovic et al., 2009). By recording the concentration in all rooms, the direction of the
systematic bias that the single zone assumption introduces to the ventilation rate calculated for
particular rooms is known.
The calculated ventilation rate for a single room will be systematically higher than the indoor-
outdoor ventilation rate when the room in question has CO2 concentration higher than the
adjacent spaces. This is because some of the reduction in CO2 concentration is due to air
exchange between indoor spaces rather than exchange with outdoors, so that the ‘true’ indoor-
outdoor ventilation rate is lower than that measured. Conversely, the calculated value for a
single room will be systematically lower when the CO2 concentration in this room is lower than
the adjacent spaces. This is because some additional CO2 may be flowing into the room in
question from the adjacent rooms, causing a reduction in the measured rate of CO2 decay. The
closer the indoor spaces to the concentration in the room in question, the closer the measured
value will be to the indoor-outdoor ventilation rate. These measured ventilation rates in
particular rooms might be considered ‘effective’ ventilation rates: when the CO2 source is
removed and given the distribution of CO2 in the building, the capacity of the building to act as
a fresh air reservoir and the configuration of the doors and windows, this quantifies how quickly
the CO2 decays to background concentrations.
Since the building is continuously monitored, the similarity of the configuration of the building
during occupied times and unoccupied ventilation measurement periods can be assessed. This
means it is possible to understand how the occupants use the building when they are present,
and whether their use of internal doors means that the building is likely to behave as one single
zone or if they are likely to experience different ventilation rates depending which room they
are in during occupied times. This provides much more information on the ventilation
conditions that occupants are likely to experience, and how this may vary, than is possible using
standard single zone methods.
It should be noted that the binary nature of the window and door sensor limits the extent to
which the occupied and unoccupied conditions can be compared – there is no record of whether
doors or windows are ajar or fully open.
Weather conditions will also affect the ventilation in a building. The sensors required are silent
and reasonably small, meaning that participants may be willing to accept the presence of the
equipment for an extended period. An extended measurement campaign allows the variation of
ventilation under different weather conditions to be explored. However, the cost of the
equipment and management of the measurement campaign are important restrictions to the
application of this method.
The proposed method allows detailed insights into the ventilation conditions in occupied
buildings. The occupancy algorithm may be used to analyse large volumes of data over
extended measurement campaigns with much reduced workload compared to manual selection.
This method enables the investigation of the varying nature of ventilation rate, and an
understanding of how the occupant’s use of the building may affect the ventilation rates they
experience.
5 CONCLUSIONS
Estimating the ventilation rate in occupied dwellings is challenging, and the method employed
depends on the desired insights, occupant acceptance and resource limitations. Issues include
the timescale over which estimates are required (e.g. average over a period, or only during
occupied hours), the spatial scale (e.g. specific room or whole building), the intrusiveness of
measurements and occupant acceptance, and the researcher time required.
The method developed in this paper is based on metabolic CO2 tracer gas decay, and uses an
automated algorithm to detect occupancy. This method enables the variation in ventilation to
be explored. The measurements may be acceptable to occupants since CO2 is naturally present
and the sensors can be silent and reasonably small. These sensors are not expensive, and are
combined with sensors that determine the open/closed state of windows and doors, to
investigate the differences in ventilation rates in different spaces and under different
configurations. The window and door sensors also allow investigation into the extent to which
different configurations are experienced in occupied and unoccupied times, and the impact that
this has on the ventilation rate that the occupant experiences.
The proposed method uses an automatic algorithm for determining when the building is
occupied. The algorithm has been shown to be agree with occupant records in over 80% of
cases, which will increase with the use of window sensors and event rather than state loggers.
This algorithm reduces burden on the researcher and occupant to manually record or interpret
data, enabling the analysis of large volumes of data.
The proposed method represents a step towards appropriate measurement of ventilation and its
spatial and temporal variation in occupied buildings. It may enable the measurement of
buildings over an extended period, supporting ventilation rate estimation in many different
configurations and weather conditions. Such results will support insights into the ways that a
building is used by occupants and the ventilation rates that they are likely to experience given
their use of the building. This will provide insight into the thermal comfort of occupants, as
well as the IAQ they experience. The exposure of occupants to contaminants is complex,
depending on ventilation rates, source strength, source location and exposure time: this method
provides a detailed picture of the ventilation in occupied buildings, and could help to better
understand the association between ventilation rates and health.
6 ACKNOWLEDGEMENTS
This research was made possible by support from the EPSRC Centre for Doctoral Training in
Energy Demand (LoLo), grant numbers EP/L01517X/1 and EP/H009612/1.
7 REFERENCES
ASTM (2011) E741-11: Standard Test Method for Determining Air Change in a Single Zone
by Means of a Tracer Gas Dilution. doi: 10.1520/E0741-11.
Bekö, G. et al. (2016) ‘Diurnal and seasonal variation in air exchange rates and interzonal
airflows measured by active and passive tracer gas in homes’, Building and Environment, 104,
pp. 178–187. doi: 10.1016/j.buildenv.2016.05.016.
Bornehag, C. G. et al. (2005) ‘Association between ventilation rates in 390 Swedish homes and
allergic symptoms in children’, Indoor Air, 15, pp. 275–280. doi: 10.1111/j.1600-
0668.2005.00372.x.
Bryman, A. (2004) Social Research Methods. 2nd edn. Oxford: Oxford University Press.
Chen, Z., Jiang, C. and Xie, L. (2018) ‘Building occupancy estimation and detection: A review’,
Energy and Buildings. Elsevier B.V., 169, pp. 260–270. doi: 10.1016/j.enbuild.2018.03.084.
CIBSE (2015) CIBSE Guide A: Environmental Design. 8th edn. London.
Dedesko, S. et al. (2015) ‘Methods to assess human occupancy and occupant activity in hospital
patient rooms’, Building and Environment. Elsevier Ltd, 90, pp. 136–145. doi:
10.1016/j.buildenv.2015.03.029.
Fisk, W. J. (2018) ‘How home ventilation rates affect health: A literature review’, Indoor Air,
28(4), pp. 473–487. doi: 10.1111/ina.12469.
Guo, L. and Lewis, J. O. (2007) ‘Carbon dioxide concentration and its application on estimating
the air change rate in typical Irish houses’, International Journal of Ventilation, 6(3), pp. 235–
245. doi: 10.1080/14733315.2007.11683780.
Harrje, D. T. et al. (1990) ‘Tracer Gas Measurement Systems Compared in a Multifamily
Building’, in Sherman, M. H. (ed.) Air Change Rate and Airtightness in Buildings, ASTM STP
1067. Philadelphia: American Society for Testing and Materials, pp. 5–20.
Iwashita, G. and Akasaka, H. (1997) ‘The effects of human behavior on natural ventilation rate
and indoor air environment in summer - a field study in southern Japan’, Energy & Buildings,
25, pp. 195–205.
Keig, P., Hyde, T. and McGill, G. (2016) ‘A comparison of the estimated natural ventilation
rates of four solid wall houses with the measured ventilation rates and the implications for low-
energy retrofits’, Indoor and Built Environment, 25(1), pp. 169–179.
Kvisgaard, B. and Collet, P. F. (1990) ‘The User’s Influence on Air Change’, in Sherman, M.
H. (ed.) Air Change Rate and Airtightness in Buildings. American Society for Testing and
Materials, pp. 67–76.
Liddament, M. (1996) A Guide to Energy Efficient Ventilation. Warwick: Air Infiltration and
Ventilation Centre, International Energy Agency.
Mumovic, D. et al. (2009) ‘A methodology for post-occupancy evaluation of ventilation rates
in schools’, Building Services Engineering Research and Technology, 30(2), pp. 143–152. doi:
10.1177/0143624408099175.
Oreszczyn, T. et al. (2005) ‘The reduction in air infiltration due to window replacement in UK
dwellings: Results of a field study and telephone survey’, International Journal of Ventilation,
4(1), pp. 71–78. doi: 10.1080/14733315.2005.11683700.
Penman, J. M. and Rashid, A. A. M. (1982) ‘Experimental determination of air-flow in a
naturally ventilated room using metabolic carbon dioxide’, Building and Environment, 17(4),
pp. 253–256. doi: 10.1016/0360-1323(82)90017-8.
Persily, A. (2006) ‘What we think we know about ventilation’, International Journal of
Ventilation, 5(3), pp. 275–290. doi: 10.1080/14733315.2006.11683745.
Persily, A. K. (2016) ‘Field measurement of ventilation rates’, Indoor Air, 26(1), pp. 97–111.
doi: 10.1111/ina.12193.
Roulet, C.-A. and Foradini, F. (2002) ‘Simple and Cheap Air Change Rate Measurement Using
CO2 Concentration Decays’, International Journal of Ventilation, 1(1), pp. 39–44. doi:
10.1080/14733315.2002.11683620.
Sharpe, T. et al. (2015) ‘Occupant interactions and effectiveness of natural ventilation strategies
in contemporary new housing in Scotland, UK’, International Journal of Environmental
Research and Public Health, 12(7), pp. 8480–8497. doi: 10.3390/ijerph120708480.
Sherman, M. (1998) The Use of Blower-Door Data. Lawrence Berkley Lab Report 35173.
Berkelely, California: Lawrence Berkeley Laboratory. doi: 10.1111/j.1600-0668.1995.t01-1-
00008.x.
Sherman, M. H. (1990) ‘Tracer-gas techniques for measuring ventilation in a single zone’,
Building and Environment, 25(4), pp. 365–374. doi: 10.1016/0360-1323(90)90010-O.
Smith, P. N. (1988) ‘Determination of ventilation rates in occupied buildings from metabolic
CO2 concentrations and production rates’, Building and Environment, 23(2), pp. 95–102. doi:
10.1016/0360-1323(88)90023-6.
Sundell, J. et al. (2011) ‘Ventilation rates and health: Multidisciplinary review of the scientific
literature’, Indoor Air, 21(3), pp. 191–204. doi: 10.1111/j.1600-0668.2010.00703.x.
UNFCCC (2015) Convention on Climate Change: Paris Agreement.
Wallace, L. A., Emmerich, S. J. and Howard-Reed, C. (2002) ‘Continuous measurements of air
change rates in an occupied house for 1 year: the effect of temperature, wind, fans, and
windows.’, Journal of Exposure Analysis and Environmental Epidemiology, 12(4), pp. 296–
306. doi: 10.1038/sj.jea.7500229.