Science of the Total Environment 502 (2015) 557–570

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

An integrated evaluation study of the ventilation rate, the exposure and

the indoor air quality in naturally ventilated classrooms in the
Mediterranean region during spring
Paraskevi Vivian Dorizas ⁎, Margarita-Niki Assimakopoulos, Constantinos Helmis, Mattheos Santamouris
Faculty of Physics, Departments of Environmental Physics and Meteorology, University of Athens, University Campus, Athens, 157 84, Greece

H I G H L I G H T S

• Ventilation rates were adequate for certain cases for the majority of the schools.
• Increased PM10 due to the students' presence, inadequate ACH and chalk use.
• The excessive exceedence of 1 of the PM I/O ratio shows strong indoor sources.
• Schools using chalk had higher PM10 levels compared to schools using marker.
• Schools using marker had greater CO and VOCs levels compared to schools using chalk.

a r t i c l e i n f o a b s t r a c t

Article history: Ventilation rates and indoor air pollutants have been extensively monitored in nine naturally ventilated primary
Received 20 February 2014 schools of Athens, Greece during spring. The ventilation rates and pollutant levels were studied during the
Received in revised form 17 September 2014 teaching and non-teaching periods and ventilation profiles were created for each of the schools. The median
Accepted 19 September 2014
ventilation rates per school ranged between 0.7 and 8 ACH while the average ventilation rate in all schools
Available online 7 October 2014
(11.7 l/s/p) was greater than the minimum recommended rates by ASHRAE for school classrooms. The average
Editor: Pavlos Kassomenos carbon dioxide (CO2) concentrations per school varied between 893 and 2082 ppm, while the majority of the
cases were slightly above the recommended limit values. CO2 concentrations were also positively correlated to
Keywords: the number of students and negatively correlated to the ventilation rates. Particles of several size ranges
IAQ (PM10, PM5, PM2.5, PM1, PM0.5 and UFP) were also measured and analyzed. PM10 concentrations exceeded the
PM recommended limit values by more than 10 times for the majority of the cases. There were also many cases
CO2 that the PM2.5 concentrations exceeded their limit values. PM concentrations were significantly affected by the
Ventilation ventilation rates and the presence of students. All of the measured particle sizes were greater during teaching
Health
than the non-teaching hours. For most of the cases the indoor to outdoor (I/O) concentrations ratios of PM10
Naturally ventilated schools
and PM2.5 were much greater than one, indicating that the indoor environment was being mostly affected by
indoor sources instead of the outdoor air. Furthermore it was found that chalk and marker boards' usage
significantly affect indoor pollutant concentrations. Overall, the measured levels of exposure were for most of
the cases greater than the recommended guideline values due to the intense presence of indoor pollution
sources, even though the ventilation rates were in general satisfactory.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction in urban areas (Sarigiannis et al., 2004). Studies conducted in indoor

There is epidemiological evidence that the exposure of human to
air pollutants such as particulate matter (PM) can lead to hazardous
health effects such as asthma (Pope and Dockery, 2006; Bakke et al.,
2012). Atmospheric PM is mainly responsible for adverse health effects
⁎ Corresponding author. Tel.: +30 2107276849.
E-mail address: pdoriza@phys.uoa.gr (P.V. Dorizas).
environments and human exposure, have related poor air quality with
asthma, rhinitis, allergic diseases and eczema increases (Annesi-
Maesano et al., 2003; Sarigiannis et al., 2009; Kim et al., 2010; Jantunen
et al., 2011; Bakke et al., 2012). As the indoor air quality (IAQ) can influ-
ence by far human health, numerous studies have been carried on in
several indoor environments such as dentistry clinics, underground
trains, residences and even air traffic control towers (Helmis et al.,
2007; Chau et al., 2008; Helmis et al., 2009; Assimakopoulos et al.,
2013).

http://dx.doi.org/10.1016/j.scitotenv.2014.09.060
0048-9697/© 2014 Elsevier B.V. All rights reserved.
558 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570
Infants and young children are highly vulnerable to the health effects
compared to adults, to the exposure to indoor air pollution, due to their
higher resting metabolic rate in proportion to their body weight. In
particular, the air intake of an infant is proportionally about double
that of an adult meaning that children have significantly greater expo-
sures to any kind of toxicants in the air compared to adults (WHO,
2005a; Annesi-Maesano et al., 2003).
In developing countries and urban areas children spent a substantial
amount of their day indoors and mainly in schools (Silvers et al., 1994;
Bai et al., 2007; Hussein et al., 2012). Several studies conducted in school
environments have concluded that the indoor air quality of classrooms
is significantly unhealthy (Dorizas et al., 2013a; Fromme et al., 2007;
Fromme et al., 2005; Goyal and Khare, 2009; Siskos et al., 2001;
Synnefa et al., 2003; Tippayawong et al., 2009). It has also been found
that mean cancer risk for urban school children is higher than the corre-
sponding sub-urban school children (Demirel et al., 2014). Poor indoor
environment in classrooms may be linked to many factors such as inad-
equate levels of ventilation as well as to the lack of necessary resources
for the adequate operation and maintenance of facilities. Recent studies
made by Myhrvold et al. (1996), Mendell and Heath (2005), Wargocki
et al. (2005), and Wargocki and Wyon (2007) have found that poor
IAQ and inadequate ventilation rate can also be linked to reduced
academic achievements and general performance of schoolwork.
Furthermore, studies related with the indoor and outdoor air quality
of schools have found that the outdoor environment plays a decisive
role on the indoor pollutant levels (Chaloulakou et al., 2003; Diapouli
et al., 2007; Goyal and Khare, 2009; Guo et al., 2010; Tippayawong
et al., 2009; Yoon et al., 2011). Further relationships between the indoor
and outdoor levels of particulate matter have been examined by
Jamriska et al., 1999; Jones et al., 2000 and Morawksa et al., 2001.
Despite the fact that a great number of studies have been carried out
in school environments, the ventilation requirements in naturally venti-
lated buildings in conjunction to the indoor air pollutant levels are still
not well understood. Therefore, additional information concerning ven-
tilation rate and indoor air pollutants are necessary. The main objectives
of this study are: 1. to extensively analyze the airflow rates of naturally
ventilated classrooms in warm (Mediterranean) climates, 2. to examine
the concentration levels of the main indoor air pollutant concentrations
such as CO2, CO, VOCs, and PM of several size ranges within classrooms,
3. to investigate the level at which the outdoor air affects the indoor air
pollutant levels, and 4. to identify to which extend factors such as
human presence, ventilation rate and the use of chalk or marker boards
can influence indoor air pollutants concentrations.
Fig. 1. Map of Attika, Greece (left), and location of the 9 schools (right).
2. Methodology
2.1. Sampling site description & measurement period
The measurements were carried out during spring 2013 in nine nat-
urally ventilated primary schools of Attika basin in Greece (Dorizas
et al., 2013b). Eight of them are located in the north-western part of
Attika (Thrakomakedones, Acharrnae) and one school was located in
the eastern regional area of Attika (Pallini) (Fig. 1). According to
CORINE 2000 land cover database (Geodata, 2010), the majority (6
out of 9) of the schools (code names: 1, 14, 4, 18, 2, 8, see Table 1) are
in areas characterized as ‘discontinuous urban fabric’ in which a great
percentage of the land is covered by structures (EEA, ETC./TE, 2004)
(Fig. 2, left). One of the schools (code 12) is at an area of ‘continuous
urban fabric’ where buildings and roads cover more than 80% of the
total surface (Fig. 2, right), and two other schools (codes: 3 & 11) are
in areas having ‘complex cultivation patterns ‘where small areas of
annual crops are present (Fig. 3).
It is worth mentioning that our aim was to conduct measurements at
an adequate number of schools for a period of the order of one week at
each location. On the other hand according to our preliminary measure-
ments, the values of concentration of pollutants and ventilation rates in
the schools were within the same range, under the same activity pattern
and meteorological conditions. Also there were limitations in time given
by the school directors. Thus, all conducted measurements, even with
different measuring periods, were included in the data base, in order
to use measurements from an adequate number of schools.
The monitoring was carried out in one classroom per school, the 6th
grade where students are of 11–12 years old. Table 1 summarizes some
of the main characteristics of the schools, buildings and classrooms. The
occupancy density ranged from 1.84 to 3.67 m2/student, while the aver-
age floor area of the monitored classrooms is 51.78 m2 which is similar
for the majority of the schools. Six out of the nine schools used chalk
whiteboards (codes: 1, 14, 4, 3, 18, 11) while the remaining three
used marker white boards (codes: 12, 2, 8). The age of the school
buildings varied from 10 to 35 years. The experimental campaign was
performed during mid-season (spring), from the beginning of April
until the ending of May 2013. The reason that the monitoring took
place outside the heating season was in order to assess the ventilation
conditions and pollutant levels into a free-running mode of the school
buildings, as the variations in temperature and relative humidity from
a mechanically operated heating system would influence by far their
diurnal distribution and fluctuations (Teli et al., 2013). In this case the
 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570 559

Table 1
Main characteristics of the schools and the classrooms.

School name School code Measurement Year of construction Classroom's Classroom's Classroom's Classroom's Classrooms'
name period (number of Classroom's floor volume (m3) number of number of board type orientation
measurement days) area (m2) students Students

Acharnae 14 14 1–5/4/13 2001 53 165 17 Chalk North

(5 days)
Thrakomakedones 1 1 8–12/4/13 1978 64 198 25 Chalk Northwest
(5 days)
Axharnae 4 4 14–18/4/13 & 24/4/13 1986 50 155 24 Chalk Southwest
(5 days)
Pallini 3 3 19 & 22/4/13 N/A 46 137 25 Chalk West
(2 days)
Acharnae 18 18 23/4/13 1991 47 138 18 Chalk South
(1 day)
Acharnae 12 12 13–17/5/13 1980 49 157 25 Marker South
(5 days)
Thrakomakedones 2 2 20–24/5/13 2003 50 162 25 Marker East
(5 days)
Acharnae 8 8 27–29/5/13 1999 52 159 19 Marker West
(3 days)
Acharnae 11 11 31/5/13 1994 55 172 15 Chalk South
(1 day)

effects of indoor sources such as the window opening, the presence of
students etc. were the key factors to be examined.
2.2. Variables measured and measuring instruments
Several parameters were measured in the indoor and outdoor envi-
ronment of the classrooms. In particular, carbon dioxide (CO2), carbon
monoxide (CO) and volatile organic compounds (VOCs) were measured
using a multigas monitor MultiRAE IR Model PGM-54 (RAE Systems) in
units of parts per million (ppm). At this point it must be noted that the
VOCs are the TVOCs and that the gas used for the calibration of the
VOCs sensor was isobutylene in air of 100 ppm. The accuracy of
MultiRAE IR was checked by exposing the sensors to known concentra-
tion calibration gases right before the beginning and once during the
monitoring period and according to the manufacturer had to be within
0 ± 20% of the actual. The indoor PM10, PM5, PM2.5, PM1 and PM0.5 con-
centrations were measured using Handheld3016, IAQ (Lighthouse,
worldwide solutions) which is designed for ISO 21501-4 compliance
with unit to unit accuracy and repeatability. The counting efficiency of
Handheld 3016-IAQ is 100% for particles greater than 0.45 μm. This mon-
itor is calibrated according to ISO 21501-4 which is standard in the in-
dustry for particle counter calibration using Polystyrene Spheres Latex
(PSL particles). The PSL particles are spherical and white which are per-
fect attributes for performing size calibration on light scatting sensors
focusing on sieve analysis. The outdoor PM10, PM2.5 and PM1 concentra-
tions were sampled using Osiris, an airborne particulate monitor
(Turnkey Instruments, Ltd) in units of mass per unit volume (μg/m3).
The lower detection limit of Osiris is 0.1 μg/m3 and measurements
range up to 6000 with an accuracy of 0.1 μg/m3. In order to compare
the concentrations from the two instruments (Handheld 3016 and
Osiris) measuring PM in the indoor and outdoor environment, a correc-
tion factor was calculated correcting the values of the one instrument as
a function of the other one. This correction factor was estimated by mea-
suring at a lab under constant conditions without any pollutant sources
functioning, using both of the instruments. Ultrafine particle (UFP)
concentrations were measured using the portable condensation particle
counter P-Trak (TSI, Model 8525), in units of number of particles
per unit volume (pt/cm3), while the detection range of P-Trak is
from 20 nm to 1 μm. The relative accuracy of P-Trak to the SMPS/CPC
was equal to 95% and the precision was between 3 and 12% (Hagler et
al., 2009). However the accuracy of P-Trak by itself is not specified
by the manufacturer. The Handled 3016, Osiris and P-Trak were cali-
brated by their manufacturers' right before the beginning of the mea-
surements. Finally, the air temperature (T) and relative humidity (RH)
were also measured in the indoor and outdoor environment using
Tiny Tag data loggers (Model: TGP-4500, Gemini Data loggers) with
self-contained sensors with an accuracy of ± 0.6 °C and ± 3.0% RH at
25 °C.
2.3. Calculation of air exchange rates
The ventilation and infiltration rates of the naturally ventilated class-
rooms were calculated using the tracer gas decay method which in-
volves the injection of a tracer gas, its mixing through fans in a space

Fig. 2. Left: discontinuous urban fabric area (School 14); right: continuous urban fabric area (School 12).
560 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570
Fig. 3. Complex cultivation pattern area (School 11).
and then the measurement of its decay rate with the appropriate mea-
suring instrument. This is one of the most commonly used methods for
the measurement of the ventilation rates reported in a great number of
studies (Johnson et al., 2004). In this study the tracer gas used was sulfur
hexafluoride (SF6) whose decay was measured with the INNOVA model
1312, a photo-acoustic multi-gas monitor whose measuring principle is
based on the photoacoustic infra-red detection method. The accuracy of
the monitor according to the manufacturer is ±1.5 ppm. The air flow
rate is defined as the slope of the SF6 concentration decay curve for a
certain period of time using a semi-logarithmic plot (Eq. (1)) (Nantka,
1990):
ln Cðt1Þ−ln Cðt2Þ
ACH ¼
t2−t1
where C(t1) and C(t2) are the concentrations of SF6 in ppm, at times t1
and t2 respectively and t2-t1 is the total measurement period in hours.
For the majority of the ventilation measurements using the tracer
gas method some assumptions are made concerning the zone of
measurement and the airflow paths. The main assumption is that the
zone under investigation can be treated as uniformly well-mixed in
terms of the tracer gas and any injected tracer is also assumed to be in-
stantaneously well-mixed (Sherman and Dickerhoff, 1989). In practice,
the idealized condition of perfectly well-mixed zones is rarely realized;
however it has been observed that it can be reasonably approached with
the use of fans that mix the indoor air (Sherman and Dickerhoff, 1989).
In this campaign, a fan was used to ensure that the tracer gas concentra-
tion is the same at all points of the zone, located away from points
where the flows between the zones can be changed such as near a
window or at a leakage site (Sherman and Dickerhoff, 1989).
Throughout each day's monitoring, several window opening scenar-
ios ranging from fully closed (infiltration rates, will be studied in para-
graph 3.2.1), slightly open, half open and fully open operated in each
of the classrooms. In order to calculate the ventilation rates of every
single opening, at the end of each day's monitoring these certain
window openings were repeated and the ventilation measurements
were performed using the tracer gas decay method mentioned above.
The procedure for measuring the ventilation rates for all window open-
ings scenarios was exactly the same. The tracer gas SF6 was released at
several different points of the classroom and for good mixing of the
air, the fun operated during the entire measurement assuming a steady
air flow rate from outdoors. In order to verify that good distribution was
achieved, the tracer gas decay was measured at five different points of
the classroom and at the end, the average value of these points was
taken into account. The reason for not making these measurements
during the day was because the tracer gas technique requires the injec-
tion of the gas SF6 which even though it is non-toxic, its release could
provoke a negative reaction of the pupils. At this point it is worth men-
tioning that children's presence may influence the indoor mixing of the
air either by promoting mixing due to convection flows or by damaging
mixing as the air can be accumulated close to the classroom's ceiling.
This fact constitutes a limitation of the selected methodology; however
it was assumed that this error was of the order of magnitude of about
5–10% and was not taken under consideration. Differences on the height
of the classrooms could have also had an important impact on the results
however all classrooms had similar height ranging from 3 to 3.2 m. Fur-
thermore, the fact that the ventilation and infiltration measurements
were performed at different time of the day from the actual measure-
ments, could lead to inaccurate estimation of the actual rates as there
were cases that the weather conditions changed during the day. In par-
ticular, the temperature was usually slightly increasing by 2 °C from the
beginning to the end of the measurements while the relative humidity
was decreasing by approximately 10%. The error was estimated to be be-
tween 5 and 8%. The wind velocity was slightly changing all through the
day. Special attention was paid so as to take into account only the mea-
surements of similar weather conditions of the ones during the day to
the ones during the time the ventilation measurements were performed.
Finally, certain properties should be met in the zones (classrooms)
where the ventilation measurements using the tracer gas method are
carried on: (i) In matters of homogeneity, the SF6 properties should be
the same at every point of the zone. (ii) The zone should be isolated
without any re-entrainment of tracer. (iii) Finally the zone should be
well mixed meaning that the injected tracer should be instantaneously
dispersed within the zone. However, practically the air requires a finite
time to be well mixed (Sherman, 1990). If the aforementioned condi-
tions are not met, there is a possibility of introduction of significant
errors and inaccuracies. Santamouris et al. (2008) estimated that the
error due to non-perfect homogeneity was between 4 and 7%. Special
effort was made in this study to meet the required properties of the
ð1Þ
zone so as to reduce the inaccuracies.
In the schools that the measurements lasted for more than one day,
the ventilation measurements were repeated for three times in three
different days. The averaged ventilation rate of these three measure-
ments per each opening scenario was finally used for each school.
In all classrooms single sided ventilation strategy was used and no
other opening types apart from windows and doors were available. It
is worth mentioning that excluding schools 12 and 18 where the
classroom's door was connecting the classroom to the outdoor environ-
ment, for the rest of the schools the door was connecting the classroom
to an indoor corridor. It should be also noted that the doors and win-
dows of the corridors outside the classrooms during the ventilation
measurements were kept as they were during the school day so as to
conduct the measurements under the same conditions and therefore
reduce the measurements errors. Table 2 summarizes the open window
areas for the ‘fully open’ window scenario and floor areas for each of the
classrooms that the measurements were carried on.
2.4. Sampling protocol
In total, 32 days of measurement were performed in the 9 schools.
Table 1 summarizes in detail the schools and the exact dates in which
Table 2
Fully open window areas and floor areas per classroom per school.
Schools Open window area in the ‘fully Classroom's
open’ window scenario (m2) floor area (m2)
14 3.50 53
1 3.90 64
4 4.60 50
3 2.40 46
18 1.70 47
12 4.30 49
2 2.90 50
8 1.20 52
11 0.45 55
 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570 561

the measurements were conducted. In each of the nine schools the 70 45

measurements lasted from one to five days depending on the availabil-

Averaged RH per school (%)

60

Avergaed T per school (0C)

ity of the schools. Measurements started 40 min prior to the arrival 40 Indoor RH
of tudents in the classrooms (at around 7:30 local time-LT) and lasted Outdoor RH
50
40 min after the students left school (until about 14:40 LT). Sampling 35
took place in one classroom per school and the experimental equip- Upper RH (50%)
40
ment was placed at 1.1 m above the floor according to the standard Lower RH (45%)
30
ISO7726:1998 for seated persons, as close as possible to the center of
30
the classroom. Fig. 4 presents the sampling point locations in one of Indoor T
the classrooms (left) and the outdoor environment (right). The instru- 25
20 Outdoor T
ments measuring the outdoor air were placed at the roof terrace so as
to collect representative sample of the atmospheric air. All the above 10 20 Upper T (26oC)
mentioned parameters were monitored with a sampling interval of
Lower T (19oC)
5 min and the duration of each measurement was approximately 0 15
7 hours per day. Any kind of indoor activities that could possibly affect 14 1 4 3 18 12 2 8 11
the indoor air pollutant concentrations such as the presence and num- Schools
ber of students, window opening scenarios etc. were written on daily
diaries marking the exact starting, ending time and duration of the Fig. 5. Averaged temperature and relative humidity per school.
activity.

3. Results & discussion
3.1. Thermal environment
Fig. 5 presents the distribution of the averaged indoor and outdoor
temperatures and relative humidity per school for the entire measure-
ment period. According to the directives of the Technical Chamber of
Greece, the recommended temperature for school buildings is between
19 and 26 °C and the range for relative humidity is between 45 and 50%
(Santamouris et al., 2007). Both the temperature and also relative
humidity indoors were higher than outdoors and the indoor trends
followed the outdoor ones. The indoor temperature for the majority of
the cases was between the recommended bands; however the indoor
relative humidity for 6 out of the 9 schools was outside the recommend-
ed bands.
3.2. Air exchange rates in classrooms
In this chapter the infiltration and ventilation rates are presented for
each of the classrooms, are then compared with each other and are fi-
nally presented separately for the teaching and the total measurement
period.
3.2.1. Infiltration rates
The infiltration rates refer to the flow of air into an area, through the
cracks of the building's envelop, with fully closed windows and doors
and it is an indicator of how air tight the construction is. Fig. 6 presents
the infiltration rates (or minimum ventilation rates, green thick dotted
bars) and the percentage of time the classrooms had the windows
closed (pink thin bars) and therefore the infiltration rates were func-
tioning in each of the 9 schools. Fig. 6 is not a correlation diagram. The
infiltration rates ranged between 0.11 (School 3) and 0.82 ACH (School
11). The higher rates (N 0.7 ACH) in schools 11 and 14 could be attribut-
ed to the increased wind speed on the days the measurements were
conducted, while the building's cracks and air leakages from the win-
dows' frames due to the older building construction could be the reason
for the increased infiltration rates in school 12. Moreover, closed win-
dows for long time (N50%, e.g. school 1) throughout the day could lead
to the accumulation of air pollutants indoors. However, the prolonged
window opening could also increase the indoor air pollutant levels in
the case the outdoor environment is more polluted compared to the in-
door one. The estimated infiltration rates in the schools of this experi-
mental campaign were slightly lower compared to the corresponding
findings of Park et al. (2008) in which the rates ranged between 0.5
and 1.5 ACH and were about the same compared to the findings of
Bartlett et al. (2004) (0.0148–0.738 ACH).
3.2.2. Ventilation rates
Fig. 7 presents in box plots the distribution of the ventilation rates in
ACH (left) and in liters per second per person (l/s/p) (right) throughout
the entire measurement period in each of the 9 schools. As aforemen-
tioned in chapter 3.2.1 the minimum ventilation rates represent the
infiltration rates (ventilation rates when all openings are fully closed).
In order to calculate the distribution of the rates, a time series of the

Fig. 4. Sampling point locations in the classroom (left) at the roof terrace (right).
562 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570

0.9 90

0.8 80

Percentage of time with windows closed (%)

0.7 70 Infiltration
-min

Infirtration rates (ACH)

0.6 60 ventilation
rates
0.5 50

0.4 40

0.3 30 % of time
min
0.2 20 ventilation
rates are
0.1 10 obtained

0 0
3 4 8 1 18 2 14 12 11
Schools code names

Fig. 6. Infiltration rates in the schools of measurement & percentage of time with fully closed windows per school.

ventilation and infiltration rates throughout the day was created.
This was done by taking into account the exact time and duration of
every window opening. This information was taken from the daily
diaries (logbooks) that were filled in by the people performing the ex-
perimental campaign in cooperation with the teachers every single
day of the monitoring. Students' behavior on the window opening was
also logged.
The ventilation rates for the majority of the schools present a large
dispersion around the median value meaning that several window
opening scenarios operated and therefore several ventilation rates
occurred in these certain schools (Fig. 7). Schools 1 and 2 are the excep-
tions to this in which only the scenarios of fully closed and fully open
took place (Fig. 7, left, only two extreme values). The distributions
extended from about 1 to 22 ACH; however the averaged median values
per school ranged between 0.7 and about 8 ACH. The increased ventila-
tion rates in school 3 could have been due to the wind's direction which
was against the classroom's windows for many cases. Furthermore the
window opening area in the fully open window scenario of school 4
was the greatest (Table 2) compared to the rest of the schools; therefore
greater ventilation rates were introduced into the classroom.
In the diagram, on the right, the ventilation rates are expressed in
l/s/p. In order to convert the ACH into l/s/p the ventilation rates were
divided by the classroom's volume and the number of people being
present in the classroom at that time. As in many cases the classrooms
were empty, in order to make the division, it was assumed that the
classrooms were occupied by one person; therefore the number arisen
is significantly increased, due to its division by a very small number. The
median ventilation rates per school ranged from 1.03 to 15.41 l/s/p with
an average equal to 1 l.71/s/p, a value higher than the ASHRAE Standard
62 recommended minimum of 8 l/s/p for classrooms. These findings
are also higher than the corresponding ones by Shaughnessy et al.
(2006) who have found mean ventilation rates in classrooms equal to
3.88 l/s/p from a range of 0.9–11.74 l/s/p. Also, the average ventilation
rates in Portuguese classrooms were equal to 2.4 l/s/p, and in Finnish

Fig. 7. Ventilation rate distributions in ACH (left) and l/s/p (right) in each of the 9 schools for the entire period of measurement.
 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570 563

1.0 1.0
0.9 0.9
0.8 0.8

P(CO2<CO2thres)
0.7
(P(vent<ventthres)

0.7 ACH-
Teaching 0.6
0.6 Total
ACH- 0.5
0.5 Total 0.4 Teaching
0.4 l/s/p- 0.3
0.3 Teaching Non-
0.2 teaching
0.2 l/s/p- 0.1
Total
0.1 0.0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

5,000
0.0
0 5 10 15 20 25 30 35 40 45
Ventilation rates thresholds (Ventthres) CO2 concentration thresholds (CO2thres) (ppm)

Fig. 8. Cumulative frequency distribution of the ventilation rates in ACH and l/s/p for the Fig. 9. Cumulative frequency distribution of CO2 concentrations (ppm) in the nine schools.
total and teaching periods in all the schools.

classrooms the rates were equal to 7.2 l/s/p, much lower than the
findings of the present study (Canha et al., 2013).
Furthermore, the measurement period was separated and studied in
two different periods, the teaching and the non-teaching periods. The
non-teaching periods were periods with less than 10 students within
the classroom, which was usually either an empty classroom or a tran-
sient period where the students were entering or going out of their
classrooms. The fact that the non-teaching period for certain cases was
a transient period, could be further linked to intense dynamics. The
teaching periods account 60% of the total time of measurements and
the non-teaching periods account the remaining 40%. The separation
in teaching and non-teaching periods is essential as the non teaching
hours account a significant percentage of the total monitoring period
and it also gives a more clear profile of the diurnal dynamic behavior
of the pollutants. Furthermore, the non-teaching period is a period
proceeded the teaching period which is important as it can significantly
affect it. The separation is also used for comparison purposes to other
studies.
The ventilation rates of all the schools for the teaching and the total
measurement periods are presented in cumulative frequency distribu-
tion diagrams in units of ACH and l/s/p in Fig. 8. It is worth mentioning
that the two periods of teaching and total refer to a different occupancy.
The teaching period refers to a fully occupied classroom while the total
period also includes periods at which the classrooms were empty.
Taking into account the different occupancy and the fact that the venti-
lation rates expressed in l/s/p involve the occupancy of the classrooms it
is expected to have differences between the ventilation rates of the
teaching and total period. As for the case of ACH, it has been found
that the ventilation rates during the teaching and total period were ap-
proximately the same as the curves were almost overlapping. This could
be attributed to the fact that for a lot of the cases the windows were kept
open during the teaching periods and the breaks and therefore the
ventilation rates were similar to the total period of measurement.
The median flow rate in all of the schools for the total period of mea-
surement is 4.16 ACH and 11.7 l/s/p and for the teaching period is 4.34
ACH and 7.51 l/s/p. A similar study conducted in school buildings by
Santamouris et al. (2008) found a median value close to 4.5 l/s/p during
the teaching period and for a data of naturally ventilated classrooms the
value was about 3 l/s/p. According to Santamouris et al. (2008) this
difference could be linked to the fact that their study had at least
some windows open while in the data base of naturally ventilated build-
ings the windows were kept closed for most of the cases. They conclud-
ed that their study is more representative for schools in warm climates
while the cases of the data base represent more of an average situation.
Overall, the classrooms of this study seemed to be better ventilated
compared to the other studies. This can be linked to the fact that the
measurements took place during spring time (April and May, please
refer to Table 1) where for a great percentage of the monitoring period
the windows were kept open (Fig. 6, pink bars) during the measure-
ments due to increased temperatures. It may be concluded that the
findings of this study as in the study of Santamouris et al. (2008) are
more representative for schools in warm climates.
3.3. Levels of indoor air pollutant concentrations
3.3.1. Analysis of CO2, CO and VOC concentration levels in classrooms
The descriptive statistics of the concentration levels of CO, CO2 and
VOC in each of the nine schools of measurement are presented in
Table 3. The increased concentrations of CO in some of the cases could
be possibly linked to the vehicle emissions from adjoining streets. The

Table 3
Descriptive statistics of CO, CO2 and VOC in each of the nine schools.

CO (ppm) CO2 (ppm) VOC (ppm)

School N Min. Max. Average Std. dev. Min. Max. Average Std. dev. Min. Max. Average Std. dev.

14 399 0.0 0.8 0.1 0.1 556 2910 1219 604 0.0 9.4 1.0 1.1
1 428 0.0 0.9 0.1 0.2 538 5049 2082 933 0.0 5.9 1.1 1.0
4 429 0.0 1.4 0.2 0.4 546 3547 1105 601 0.0 5.7 1.1 1.0
3 191 0.0 1.1 0.2 0.3 577 2364 1209 461 0.0 5.0 1.5 1.1
18 89 0.0 0.0 0.0 0.0 566 2385 1118 536 0.0 0.8 0.2 0.2
12 406 0.2 4.0 1.3 0.8 558 4365 1437 942 0.0 51.9 6.4 9.0
2 424 0.2 8.3 2.9 2.2 587 1729 893 205 0.0 27.2 6.6 6.8
8 268 1.3 11.1 7.2 2.0 573 2207 1018 301 0.3 39.9 7.8 6.4
11 81 4.2 13.9 12.1 2.3 689 1172 971 135 3.0 39.7 15.5 8.7
564 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570

4,500 30

4,000
25
3,500

CO2 concentrations (ppm)

Number of students
3,000 20

2,500
15
2,000

1,500 10

1,000
5
500 CO2

0 0
10:08 # of
10:28
10:48
11:08
11:28
11:48
12:08
12:28
12:48
13:08
13:28
13:48
7:48
8:08
8:28
8:48
9:08
9:28
9:48

students
Time

Fig. 10. Diurnal variation of CO2 concentrations as a function of the number of students and the ventilation rates (8/4/13, School 1).

findings of the CO2 concentrations of this experiment are about the
same with the ones of Madeco et al., 2013 (999 ppm), Mumovic et al.,
2009 (average: 1154 ppm) and Santamouris et al., 2008 (1070 ppm)
and are slightly lower compared to the average CO2 concentrations
measured by Haverineen-Shaughnessy et al., 2011 (1799 ppm) in 100
classrooms. The increased indoor CO2 concentrations in certain class-
rooms (e.g. school 1) could be attributed to overcrowded classrooms
and to inadequate ventilation rates. The use of marker boards (please
refer to Table 1) in schools 12, 2 and 8 could have lead to the increased
VOCs concentrations. On the other hand, even though school 11 used
chalk white boards, the VOC levels were high, perhaps due to the use
of cleaning and disinfecting chemicals for the classroom's cleaning,
even though no particular odor was detected. However it is worth
mentioning that there do exist odorless VOCs and also that odor does
not always indicate the levels of risk from inhalation of chemicals. The
fact that CO and VOC concentrations in schools 8 and 11 were the
highest of all the other schools while at the same time CO2 levels were
of the lowest compared to the rest of the schools might seem odd;
however it could be related to the fact that these pollutants result
from different type of sources.
Apart from the ventilation rates, the CO2 concentration distributions
were also studied during the teaching and the non-teaching periods.
Fig. 9 presents the cumulative frequency distributions of CO2 concentra-
tions in all schools for the total period of measurement (blue), the
teaching period (pink crossed) and the non-teaching period (green). It
is obvious that the concentrations of the non-teaching period are
lower compared to the ones of the teaching period due to lower CO2
levels during the morning before the arrival of students, and also due
to lower occupancy density. In particular, the mean value of CO2 con-
centrations for the non-teaching period was 1055 ppm while for the
teaching period was 1482 ppm. These findings are in accordance with
the findings of Santamouris et al. (2008).
Fig. 10 indicates the diurnal variation of CO2 concentrations as a
function of the number of students of a representative day in school 1.
From about 8.10 LT when the students enter the classroom and the win-
dows are closed, the CO2 concentrations gradually increase to very high
concentrations even exceeding 3000 ppm. The concentrations begin to
drop at about 9.40 LT, time at which the first break starts, the windows
are fully opening and the students are coming out of their classrooms.
10 min after the windows are fully open (at about 9.50 LT), the ACH
from the background infiltration rate of 0.47 ACH reach 7.88 ACH and
lead to a decrease by more than four times in CO2 concentrations. At
about 10.00 LT the students enter the classrooms again after the
20 min break, the windows close and the concentrations are dramatically

4000 4000
3500
CO2 concentrations (ppm)

CO2 concentrations (ppm)

3500
3000 3000
2500 2500
y = 2115.1x-0.29
y = 1721.1x-0.297 R² = 0.5939
2000 2000
R² = 0.638
1500 1500
1000 1000
500 500
0 0
0 5 10 15 20 25 0 10 20 30 40 50
Ventilation rates (ACH) Ventilation rates (l/s/p)

Fig. 11. Scatter plot of the averaged per teaching hour CO2 concentrations and corresponding ventilation rates in ACH (left) and l/s/p (right).
 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570
Fig. 12. Indoor particulate matter concentration distributions in the 9 classrooms.
increasing. The same pattern is repeated until the end of the day's
measurements; however, the CO2 concentrations do not extend to very
high concentrations as in the first two teaching hours of the day, due to
the shorter duration of the last three teaching hours of the day. The
outdoor CO2 concentrations remained constant at about 400 ppm.
Fig. 11 presents a scatter plot of the CO2 concentrations and
corresponding air flow rates averaged per teaching hour (left: ACH,
right: l/s/p) in all of the schools. A moderate negative correlation stands
between CO2 and ventilation rates of the teaching hours (R2 = 0.638
and 0.5939 for ventilation in ACH and in l/s/p respectively) showing
that the CO2 concentrations seem to decrease exponentially for greater
ventilation rates. The main conclusion from this diagram is that ade-
quate ventilation rates can lead to the reduction of CO2 concentrations
indoors during the teaching hours. The same issue was examined for
the entire period of measurement however the correlations were not
significant. A reason for the correlation's strengthening through the
averaged values per teaching hour is that “averaging” filters out random
fast variations that are not influenced by air change but they are
random. During teaching hours the ventilation rates and the number
of students were usually constant. Therefore air change rates did not
change abruptly as they did during the non-teaching hours when very
fast air changes caused dynamic phenomena that go beyond a simple
linear correlation.
3.3.2. Analysis of PM concentrations in classrooms
The distributions of PM10, PM2.5 and PM1 for the entire period of
measurement in each of the 9 classrooms are presented in box plots in
Fig. 12 (Dorizas et al., 2013b). The two horizontal lines represent
the limit values of 50 μg/m3 and 25 μg/m3 for PM10 (red) and PM2.5
(green dashed) respectively (WHO, 2005b). For all schools the PM10
concentrations had similar profiles and they extended to very high
levels, exceeding by far the recommended limit value. The average
value of PM10 in all schools is 244.5 μg/m3 while Habil and Taneja
(2011) found 524.76 and 153.37 μg/m3 average values for winter and
summer respectively. The increased concentrations of particles of great-
er size could be attributed to the increased levels of indoor dust and to
565
the presence of students whose movement makes the dust raise. Inade-
quate levels of ventilation could be also linked to increased concentra-
tions of PM. PM2.5 and PM1 concentration distributions presented
differences from school to school and there were also many cases that
the PM2.5 exceeded their limit value. The vehicle emissions from the ad-
joining streets could raise the concentrations of particles of smaller size.
The descriptive statistics of all the sizes of the measured particulate
matter indoors in each of the nine schools are summarized in Table 4.
It is obvious that the concentration ranges for most of the particulate
matter size ranges vary significantly from school to school.
The averaged indoor concentration distribution of PM in all the
schools for the teaching (pink) and the non-teaching (striped blue
azure) periods are shown in box plots in Fig. 13. During the teaching
period the concentrations were slightly greater compared to the non-
teaching period. This difference is increasing for particles of greater
size. The reason for this could be the fact that particles of greater size
are affected a lot by the presence of students during the teaching
hours and ventilation rates during non teaching hours, contrary to the
particles of smaller size whose sources are mainly combustion processes.
The diurnal variation of the indoor (purple) and outdoor (orange
crossed) PM10 concentrations of a typical day in school 1 (same as
Fig. 10 for CO2) is presented in Fig. 14 versus the number of students
and the ventilation rates expressed in ACH. At about 8.10 LT the
students enter the classroom, the windows are closed and the PM10
concentrations gradually increase. At around 9:40 LT, time of the first
break of the day, the increase of the ventilation rates from 0.47 ACH (in-
filtration with windows closed) to 7.88 ACH (fully open windows)
through the window openings results in a significant drop of PM10
concentrations. It is worth mentioning that at the time of the break
the students were away from the classrooms. The same pattern is re-
peated for the next long teaching hour of the day which starts at around
10.00 LT and lasts until 11.30 LT. As the duration of the third, fourth and
fifth teaching hours of the day are shorter compared to the first two
teaching hours, the concentrations do not extend to very high levels.
The outdoor concentrations remain in rather low levels (b20 μg/m3)
throughout the day. Overall, it is obvious that the diurnal fluctuations
566 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570

of PM10 concentrations and the number of students being in the class- Table 4
room are positively correlated while the concentrations are negatively Descriptive statistics of PM indoors in each of the nine schools.

correlated to the ventilation rates through the window openings. School Min Max Average Median St.dev

PM0.5 14 0.66 7.60 2.77 2.55 1.58

3.4. Indoor to outdoor (I/O) ratios of PM concentrations 1 0.39 11.53 4.39 4.22 2.22
4 1.73 19.57 4.26 3.17 3.27
The I/O ratios of PM10 and PM2.5 for the entire period of measure- 3 3.04 9.23 6.33 6.28 1.57
18 6.65 13.70 9.76 10.00 2.25
ment in each of the 9 schools are presented in Table 5. For both of the
12 1.57 13.71 6.93 6.66 3.37
size ranges in all of the schools the ratios are greater than 1 indicating 2 0.53 6.87 2.09 2.02 0.90
the strong presence of indoor PM sources (e.g. human presence and 8 0.62 3.88 1.36 1.07 0.83
movement, use of chalk white boards) and that the outdoor environ- 11 1.44 2.23 1.69 1.62 0.21
ment does not significantly affect the indoor one. The I/O ratios of this PM1 14 1.38 15.78 5.80 5.31 3.01
1 0.82 18.13 8.35 8.66 3.50
study exceed by far the corresponding ratios measured by Madeco 4 2.60 28.05 6.53 5.25 4.43
et al., 2013 in primary schools which ranged between 0.9–2.8 and 0.9– 3 4.38 13.71 9.61 9.56 2.31
2.5 for the heating and cooling seasons respectively. 18 10.44 18.61 14.21 15.31 2.71
The I/O ratios of PM10 and PM2.5 are further presented in box plots 12 2.21 20.85 9.98 9.53 4.75
2 2.41 18.66 5.52 5.15 2.25
for the teaching (pink) and non-teaching (blue azure striped) periods
8 1.16 5.98 2.57 2.10 1.31
in Fig. 15. It is obvious that the distributions of the ratios during the 11 2.35 4.73 2.91 2.63 0.69
teaching periods are greater compared to the corresponding ones of PM2.5 14 4.31 68.78 20.25 17.46 11.43
the non-teaching periods. This finding is in accordance with the findings 1 3.00 50.87 26.88 28.80 11.06
of Alshitawi et al. (2009). The differences between the I/O ratios in the 4 4.51 37.64 14.32 13.38 6.09
3 7.41 52.47 20.43 21.18 6.52
two periods are greater for the ratios of PM10.
18 16.11 42.72 26.42 25.90 5.40
The indoor to outdoor ratios of PM are also studied as a function of 12 3.80 32.15 16.73 16.86 5.84
the CO2 concentrations. Fig. 16 presents a scatter plot of the averaged 2 6.13 44.19 21.60 21.92 8.97
CO2 and corresponding I/O ratios per school. Moderate positive correla- 8 2.82 18.41 7.17 6.56 3.68
11 5.48 14.50 7.61 6.72 2.47
tions stand between the I/O ratios of PM10 and PM2.5 and the indoor CO2
PM5 14 30.07 709.97 163.92 147.19 106.12
concentrations. One could conclude that the indoor CO2 concentrations 1 11.61 594.73 225.43 239.12 114.87
have common behavior with both the internal and external concentra- 4 20.51 338.88 109.63 105.34 54.94
tions of PM10 and PM2.5. This common behavior could have as a driving 3 21.54 605.23 145.07 132.03 80.10
force the common sources of the two pollutants mentioned before, such 18 60.90 445.41 174.88 140.93 93.42
12 19.68 220.60 95.34 89.48 41.10
as the inadequate ventilation rates and the overcrowded classrooms.
2 20.62 287.21 143.91 155.92 67.47
8 13.27 202.83 49.92 48.72 30.32
3.5. Effects of chalk and marker on air pollutant concentrations 11 30.46 71.66 49.52 48.28 10.93
PM10 14 55.35 1618.53 316.23 282.71 212.06
1 27.06 1603.72 430.56 454.07 219.44
In this section, the schools are separated to the ones using chalk
4 54.46 687.23 223.21 215.54 108.25
blackboards (codes: 14, 1, 4, 3, 18, 11) and to the ones using marker 3 37.72 1371.89 286.09 257.83 175.27
whiteboards (codes: 12, 2, 8, please refer to Table 1 columns 1 & 8). 18 112.47 1093.13 360.05 290.85 214.59
The ratios of the averaged parameters in the schools using chalk to the 12 38.68 439.25 192.87 186.24 84.69
schools using marker for the entire period of measurement, are present- 2 39.33 509.33 206.85 225.33 91.15
8 21.91 526.40 91.78 87.47 63.30
ed in Fig. 17. The ratios are greater than 1 for the cases of PM, CO2 and
11 48.48 136.87 92.87 90.17 21.97
ACH, while are lower than 1 for T, CO and VOCs. The number of people TPM 14 57.52 1867.23 349.96 316.73 234.05
was approximately the same in the two cases while the relative humid- 1 38.61 1827.36 479.48 496.07 240.81
ity (RH) was slightly greater in the schools using chalk. Overall it was 4 60.56 765.53 250.09 242.69 119.81
3 41.42 1602.65 335.13 303.89 205.71
found that chalk black boards lead to increased levels of PM even
18 120.60 1294.58 402.38 320.15 246.99
though the ventilation rates were greater, while marker white boards 12 42.56 503.14 220.88 213.56 97.82
lead to increased VOCs and CO concentrations. One could conclude 2 42.41 661.60 226.20 240.87 97.20
that the gain from the reduction of CO and VOCs is lost in the increase 8 23.20 695.87 110.85 106.49 78.36
of PM when using chalk blackboards. 11 52.27 169.29 112.12 111.13 29.03
UFP 14 3062 24292 9494 9025 3871
1 751 47060 5820 3506 5410
3.6. Pollutant limit value exceedance 4 1493 12818 3502 3290 1627
3 1357 36641 8961 6117 7430
Table 6 summarizes the percentages of exceedance of the recom- 18 1792 3985 2445 2327 518
mended limit values of certain pollutants for the entire measurement 12 1018 15693 4112 3551 2486
2 790 70718 6496 4830 5569
period in a scale of colors. The percentages of greater than 80% are 8 883 18937 5572 4330 4368
marked with red, the ones between 60 and 80% are marked with orange, 11 769 8729 3851 3745 2695
the percentages between 40 and 60% are shown in yellow, the ones
between 20 and 40% are shown in green and the ones below 20% are
marked in blue. According to Molhave (1990) the discomfort scale of attributed to the usage of marker in white boards. The great majority
exposure to VOCs with possible occurrence of headaches starts when of the CO concentrations were below their limit value. CO2 concentra-
the VOC concentrations are from 0.8 ppm and above. According to the tions exceeded the limit value of 1000 ppm in all schools. In school 1
same scale, when the concentrations are between 0.8 and 6.64 ppm the limit values were exceeded by more than 80% of the measurements
they may cause headaches to the building's occupants, while concentra- period which can be linked to the fact that school 1 had the windows
tions greater than 6.64 ppm can cause more serious health effects such closed for more than 80% of the time of the measurements (paragraph
as neurological problems. VOCs exceeded the limit value of 0.8 ppm 3.1.1 Fig. 6). Therefore, the ventilation rates were low and the CO2 con-
in all the schools apart from school 18. The greater percentages of centrations were trapped in the indoor environment. Seppanen et al.
exceedance of the VOC limit values in schools 12, 2 and 8 could be (1999) found that CO2 concentrations below 1,000 ppm are not always
 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570 567

Fig. 13. Particulate matter concentration distributions in teaching and non-teaching periods.

adequate for the removal of indoor air pollutants. The percentages of 4. Conclusions
exceedance of the recommended limit value of PM2.5 were rather low
for the majority of the schools. The highest percentage was in school 1 The main conclusions arisen from this study are:
(66%) which could be attributed to the inadequate ventilation rates.
The PM10 concentrations in all the schools for approximately the entire 1. The average ventilation rates in all schools were in general greater
period of measurement were above their limit values. Only in school than the minimum recommended ventilation rates, showing that
8 the percentage was slightly lower (74%) compared to the others. the classrooms were well ventilated for certain cases.
This could be attributed to the less students per meter squared in the 2. The increased infiltration rates for certain classrooms were attributed
classroom compared to the other schools and also to the use of marker to increased values of wind speed and also to air leakages due to
instead of chalk board. older building constructions.

1,000 30
Indoor & outdoor PM10 concentrations

Ventilation rates (ACH) & Number of

25

100 20
students
(µg/m3)

15

10 10

PM10-in
1 0
PM10-out

ACH
Time
#of students

Fig. 14. Diurnal variation of indoor and outdoor PM10 concentrations as a function of the number of students and the ventilation rates (8/4/13, School 1).
568 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570

Table 5 2500

CO2 concentrations (ppm)

Averaged I/O ratios of PM10 and PM2.5 in each of the schools of measurement.

Schools I/O-PM10 ratios I/O-PM2.5 ratios 2000

y = 199.07x + 693.01
14 16.41 2.78 R² = 0.5373 I/O PM10
1 36.42 4.97 1500
y = 31.248x + 807.61
4 19.11 2.83 R² = 0.5789 I/O PM2.5
3 20.00 4.14 1000
18 17.18 3.80
12 12.16 2.92
2 2.09 1.02 500
8 3.41 1.02
11 4.25 1.36 0
0 10 20 30 40
Indoor to Outdoor PM ratios
3. The ventilation rates and the presence of students significantly af-
fected the concentration profile of CO2. CO concentrations for most Fig. 16. Correlation between I/O ratios of PM and CO2 concentrations.
of the schools were below their limit values and VOC concentrations
were higher in the classrooms using marker white boards.
4. The average indoor concentrations of PM10 exceeded their recom- ventilation measurements were performed at the end of the measure-
mended limit value by more than 4 times for most of the schools. ment day which might introduce an error since the presence of students
Also, PM concentrations during the teaching periods were higher could affect the air mixing and the fact that blower door measurements
compared to the ones of the non teaching period possibly attributed were not performed in order to further estimate the air tightness of the
to the presence of students. It was further found that appropriate classrooms. Also the infiltration rate values strongly depend on the
ventilation rates contributed to the reduction of the indoor PM levels. weather conditions which are changing during the day. Based on the
5. The indoor to outdoor (I/O) ratio of PM was greater than 1 for all the above some suggestions could be made for future work including a
cases indicating that the outdoor environment does not really affect greater number of schools so as to enhance the statistical sample, the
the indoor one possibly due to the intense presence of indoor PM involvement of all year around measurements so as to evaluate the per-
sources. Furthermore, a positive moderate correlation was found formance of natural ventilation, the continuous monitoring of wind
between the I/O ratios of PM10 and PM2.5 to the CO2. speed and direction during the experiment for all schools as well as
6. Finally, it was found that the use of chalk and marker boards had a the use of blower door measurements in order to estimate the air tight-
significant effect on certain indoor air pollutants. In particular, it ness of the classrooms.
was found that in classrooms where marker was used had higher
VOC and CO concentrations while classrooms using chalk suffered Acknowledgments
from increased levels of PM concentrations.
The authors would like to acknowledge all of the participants in the
It is worth to mention that the realization of the experiments incor- experimental campaign. In particular, we gratefully acknowledge the
porates some limitations of the methodology such as the fact that the contribution of John Syfnaios and Katerina Stathi in the experimental

Fig. 15. Distributions of I/O ratios of PM during teaching and non-teaching hours for all of the schools and the entire period of measurement.
 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570 569

2 Canha N, Almeida M, Freitas M, Taubel M, Hanninen O. Winter ventilation rates at prima-

ry schools: comparison between Portugal and Finland. J Toxicol Environ Health A
2013;76:400–8. http://dx.doi.org/10.1080/15287394.2013.765372.
1,8 Chaloulakou A, Mavroidis I, Duci A. Indoor and outdoor carbon monoxide concentration
relationships at different microenvironments in the Athens area. Chemosphere
concentrations: (classes with chalk)/
1,6 2003;52:1007–19. http://dx.doi.org/10.1016/S0045-6535(03)00263-7.
Chau CK, Hui WK, Tse MS. Valuing the health benefits of improving indoor air quality in
residences. Sci Total Environ 2008;394:25–38. http://dx.doi.org/10.1016/j.scitotenv.
Ratio of average pollutant

1,4 2008.01.033.
( classes with marker)

Demirel G, Özden Ö, Döğeroğlu T, Gaga EO. Personal exposure of primary school children
to BTEX, NO2 and ozone in Eskişehir, Turkey: relationship with indoor/outdoor
1,2 concentrations and risk assessment. Sci Total Environ 2014;473–474:537–48.
http://dx.doi.org/10.1016/j.scitotenv.2013.12.034.
1 Diapouli E, Chaloulakou A, Spyrellis N. Indoor and outdoor particulate matter concentra-
tions at schools in the Athens area. Indoor Built Environ 2007;16(1):55–61. http://dx.
PM0.5
PM1
PM2.5
PM5
PM10
TPM
CO2
ACH
#people
RH
T
CO
VOC
doi.org/10.1177/1420326X06074836.
0,8 Dorizas PV, Kapsanaki-Gotsi E, Assimakopoulos MN, Santamouris M. Correlation of
particulate matter with airborne fungi in schools in Greece. Int J Vent 2013a;12(1):
1–16. http://dx.doi.org/10.5555/2044-4044-12.1.1.
0,6 Dorizas PV, Assimakopoulos MN, Helmis C, Santamouris M. Analysis of the indoor air
quality in Greek primary schools. 34th AIVC-3rd TightVent-2nd Cool Roofs'-1st
0,4 Venticool Conference, Athens, Greece25-26 September 2013; 2013b. p. 110–3.
Fromme H, Lahrz T, Hainsch A, Oddoy A, Piloty M, Ruden H. Elemental carbon and
respirable matter in the indoor air of apartments and nursery schools and ambient
0,2 air in Berlin. Indoor Air 2005;15:335–41. http://dx.doi.org/10.1111/j.1600-0668.
2005.00377.x.
Fromme J, Twardella D, Dietrich S, Heitmann D, Schierl R, Liebl B, et al. Particulate matter
0 in the indoor air of classrooms-exploratory results from Munich and surrounding
area. Atmos Environ 2007;41:854–66. http://dx.doi.org/10.1016/j.atmosenv.2006.
Fig. 17. Ratio of averaged values in classrooms using chalk to the ones using marker. 08.053.
Geodata. Available at: http://www.geodata.gov.gr/maps/, 2010. Accessed on September
20th, 2013.
Goyal RG, Khare M. Indoor–outdoor concentrations of RSPM in classroom of a naturally
ventilated school building near an urban traffic roadway. Atmos Environ 2009;43:
Table 6 6026–38. http://dx.doi.org/10.1016/j.atmosenv.2009.08.031.
Percentage of exceedance of the recommended limit values of pollutants (%) for the Guo H, Morawska L, He C, Zhang YL, Ayoko G, Cao M. Characterization of particle number
measurement period. concentrations and PM2.5 in a school: influence of outdoor air pollution on indoor
air. Environ Sci Pollut Res Int 2010;17(6):1268–78. http://dx.doi.org/10.1007/
CO VOC CO2 PM2.5 PM10
s11356-010-0306-2.
School >10 ppm >0.8 ppm >1000 ppm >25 mg/m3 >50 mg/m3 Habil M, Taneja A. Children's exposure to indoor particulate matter in naturally ventilates
14 0 43 44 31 100 schools in India. Indoor Built Environ 2011;20(4):430–48. http://dx.doi.org/10.1177/
1420326X11409455.
1 0 49 81 66 96 Hagler GSW, Baldauf RW, Thoma ED, Long TR, Snow RF, Kinsey JS. Ultrafine particles near
4 0 52 33 4 100 a major roadway in Raleigh, North Carolina: Downwind attenuation and correlation
with traffic-related pollutants. Atmos Environ 2009;43:1229–34. http://dx.doi.org/
3 0 69 57 21 97
10.1016/j.atmosenv.2008.11.024.
18 0 0 29 57 100 Haverineen-Shaughnessy U, Moschandreas DJ, Shaughnessy RJ. Association between
12 0 71 42 11 100 substandard classroom ventilation rates and students' academic achievement. Indoor
Air 2011;21:121–31. http://dx.doi.org/10.1111/j.1600-0668.2010.00686.x.
2 0 80 23 35 98
Helmis CG, Tzoutzas J, Flocas HA, Halios CH, Stathopoulou OI, Assimakopoulos VD,
8 12 93 43 0 74 et al. Indoor air quality in a dentistry clinic. Sci Total Environ 2007;377:349–65.
http://dx.doi.org/10.1016/j.scitotenv.2007.01.100.
11 89 100 49 0 98
Helmis CG, Assimakopoulos VD, Flocas HA, Stathopoulou OI, Sgouros G, Hatzaki M. Indoor
air quality assessment in the Air Traffic Control Tower of the Athens Airport, Greece.
Environ Monit Assess 2009;148:47–60.
Hussein T, Paasonen P, Kulmala M. Activity attern of a selected group of school occupants
campaign and also the schools' directors, teachers and students for their and their family members in Helsinki-Finland. Sci Total Environ 2012;425:289–92.
Jamriska M, Thomas S, Morawska L, Clark BA. Relation between indoor and outdoor
support during the monitoring period. This research has been co- exposure to fine particles near a busy arterial road. Indoor Air 1999;11:35–48.
financed by the European Union (European Social Fund—ESF) and http://dx.doi.org/10.1111/j.1600-0668.1999.t01-2-00002.x.
Greek national funds through the Operational Program ‘Education and Jantunen M, Oliveira Fernandes E, Carrer P, Kephalopoulos S. Promoting actions for
healthy indoor air (IAIAQ). Luxembourg: European Commission Directorate General
Lifelong Learning’ of the National Strategic Reference Framework
for Health and Consumers; 2011.
(NSRF)—Research Funding Program: Heracleitus II (MIS 346942). Johnson T, Myers J, Kelly T, Wisbith A, Ollisonc W. A pilot study using scripted ventilation
Investing in knowledge society through the European Social Fund. conditions to identify key factors affecting indoor pollutant concentration and air
exchange rate in a residence. J Expo Anal Environ Epidemiol 2004;14(1):1–22.
Jones NC, Thornton CA, Mark D, Harrison RM. Indoor/outdoor relationships of particulate
References matter in domestic homes with roadside, urban and rural locations. Atmos Environ
2000;34:4375–83. http://dx.doi.org/10.1016/S1352-2310(99)00489-6.
Alshitawi M, Awbi H, Mahyuddin N. Particulate matter mass concentration (PM10) under Kim HH, Kim CS, Lim YW, Suh MA, Shin DC. Indoor and outdoor air quality and its relation
different ventilation methods in classrooms. Int J Vent 2009;8(2):93–108. to allergic diseases among children: a case study at a primary school in Korea. Asian J
Annesi-Maesano I, Agabiti N, Pistelli R, et al. Subpopulations at increased risk of adverse Atmos Environ 2010;4–3:157–65. http://dx.doi.org/10.5572/ajae.2010.4.3.157.
health outcomes from air pollution. Eur Respir J Suppl 2003;40:57–63. http://dx. Madeco A, Magalhaes O, Brito A, Mayan O. Characterization of indoor environmental
doi.org/10.1183/09031936.03.00402103. quality in primary schools in Maia: a Portuguese case study. Hum Ecol Risk Assess
Assimakopoulos MN, Dounis A, Spanou A, Santamours M. Indoor air quality in a metro- 2013;19:126–36. http://dx.doi.org/10.1080/10807039.2012.683751.
politan area metro using fuzzy logic assessment. Sci Total Environ 2013;449:461–9. Mendell MJ, Heath GA. Do indoor pollutants and thermal conditions in schools influence
http://dx.doi.org/10.1016/j.scitotenv.2012.12.043. student performance? A critical review of the literature. Indoor Air 2005;15:27–52.
Bai Z, You Y, Gao X, Yu H, Yan L, Zhu T. Comparison of time-activity patterns in http://dx.doi.org/10.1111/j.1600-0668.2004.00320.x.
different patterns in different periods in Tianjin, China. Epidemiology 2007;17(6): Molhave L. Volatile organic compounds, indoor air quality and health. Indoor Air 1990;4:
S454. http://dx.doi.org/10.1097/00001648-200611001-01218. 357–76. http://dx.doi.org/10.1111/j.1600-0668.1991.00001.x.
Bakke J, Wieslander G, Norback D, Moen B. Eczema increases susceptibility to Morawksa L, He C, Hitchins J, Gilbert D, Parappukkaran S. The relationship between
PM10 in office indoor environments. Arch Environ Occup Health 2012;67(1): indoor and outdoor airborne particles in the residential environment. Atmos Environ
15–21. http://dx.doi.org/10.1080/19338244.2011.564236. 2001;35:3463–73. http://dx.doi.org/10.1016/S1352-2310(01)00097-8.
Bartlett K, Martinez M, Bert J. Modeling of occupant-generated CO2 dynamics in naturally Mumovic D, Davies M, Ridley I, Altamirano-Medina H, Oreszczyn T. A methodology for
ventilated classrooms. J Occup Environ Hyg 2004;1(3):139–48. http://dx.doi.org/10. post-occupancy evaluation of ventilation rates in schools. Build Serv Eng Res Technol
1080/15459620490424393. 2009;30(2):143–52. http://dx.doi.org/10.1177/0143624408099175.
570 P.V. Dorizas et al. / Science of the Total Environment 502 (2015) 557–570
Myhrvold AN, Olsen E, Lauridsen O. Indoor environment in schools—pupils health and
performance in regard to CO2 concentrations. Proceedings of seventh international
conference on indoor air quality and climate, 4. ; 1996. p. 369–74.
Nantka M. Comparison of different methods for airtightness and air change rate determi-
nation. In: Sherman MH, editor. Air change rate and airtightness in buildings, ASTM
STP 1067. Philadelphia: American Society for Testing and Materials; 1990. p. 267–82.
Park J, Jang Y, Ahn Y, Cheong S, Lee J. Estimation of appropriate capacity of ventilation
system based on the air infiltration rate in Korean classrooms. J Mech Sci Technol
2008;22:788–97. http://dx.doi.org/10.1007/s12206-008-0111-8.
Pope III CA, Dockery DW. Health effects of fine particulate air pollution: lines that connect.
JAMA 2006;56:709–42. http://dx.doi.org/10.1080/10473289.2006.10464485.
Santamouris M, Mihalakakou G, Patarias P, Gaitani N, Sfakianaki K, Papagralstra M, et al.
Using intelligent clustering techniques to classify the energy performance of school
buildings. Energy Build 2007;39:45–51. http://dx.doi.org/10.1016/j.enbuild.2006.04.
018.
Santamouris M, Synnefa A, Assimakopoulos M, Livada I, Pavlou K, Papaglastra M, et al.
Experimental investigation of the air flow and indoor carbon dioxide concentration
in classrooms with intermittent natural ventilation. Energy Build 2008;40:1833–43.
http://dx.doi.org/10.1016/j.enbuild.2008.04.002.
Sarigiannis DA, Soulakellis NA, Sifakis NI. Information fusion for computational assess-
ment of air quality and health effects. Photogramm Eng Remote Sens 2004;70(2):
235–45.
Sarigiannis DA, Karakitsios SP, Gotti A, Papaloukas CL, Kassomenos PA, Pilidis GA.
Bayesian algorithm implementation in a real time exposure assessment model on
benzene with calculation of associated cancer risks. Sensors 2009;9(2):731–55.
Seppanen OA, Fisk WJ, Mendel MJ. Association of ventilation rates and CO2 concentrations
with health and other responses in commercial and institutional buildings. Indoor Air
1999;9:333–41.
Shaughnessy RJ, Haverinen-Shaughnessy U, Nevalainen A, Moschandreas D. A prelimi-
nary study on the association between ventilation rates in classrooms and student
performance. Indoor Air 2006;16:465–8. http://dx.doi.org/10.1111/j.1600-0668.
2006.00440.x.
Sherman M. Tracer-gas techniques for measuring ventilation in a single zone. Build
Environ 1990;25(4):365–74. http://dx.doi.org/10.1016/0360-1323(90)90010-O.
Sherman M, Dickerhoff D. A multigas tracer system for multizone air flow measurements.
Proceedings ASHRAE/DOE/BTECC conference, Thermal Performance of the Exterior
Envelopes of Buildings Conference IV. American Society of Heating, Refrigerating
and Air-conditioning Engineers; 1989. p. 417–32.
Silvers A, Florence BT, Rourke DL, Lorimor RJ. How children spend their time—a sample
survey for use in exposure and risk assessment. Risk Anal 1994;14(6):931–44.
http://dx.doi.org/10.1111/j.1539-6924.1994.tb00062.x.
Siskos PA, Bouba KE, Stroubou ApA. Determination of selected pollutants and measure-
ment of physical parameters for the evaluation of indoor air quality in school build-
ings in Athens, Greece. Indoor Built Environ 2001;10(3–4):185–92. http://dx.doi.
org/10.1159/000049235.
Synnefa A, Polichronaki E, Papagiannopoulou E, Santamouris M, Mihalakakou G, Doukas
P, et al. An experimental investigation of the indoor air quality in fifteen school build-
ings in Athens, Greece. Int J Vent 2003;2(3):185–201. [DOI: http://www.ijovent.org/
doi/abs/10.5555/ijov.2003.2.3.185?journalCode=ijov].
Teli D, Patrick A, James B, Jentsch MF. Thermal comfort in naturally ventilated primary
school classrooms. Build Res Inf 2013;41(3):301–16. http://dx.doi.org/10.1080/
09613218.2013.773493.
Tippayawong N, Khuntong P, Nitatwichit C, Khunatorn Y, Tantakitti C. Indoor/outdoor
relationships of size-resolved particle concentrations in naturally ventilated school
environments. Build Environ 2009;44:188–97. http://dx.doi.org/10.1016/j.buildenv.
2008.02.007.
Wargocki P, Wyon DP. The effects of outdoor air supply rate and supply air filter condition
in classrooms on the performance of schoolwork by children. HVAC&R Res 2007;
13(2):165–91. http://dx.doi.org/10.1080/10789669.2007.10390950.
Wargocki P, Wyon DP, Matysiak B, Irgens S. The effects of classroom air temperature and
outdoor air supply rate on the performance of school work by children. Proceedings
of indoor air, 1. Beijing, the 10th international conference on indoor air quality and
climate; 2005. p. 368–72.
WHO, World Health Organization. Children's health and the environment—a global
perspective. A resource guide for the health sector; 2005a.
WHO, World Health Organization. Air quality guidelines- global update. World Health
Organization Regional Office for Europe. Denmark: Copenhagen; 2005b.
Yoon C, Lee K, Park D. Indoor air quality differences between urban and rural preschools
in Korea. Environ Sci Pollut Res 2011;18:333–45. http://dx.doi.org/10.1007/s1135.