Investigation and Evaluation of Winter Indoor Air Quality of Primary Schools in Severe Cold Weather Areas of China

The indoor air quality (IAQ) in classrooms has attracted more and more attention. Unfortunately, there is limited information relating to IAQ in the primary schools in severe cold weather areas of China. In this study, a field investigation on the IAQ of a primary school of Shenyang in northeast China was carried out by physical measurements and questionnaire surveys. The carbon dioxide (CO2) concentration in selected classrooms was continuously measured for a week, and the corresponding ventilation rate was calculated. Meanwhile, the perceptions of the IAQ, the purpose and the comfort degree of window opening have also been recorded from 106 pupils, aged 9–12. The results indicate the ventilation rate is considerably inadequate in about 99% of the class time due to the low frequency of window opening. The average daily CO2 concentration in these classrooms is 1510–3863 ppm, which is far higher than the recommended value of 1000 ppm. Most pupils understand that the purpose of opening windows in winter is to improve air quality. However, there are big differences between the measurement results and subjective judgments of indoor air quality. Contrary to the high measured CO2 concentration, around 70% pupils consider the air fresh, and only 3.7% pupils are dissatisfied and even very dissatisfied with IAQ in their classroom. It is necessary to change the existing manual window opening mode, because the pupils’ subjective judgment affects the window opening behavior.


Introduction
More and more studies have been written about the indoor air quality (IAQ) in primary and secondary school buildings [1].The health and performance of occupants are affected by indoor air quality, which affects children more than adults [2][3][4].In addition, classrooms are densely occupied environments [5,6], and most students are usually in school for 5-9 hours every weekday [7][8][9][10].Most of this time is spent in the classroom, with long periods of time spent sitting in chairs.Research shows that a good indoor environment can reduce absence rates [11,12] and improve students' learning efficiency [13][14][15][16].The air quality and temperature in the classroom are important factors in the learning process, and improving the air quality and temperature in schools is as important to student success as improving the level of teaching materials and teaching methods [17].Unfortunately, poor indoor air quality in primary schools has become a global issue, with problems including low ventilation rates, insufficient fresh air, excessive indoor particle and CO 2 concentration levels [6,9,10,[18][19][20][21].Many scholars regard CO 2 concentration as a key index for evaluating classroom air quality [21][22][23][24][25][26][27][28].The method of evaluating indoor air quality based on CO 2 measurement is convenient and inexpensive, and reasonable and accurate measuring and logging instruments can be provided [29].Moreover, high CO 2 concentration affects adult decision-making performance and neuro-physiological symptom [30,31].Simultaneously, high CO 2 concentrations are related to students' performance in class, affecting their learning efficiency, health status and absenteeism rate.Twardella et al. [23] showed that low air quality by increasing CO 2 concentration did not reduce students' short-term performance, but the error rate increased.Further, higher CO 2 concentrations can reflect inadequate ventilation.The research of Fisk [1] showed that indoor CO 2 concentrations above 1000 ppm indicate a ventilation rate of less than 7 L/s per person.Bako-Biro et al. [7] found that the CO 2 concentrations in some UK schools were much higher than recommended levels, and that low ventilation rates significantly reduced students' memory and alertness.The findings of Tofum et al. [32] have provided more evidence that inadequate classroom ventilation affects learning outcomes, so that classroom air quality should be a major priority in building or remodeling existing schools.Some studies have shown that the CO 2 concentration in passive ventilation classrooms and mechanical ventilation classrooms exceeds the recommended standard [33,34]; however, the CO 2 concentration under the condition of mechanical ventilation is significantly better than that under the condition of natural ventilation [35,36].The maximum CO 2 concentration in naturally ventilated classrooms was generally 4-6 times higher than the reference value [7,8,18,20,22,26,27,37,38].
At present, natural ventilation is widely used in primary school buildings in severe cold weather regions of China, even during the winter heating season.This ventilation mode mainly depends on opening windows manually (without any mechanical equipment).The characteristics of China's monsoon climate make cold regions colder in the winter, and the cold is more persistent than in other regions of the world at the same latitude.Thus, the heating intensity in these areas of China is much higher than that of European and American cities located in the same latitude [39].The demands of energy saving and heat preservation result in insufficient indoor ventilation, which seriously affects the indoor air quality.Some studies have begun to pay attention to the indoor environment of primary and secondary schools in China [40,41].However, there is a lack of on-site survey data regarding pupils' perception and evaluation of indoor air quality in the primary schools of China's severe cold regions.Correlation analysis between subjective evaluation and objective evaluation is even less available.
This study takes the teaching building of a primary school in Shenyang as the study object.Physical measurements and questionnaire surveys were conducted to investigate and evaluate the indoor air quality of naturally ventilated classrooms during the heating season in this severe cold area.The IAQ and ventilation capacity of three representative classrooms were assessed by means of weekly CO 2 concentration records.During the measurement period, survey questionnaires were implemented to analyze pupils' perception and satisfaction with the IAQ in their classrooms.At the same time, pupils' willingness to open windows in the winter was surveyed through collecting their opinions on the purpose and comfort of window opening.Finally, the differences between the subjective voting about air quality among classes and between genders, the correlation between subjective and objective evaluations, and the influence of subjective perceptions about window opening frequency were analyzed.

City Climate
Figure 1 shows that the longitude and latitude of Shenyang in northeast China are 123.38E and 41.8 N, respectively.In the global climate classification, Shenyang is placed in a region characterized by its temperate semi-humid continental climate.According to the design standard for energy efficiency of public buildings [42], Shenyang belongs to the severe cold area C of China.The heating season in Shenyang lasts for 150 days, from November 1 of each year to April 1 of the following year.The main Therefore, schools in Shenyang are expected to have greater heating than cooling needs 98 throughout the year.The naturally ventilated building A of a primary school in Shenyang was selected as a case 103 study (see Figure 2).The building A was constructed in 1999 with six floors, using a reinforced

Target Building
The naturally ventilated building A of a primary school in Shenyang was selected as a case study (see Figure 2).The building A was constructed in 1999 with six floors, using a reinforced concrete frame construction and brick walls without any thermal insulation materials.The layout, function and space design features of teaching building A are representative of the typical primary and secondary school buildings that have been built in the severe cold regions of China in recent decades.Building A is heated by hot water radiators during the heating season.All classrooms are ventilated by opening windows manually, and there are no mechanical ventilation facilities for use during any part of the year.There are 44 classrooms on the second to fourth floors of the building for grade 4 to grade 6 (pupils aged 8 to 13), of which 28 classrooms face south and 16 classrooms face north.The Building B is also a teaching building for grade 1 to grade 3 (pupils age 6-8).
Energies 2019, 12, x FOR PEER REVIEW 3 of 19 efficiency of public buildings [42], Shenyang belongs to the severe cold area C of China.The heating season in Shenyang lasts for 150 days, from November 1 of each year to April 1 of the following year.
The main climate indicator for the severe cold region is the average outdoor temperature of less than −10 ℃ in January.Meanwhile, the average outdoor temperature is less than 25 ℃ in July.
Therefore, schools in Shenyang are expected to have greater heating than cooling needs throughout the year.

Target building
The naturally ventilated building A of a primary school in Shenyang was selected as a case study (see Figure 2).The building A was constructed in 1999 with six floors, using a reinforced

Pupil and Classroom Information
Three 5th grade classrooms on the fourth floor of the teaching building were selected as the research object for data collection; classroom S1(S1) and classroom S2(S2) face south, and classroom N1(N1) faces north (Figure 3).These are representatives of classrooms with different orientations and locations (middle and side), and they house pupils in the same grade for consistency of comparison.Classrooms are typically crowded spaces, especially in China.When school is in session, between 32 and 38 pupils (aged 9 to 12) are present in each classroom.The per capita area of each classroom is 1.43-1.65 m 2 , which meets the standard 1.36 m 2 /p according to the code for design of school (GB50099-2011) [5], but this is much less than the per capita area of an average classroom in the United States [29] and in European countries [43].The characteristics of the classrooms studied are presented in Figure 4 and Table 1.

Pupil and classroom information
Three 5th grade classrooms on the fourth floor of the teaching building were selected as the research object for data collection; classroom S1(S1) and classroom S2(S2) face south, and classroom N1(N1) faces north (Figure 3).These are representatives of classrooms with different orientations and locations (middle and side), and they house pupils in the same grade for consistency of comparison.Classrooms are typically crowded spaces, especially in China.When school is in session, between 32 and 38 pupils (aged 9 to 12) are present in each classroom.The per capita area of each classroom is 1.43-1.65 m 2 , which meets the standard 1.36 m 2 /p according to the code for design of school (GB50099-2011) [5], but this is much less than the per capita area of an average classroom in the United States [29] and in European countries [43].The characteristics of the classrooms studied are presented in Figure 4 and Table 1.In China, the pupils will have all their lessons in the same classroom for the whole school term, except the music and experimental classes, which will be in specific music room and outdoor environment, respectively.The curriculum for each class is repeated weekly.research object for data collection; classroom S1(S1) and classroom S2(S2) face south, and classroom N1(N1) faces north (Figure 3).These are representatives of classrooms with different orientations and locations (middle and side), and they house pupils in the same grade for consistency of comparison.Classrooms are typically crowded spaces, especially in China.When school is in session, between 32 and 38 pupils (aged 9 to 12) are present in each classroom.The per capita area of each classroom is 1.43-1.65 m 2 , which meets the standard 1.36 m 2 /p according to the code for design of school (GB50099-2011) [5], but this is much less than the per capita area of an average classroom in the United States [29] and in European countries [43].The characteristics of the classrooms studied are presented in Figure 4 and Table 1.In China, the pupils will have all their lessons in the same classroom for the whole school term, except the music and experimental classes, which will be in specific music room and outdoor environment, respectively.The curriculum for each class is repeated weekly.

Field Measurements
Physical measurements were conducted in three classrooms (one indoor measuring point in each classroom) for a school week (from Monday to Friday), which is shown in Figure 3.The indoor physical parameters were measured by means of field measurements of the following factors: concentration of carbon dioxide (CO 2 ), air temperature (Ta) and relative humidity of the air (RH).The measured CO 2 concentration and the corresponding ventilation rates were compared with the standards to evaluate the indoor air quality and ventilation capacity of the classrooms, and the corresponding influencing factors were analyzed.

Measurement Instrument and Period
A HOBO (MX 1102) CO 2 logger (Onset, Bourne, MA, USA, Table 2) was used to monitor and collect values of all parameters every 5 min at each measuring point.To determine the best location for the instrument, before any measurements were made, the researchers placed two calibrated measuring instruments in the middle of S1 and on the back storage cabinet for one day to compare the measured data.Comparing the values of the two measurement points, the difference between the CO 2 concentration of the two locations is between ±100 ppm except for a few measurements (which may be disturbed by human activities).The level and trend of CO 2 concentration were similar, which indicates the characteristics of the change in overall CO 2 concentrations were similar in the classroom.Therefore, to minimize interference with the measuring instruments from the environment or by the pupils during the long-term measurement process, the instruments were placed on lockers at the back of three selected classrooms at a height of approximately 1 m above the floor, 2 m away from the windows and doors and 1m from the pupils (see Figure 3).The characteristics of the measuring instruments are presented in Table 2.The measurements were conducted from Monday, March 7th (starting at 07:00) to Friday, March 11th (ending at 18:00) of 2016, and included periods during which the classrooms were both occupied and unoccupied.According to the school's regulations, all classes were conducted from 08:00 to 16:10 from Monday to Friday (Wednesdays from 08:00 to 15:30), but most pupils entered the classroom at 07:30 and left at 16:30.Therefore, the occupied period was defined from 07:30 to 16:30, including periods called Class-Time (C-T) and Intermittent-Time (I-T).Unoccupied periods were defined as part of the time of a school week, except for the occupied periods.The timetable of a regular school day is shown in Table 3.
Class-Time (C-T) refers to the time pupils spend in the classroom according to the school's schedule, including taking courses (such as Chinese, mathematics, art, history, English and other courses) and engaging in self-study.Intermittent-Time (I-T) refers to the time between two C-T periods, including times for breaks, lunch and outdoor activities.The I-T periods of Break 1, 3, 4 and 6 lasting less than 15 mins are called short intervals.Break 2, 5 and Noon lasting 40 mins, 45 mins and 80 mins respectively are called long intervals.
During the measurement period, the frequencies with which doors and windows were opened in the three classrooms were observed and recorded manually.There was almost no window opening behavior in the three classrooms during the periods of C-T.Usually, windows were opened during I-T period by pupils according to their own ideas, but the frequency of window opening was random and short-time.
In fact, for most of the short I-T periods, many pupils stayed in the classrooms instead of going outside.It was difficult to record the changes in occupancy of the classrooms manually, because pupils entered and left the classrooms frequently.Similarly, the frequency of window openings was difficult to capture, because the pupils moved about or gathered in groups in the classrooms.Therefore, the observation data for window openings recorded by the observers may be inaccurate.

Reference Value
The lowest and highest outdoor temperatures during the measured period were obtained from the China Air Quality online Monitoring and Analysis Platform.There are 11 outdoor measurement points for Shenyang city on this platform.The data selected in this paper come from one of the measurement points which is about 3.5 kilometers nearest to the primary school.The range of outside temperatures between 07:00 and 17:00 were as follows: From Monday to Friday, the outdoor CO 2 concentration was measured by a handheld CO 2 test instrument (Telaire 7001, General Electric Company, Fairfield, Conn, USA) used in the center of the school playground at 7 a.m. and 5 p.m. respectively.The range of outdoor CO 2 concentration varies from 358 to 425 ppm, with an average of 384 ppm.380 ppm was used as the reference value of outdoor CO 2 concentration in this study.Usually, IAQ is evaluated by means of two main criteria: CO 2 concentration and ventilation rate.In primary schools, 20 m 3 /h per person (5.556 L/s per person) is recommended as the minimum ventilation rate, and 1000 ppm as the daily average value of CO 2 concentration [44].1080 ppm (700 ppm + outdoor CO 2 concentration 380 ppm) is recommended as the highest value allowed [29].Ventilation rates can be calculated from indoor and outdoor CO 2 measurements over suitable intervals using Equation (1) [25,26]: where C t is the internal concentration of carbon dioxide at time t (ppm); C ext is the external concentration of carbon dioxide (ppm); G is the generation rate of carbon dioxide in the space (cm 3 /s); Q is the internal-external air exchange rate (m 3 /s); C 0 is the initial concentration of carbon dioxide (ppm); V is the room volume (m 3 ); and t is the time (s).
During the unoccupied periods, the value of G is equal to zero.Equation ( 1) can be expressed as: The corresponding air exchange rate (AER) is: Using the method presented in [26,45], the average CO 2 emission rate per pupil is circa 0.0039 L/s, and that per teacher is circa 0.0054 L/s (The errors of the calculated values between the three classes are very small and can be neglected).
The variation rate (VR) of CO 2 concentration in each C-T period and I-T period is calculated by the difference between the initial and ending CO 2 concentrations during this period:

Questionnaire
During the time in which physical measurements were taken, questionnaire surveys were administered to 106 pupils aged 9-12 in the three study classrooms.The questionnaire design was completed after careful consideration and collaboration with the three classrooms' teachers.That was easily accepted by pupils, because it was written to take into consideration the pupils' levels of understanding, patience, and ability to comprehend technical terms.Details about the questions posed in the survey are shown in Table 3.Through the questionnaire surveys, we can understand the pupils' perception of IAQ and analyze any correlation between perception and the actual CO 2 level and ventilation rate.The voting results also allowed us to assess any possible psychological impacts of indoor air quality on children.Relevant issues have also been raised, such as the reasons of pupils open windows and the comfort level.The information of the questionnaires is as follows: (1) Background information about the pupils such as age, grade, gender and seat position; (2) Indoor air freshness and air quality assessment (i.e., at the moment they completed the questionnaire); (3) Purpose of, and attitude toward, opening windows.The main contents of the questionnaire are shown in Table 4.To ensure consistency, the questionnaire was sent out to the pupils at 10:50 on Friday morning (March 11, 2016), which was 10 minutes before the third class was over.At that time, the pupils had been in the classroom for about 30 minutes.The questionnaires were explained by the research team members before the pupils answered them.
The indoor physical conditions of the three classrooms during the questionnaire period were as follows: 1802 ppm (CO 2 ), 19.4 • C (T a ) and 27.1% (RH) in S1; 1705 ppm (CO 2 ), 21.9 • C (T a ) and 37.3% (RH) in S2; and 2450 ppm (CO 2 ), 21.6 • C (T a ) and 39.1% (RH) in N1.The outdoor temperature was 0 • C. 106 pupils answered the questionnaire completely, and their responses were recorded by the researchers.

Physical Measurement and Analysis
Figure 5a,b,c show a 5-day plot of the CO 2 level and temperature variation in the three classrooms.The CO 2 concentration changed dramatically during the occupied period (grey bars) with a maximum of 5000 ppm and a minimum of 696 ppm.In a closed classroom, the main source of CO 2 was the students' respiration, and its concentration changed with the number of students [8,21].In this study, pupils start arriving in the classroom after 07:00; the CO 2 levels started to rise when the pupils entered the classroom.The peak period for pupils entering classrooms was between 07:30 and 07:45 every morning.At about 07:45, the CO 2 concentration exceeded 1000 ppm, and at about 08:00, the CO 2 concentration exceeded 1500 ppm.After that, CO 2 concentrations were almost all above 1000 ppm during periods of classroom occupancy.The variation of CO 2 concentrations in the three study classrooms on each day was similar, increasing at times the classroom was occupied and decreasing at times when the pupils were taking a break.The magnitude of the increase and decrease was mainly related to the number of pupils in the classroom and the times when windows were opened.The CO 2 concentration decreased after school was over until the time that the classrooms were closed to the CO 2 concentration outside.The rate of decline depended mostly on whether the windows were opened or not after school was over.During the occupied period, the variation of temperature is also related to the frequency and length of window opening, which is very similar with CO 2 concentration.The range of indoor temperature variations in the three classrooms was 16.50-22.23N1 was significantly lower than that of S1 and S2.The weekly average CO 2 concentration of N1 was significantly higher than that of S2 and S1 during the occupied period, while the frequency of window opening in N1 is lowest among the three classrooms.The larger the maximum value of CO 2 concentration, the smaller the room ventilation rate, and small ventilation rates caused the CO 2 concentration to steadily increase.The daily maximum CO 2 concentration of the three classrooms occurred during the occupied periods.The maximum CO 2 concentration value of N1 occurred on Monday and Tuesday, and it exceeded 5000 ppm (the maximum limit that can be measured by the instrument).The maximum CO 2 concentration of S1 was 4982 ppm, which occurred on Tuesday, and the maximum CO 2 concentration of S2 was 3736 ppm, which occurred on Friday; both of these values are far greater than the reference maximum value of 1,080 ppm.The minimum CO 2 concentration values of the three classrooms ranged from 374 to 549 ppm, which are relatively close to the outdoor concentration, and all of these occurred during unoccupied periods.The average CO 2 concentration from 06:50 to 07:00 was calculated, and its variation range was 376 to 628 ppm.The CO 2 concentration level at this time can be regarded as the ideal initial value of the occupied period.During the periods of C-T, pupils' brains must be engaged, and their behavior is strictly controlled.It is more important to ensure the air quality during periods of C-T than at other times.According to the National Standard [44], the recommended ventilation rate for pupils is not less than 20 m 3 /s per person (circa 5.556 l/s per person).The air exchange rates required in the three study classrooms were 4.532/h in classroom S1, 3.838/h in S2 and 4.430/h in N1.The ventilation rate and air exchange rate in the three classrooms were calculated by measuring the CO 2 concentration at the beginning and end of each C-T period using formula (1) (3).The maximum, minimum and average values of ventilation rates and air exchange rates during C-T time periods are shown in Table 6.The ventilation rates of each C-T period in the three classrooms ranged from 1.291 to 1.925 L/s per person, which is much lower than the recommended value of 5.56 L/s per person.About 99% of ventilation rates of the C-T periods in three classrooms were lower than the recommended value, except that the ventilation rate of the fourth class period in Classroom S2 reached 5.79 L/s per person on Wednesday.84.6% of the ventilation rates were less than half the recommended value, and 34.6% of the ventilation rates were less than 1 L/s per person.
Corresponding to the ventilation rates, 97.4% of the air exchange rates were much lower than required, and 70% of the air exchange rates were less than 1.5/h.The air exchange rates in the N1 classroom were the lowest, with 90% of them being less than 1.5/h.
The serious insufficiency of ventilation caused excessive CO 2 concentrations.Figure 6a shows the level of CO 2 in the three study classrooms during the C-T periods.The CO 2 concentration ranged from 696 to 4982 ppm in S1; from 884 to 3736 ppm in S2; and from 1254 to 5000 ppm in N1.CO 2 concentration in all three classrooms was always at a high level, with median values of 2093 ppm (S1), 1875 ppm (S2) and 2766 ppm (N1).The percentages of measured CO 2 values that exceeded 1080 ppm were 93.94% (S1), 99.05% (S2) and 100% (N1); those exceeding 2500 ppm were 50% (S1), 48% (S2) and 60% (N1).The ventilation rates of each C-T period in the three classrooms ranged from 1.291 to 1.925 L/s per person, which is much lower than the recommended value of 5.56 L/s per person.About 99% of ventilation rates of the C-T periods in three classrooms were lower than the recommended value, except that the ventilation rate of the fourth class period in Classroom S2 reached 5.79 L/s per person on Wednesday.84.6% of the ventilation rates were less than half the recommended value, and 34.6% of the ventilation rates were less than 1 L/s per person.
Corresponding to the ventilation rates, 97.4% of the air exchange rates were much lower than required, and 70% of the air exchange rates were less than 1.5/h.The air exchange rates in the N1 classroom were the lowest, with 90% of them being less than 1.5/h.
The serious insufficiency of ventilation caused excessive CO2 concentrations.During the C-T periods, the CO 2 concentration in the three study classrooms generally showed an upward trend.Statistical results indicate that the higher the initial value of CO 2 concentration in the classrooms, the higher the average value (see Figure 6b).The initial CO 2 concentration of 94.3% in the three classrooms was higher than 1000 ppm; these values were 696-3420 ppm (S1); 957-3514 ppm (S2); and 1452-5000 ppm (N1).The maximum values of the average CO 2 concentration in the three study classrooms were 3384 ppm (S1), 4229 ppm (S2) and 5000 ppm (N1); these values are 3-5 times the reference value.The mean median CO 2 concentration values were 1843 ppm (S1), 1955 ppm (S2) and 2808 ppm (N1).
We sought to further analyze the speed by which classroom CO 2 concentration exceeded the standard during the C-T periods, and to determine the effect of window opening on reducing CO 2 concentrations during the I-T periods.According to Equation ( 4), the variation rates of CO 2 concentration in the three classrooms were calculated.The fastest variation rate and corresponding CO 2 values, ventilation rates and air exchange rates are shown in Table 7.During the C-T periods, the fastest rate of increase in the CO 2 concentration was 46.3 ppm/min in S1, meaning that CO 2 concentration increased by 1621 ppm in 35 minutes.The corresponding ventilation rate and air exchange rate were only 0.439 L/s per person and 0.358/h, respectively.The corresponding values for classroom S2 were 44.5 ppm/min, 0.556 L/s per person and 0.431/h; for classroom N1, they were 35.8 ppm/min, 0.556 L/s per person and 0.431/h.These results are also related to the initial CO 2 concentration.Given the same difference, the lower the initial concentration, the larger the ventilation rate.The rising rate of CO 2 indicates that ventilation was necessary during the C-T periods, because CO 2 increased by 700 ppm in the three classrooms within 20 minutes under closed conditions.Class breaks proved to be good times for ventilation.In this study, most of the I-T period CO 2 concentrations showed decreases.The decline rates of CO 2 concentration affected the initial concentration of the next C-T period.The faster the CO 2 decline rate, the larger the decline range in a limited time, and the closer the CO 2 concentration came to the ideal initial value.Table 7 shows the maximum decline rate of CO 2 concentration in short and long periods of I-T.Assuming that the classroom was empty during the I-T period, the ventilation and air exchange rates were calculated according to equations (2) and (3).
The maximum decline rate of CO 2 concentration in S1 and S2 during the I-T periods was 2-3 times that of the increase rate during the C-T periods.The corresponding ventilation rates in S1 and S2 were 2.307 L/s per person and 2.799 L/s per person, respectively.The fastest decline rate of CO 2 concentration in the N1 classroom in a short I-T period was only 29.7 ppm/min, and the corresponding ventilation rate was only 0.890 L/s per person.The fastest CO 2 concentration declining rates of the three classrooms during the long I-T period were similar, ranging from 52.6 ppm/min to 53.9 ppm/min; this was slightly higher than the fastest increase rate during the C-T periods.However, the corresponding ventilation rates were 2.004 L/s per person in S1, 1.814 L/s per person in S2 and 1.076 L/s per person in N1.The ventilation volume for all intervals was less than the standard value.The same as [27], although CO 2 concentration of S1 was reduced by as much as 1239 ppm in 10 minutes, it was still insufficient to maintain acceptable level.According to previous statistics, 97.3% of the initial concentration during the C-T periods exceeded 1000 ppm, indicating that there was not enough ventilation during the I-T periods to reduce the CO 2 concentration to a level that was close to that of the outdoor value.
It can be seen from the above that the IAQ of the three classrooms is very poor due to the serious insufficiency of ventilation.Opening windows during the I-T period can effectively reduce the initial value of CO 2 in the next C-T.However, the air quality in the C-T period cannot be guaranteed only by after-class ventilation.

Satisfaction Evaluation and Air Freshness Perception of IAQ
The voting results of the perception of air freshness are shown in Figure 7.The voting results of pupils in the three classrooms were very similar.Most of the pupils felt that the air was fresh.The voting rates for feeling fresh air (A) were 68.4% in S1, 71.9% in S2 and 69.4% in N1.The votes for sensing non-fresh air (B) were 10.5% in S1, 12.5% in S2 and 8.3% in N1.The number of responders who chose option C (non-fresh or smelly air) in S1 and N1 were twice that of those who chose option B. Pupils in classroom S2 chose option B as well as option C. No one thought that the classroom smelled bad for a long period of time, except for one boy in S2.
was no significant difference between genders by comparing the votes of boys and girls in three classrooms.
Usually, classroom odors come from children's metabolism and breathing, which can accumulate during prolonged presence in an enclosed space.Although most pupils have a sensitive sense of smell, they remain in classrooms for a long period of time and have a certain degree of adaptability to and tolerance of their environment [46].During the survey period, the average CO2 concentration values during the third class period in the three study classrooms were about twice the standard value: 1802 ppm in S1, 1705 ppm in S2 and 2450 ppm in N1.Nearly 70% of the pupils still thought the air was fresh, indicating their adaptability to the air environment.Pupils' votes that the air was fresh were not affected by different levels of CO2 concentration.The voting results of satisfaction with the indoor air quality of the three classrooms are shown in Table 8 and Figure 8. 80.2% of pupils expressed their satisfaction with IAQ (A+B).84.2% of S1 pupils, 77.5% of S2 pupils and 69.5% of N1 pupils cast satisfied votes.At the same time, only 2.6% to 6.2% of all pupils in the three classrooms voted that they found the IAQ unsatisfactory (D+E).The average level of unsatisfactory votes was only 3.7%, indicating that most pupils were satisfied with the indoor air quality.Although the satisfaction rate of the N1 classroom was lower than that of the S1 and S2 classrooms, the unsatisfactory votes of N1 was only 2.1%, slightly less than that of the other two classrooms.Fewer boys in the three classrooms voted that the air quality was good (A+B) than girls did, but the difference was not significant between genders.The votes of boys and girls in each class were compared.There were few differences in voting between boys and girls in the same classroom.87.5% of girls and 56.3% of boys in classroom S2 voted that they felt the air was fresh.However, it can be seen that overall, 62.5% of girls and 83.3% of boys in classroom N1 thought that the indoor air was fresh in their classrooms.The difference in voting between boys and girls in S1 was significantly smaller than that in S2 and N1.Overall, there was no significant difference between genders by comparing the votes of boys and girls in three classrooms.
Usually, classroom odors come from children's metabolism and breathing, which can accumulate during prolonged presence in an enclosed space.Although most pupils have a sensitive sense of smell, they remain in classrooms for a long period of time and have a certain degree of adaptability to and tolerance of their environment [46].During the survey period, the average CO 2 concentration values during the third class period in the three study classrooms were about twice the standard value: 1802 ppm in S1, 1705 ppm in S2 and 2450 ppm in N1.Nearly 70% of the pupils still thought the air was fresh, indicating their adaptability to the air environment.Pupils' votes that the air was fresh were not affected by different levels of CO 2 concentration.
The voting results of satisfaction with the indoor air quality of the three classrooms are shown in Table 8 and Figure 8. 80.2% of pupils expressed their satisfaction with IAQ (A+B).84.2% of S1 pupils, 77.5% of S2 pupils and 69.5% of N1 pupils cast satisfied votes.At the same time, only 2.6% to 6.2% of all pupils in the three classrooms voted that they found the IAQ unsatisfactory (D+E).The average level of unsatisfactory votes was only 3.7%, indicating that most pupils were satisfied with the indoor air quality.Although the satisfaction rate of the N1 classroom was lower than that of the S1 and S2 classrooms, the unsatisfactory votes of N1 was only 2.1%, slightly less than that of the other two classrooms.Fewer boys in the three classrooms voted that the air quality was good (A+B) than girls did, but the difference was not significant between genders.The overall comparison of the votes of boys and girls showed no significant difference, which indicats that gender did not affect the evaluation results.Compared with the votes regarding air freshness, the voting trend of satisfaction with air quality in the three classrooms was consistent with that of perception of air freshness.30.2% of the pupils (B+C+D) voted that the air was not fresh, and 19.8% of the pupils (C+D) voted that it was smelly.These feelings did not reflect the votes on IAQ satisfaction.This indicates that pupils considered the air quality to be acceptable or satisfactory, even though they felt that the air was not fresh.The average CO2 concentration was 100% over the standard value, but the unsatisfactory vote percentage (3.7%)was significantly different from the actual situation, indicating that the actual CO2 concentrations did not affect pupils' satisfaction with air quality.This also shows that most pupils did not notice or mind air quality problems caused by excessive CO2.High CO2 concentration did not bring obvious bad mood to most children.Pupils' positive evaluation of the IAQ may affect their willingness to open windows (i.e., they most often would not open them), resulting in a low frequency of window openings.

The purpose and willingness to open windows
Generally, the purpose of window ventilation is to reduce indoor temperatures [27,47] or improve indoor air quality [36].The results of the present study are that most pupils opened The overall comparison of the votes of boys and girls showed no significant difference, which indicats that gender did not affect the evaluation results.Compared with the votes regarding air freshness, the voting trend of satisfaction with air quality in the three classrooms was consistent with that of perception of air freshness.30.2% of the pupils (B+C+D) voted that the air was not fresh, and 19.8% of the pupils (C+D) voted that it was smelly.These feelings did not reflect the votes on IAQ satisfaction.This indicates that pupils considered the air quality to be acceptable or satisfactory, even though they felt that the air was not fresh.The average CO 2 concentration was 100% over the standard value, but the unsatisfactory vote percentage (3.7%)was significantly different from the actual situation, indicating that the actual CO 2 concentrations did not affect pupils' satisfaction with air quality.This also shows that most pupils did not notice or mind air quality problems caused by excessive CO 2 .High CO 2 concentration did not bring obvious bad mood to most children.Pupils' positive evaluation of the IAQ may affect their willingness to open windows (i.e., they most often would not open them), resulting in a low frequency of window openings.

The Purpose and Willingness to Open Windows
Generally, the purpose of window ventilation is to reduce indoor temperatures [27,47] or improve indoor air quality [36].The results of the present study are that most pupils opened windows to improve indoor air quality rather than to reduce the classroom temperature.Statistical results show that 82.1% of pupils in the three classrooms opened windows for ventilation to improve air quality, 12.2% to cool the classroom temperature, and 5.7% for other reasons.
Figure 9 shows pupils' votes on their perception and satisfaction with indoor temperatures.75.5% of the pupils thought the temperature was suitable, 12.3% thought the temperature was low, 11.3% thought it was warm, and 0.9% thought it was hot.86.8% of the pupils (A+B) were satisfied with the indoor temperature, while only 5.6% of the pupils (D+E) were not satisfied.The results of pupils' voting on temperature perception and satisfaction re-validate the purpose of pupils' window opening; i.e., most pupils open windows to improve air quality even at comfortable temperatures.temperature was high, and they lacked the subjective will to open windows to cool down.Second, most of the pupils in this study thought that the indoor air quality was good and very satisfactory.This may be the reason for not opening windows, because pupils did not think it necessary to open windows to improve IAQ.Consistent with the frequency of window openings, the votes for comfort with window opening in S1 were significantly higher than those in classroom N1.However, the statistical results did not show a difference between S2 and N1 in terms of votes for comfort with window opening.
Research shows that the two main purposes of window opening are to reduce temperature or to improve indoor air quality; these purposes may be affected by subjective feelings.First, most pupils in this study thought that the indoor temperature was appropriate.Few pupils thought that the indoor temperature was high, and they lacked the subjective will to open windows to cool down.
Second, most of the pupils in this study thought that the indoor air quality was good and very satisfactory.This may be the reason for not opening windows, because pupils did not think it necessary to open windows to improve IAQ.

Conclusions
During the heating season in severe cold weather areas of China, CO2 concentrations in three classrooms of a naturally ventilated primary school in Shenyang, located in northeast China, were continuously measured for one week.Meanwhile, 106 pupils were asked to complete subjective questionnaires on air quality.
The results show that the daily average concentration of CO2 in the three classrooms in the occupied periods varied from 1500 to 3863 ppm, which is far beyond the standard value of 1000 ppm, and the CO2 concentrations during the Class-Time periods were higher than during other periods.
The main reason for the excessive CO2 concentrations in the classrooms was that the ventilation rate was insufficient due to the low frequency of window opening.About 99% of the ventilation rates during the Class-Time periods were lower than the recommended value.According to the calculated increase rate of carbon dioxide in the three study classrooms, the carbon dioxide level in closed classrooms could exceed 700 ppm in 20 minutes, which indicates that it is necessary to increase the ventilation rates during the Class-Time periods.In parallel, although the indoor carbon dioxide concentration can be reduced by 1230 ppm in 10 minutes by opening windows during the Intermittent-Time periods, 97% of the initial CO2 concentrations in the Class-Time periods were more than 1000 ppm, indicating that the ventilation rate was also insufficient in the interval times.

Conclusions
During the heating season in severe cold weather areas of China, CO 2 concentrations in three classrooms of a naturally ventilated primary school in Shenyang, located in northeast China, were continuously measured for one week.Meanwhile, 106 pupils were asked to complete subjective questionnaires on air quality.
The results show that the daily average concentration of CO 2 in the three classrooms in the occupied periods varied from 1500 to 3863 ppm, which is far beyond the standard value of 1000 ppm, and the CO 2 concentrations during the Class-Time periods were higher than during other periods.The main reason for the excessive CO 2 concentrations in the classrooms was that the ventilation rate was insufficient due to the low frequency of window opening.About 99% of the ventilation rates during the Class-Time periods were lower than the recommended value.According to the calculated increase rate of carbon dioxide in the three study classrooms, the carbon dioxide level in closed classrooms could exceed 700 ppm in 20 minutes, which indicates that it is necessary to increase the ventilation rates during the Class-Time periods.In parallel, although the indoor carbon dioxide concentration can be reduced by 1230 ppm in 10 minutes by opening windows during the Intermittent-Time periods, 97% of the initial CO 2 concentrations in the Class-Time periods were more than 1000 ppm, indicating that the ventilation rate was also insufficient in the interval times.
Through the statistical voting results of the pupils' subjective evaluations (questionnaires), it was found that there was no significant difference in the voting proportion among classrooms or between genders.Although the levels of CO 2 in all three classrooms exceeded the recommended values during the questionnaire period, 69.8% of the pupils voted that they found the air in their classrooms to be fresh, which is inconsistent with the actual CO 2 measurement results.Most pupils opened windows to improve indoor air quality.However, only 3.7% of the pupils were not satisfied with the air quality, which greatly reduced their willingness to open windows and hindered the behavior of opening windows.Increasing the frequency of window opening would increase the ventilation rate, if the pupils had higher acceptance of window opening.
In summary, no relationship was found between the measured CO 2 concentrations and the pupils' perception of the freshness of the air.There was a significant difference between high indoor

Figure 1 .
Figure 1.Location of Shenyang and the severe cold area of China.

104
concrete frame construction and brick walls without any thermal insulation materials.The layout, 105 function and space design features of teaching building A are representative of the typical primary 106 and secondary school buildings that have been built in the severe cold regions of China in recent 107 decades.Building A is heated by hot water radiators during the heating season.All classrooms are 108 ventilated by opening windows manually, and there are no mechanical ventilation facilities for use 109 during any part of the year.There are 44 classrooms on the second to fourth floors of the building for 110 grade 4 to grade 6 (pupils aged 8 to 13), of which 28 classrooms face south and 16 classrooms face 111 north.The Building B is also a teaching building for grade 1 to grade 3 (pupils age 6-8).

Figure 2 .
Figure 2. Location of selected primary school in Shenyang City (a), Site map (b). 114

Figure 1 .
Figure 1.Location of Shenyang and the severe cold area of China.

Figure 1 .
Figure 1.Location of Shenyang and the severe cold area of China.
concrete frame construction and brick walls without any thermal insulation materials.The layout, function and space design features of teaching building A are representative of the typical primary and secondary school buildings that have been built in the severe cold regions of China in recent decades.Building A is heated by hot water radiators during the heating season.All classrooms are ventilated by opening windows manually, and there are no mechanical ventilation facilities for use during any part of the year.There are 44 classrooms on the second to fourth floors of the building for grade 4 to grade 6 (pupils aged 8 to 13), of which 28 classrooms face south and 16 classrooms face north.The Building B is also a teaching building for grade 1 to grade 3 (pupils age 6-8).

Figure 2 .
Figure 2. Location of selected primary school in Shenyang City (a), Site map (b).

Figure 2 .
Figure 2. Location of selected primary school in Shenyang City (a), Site map (b).

Figure 3 .
Figure 3. S1, S2 and N1 location on the 4th floor and the measurement points in the classrooms.

Figure 3 .
Figure 3. S1, S2 and N1 location on the 4th floor and the measurement points in the classrooms.

Figure 3 .
Figure 3. S1, S2 and N1 location on the 4th floor and the measurement points in the classrooms.

Figure 5 .
Figure 5. CO2 concentration and temperature values in S1(a), S2(b) and N1(c) between the 7th and the 11th of March 2016 (the shadowed areas correspond to the occupied periods).4.1.1.IAQ evaluation by CO2 concentrationThe maximum, minimum and average daily CO2 concentrations in the three study classrooms are shown in Table5.The variation range of the daily average CO2 concentration of the three classrooms during the occupied period was between 1510 ppm and 3863 ppm.The highest daily average was about 4 times the recommended value.In the periods of I-T, there was no significant

Figure 5 .
Figure 5. CO 2 concentration and temperature values in S1(a), S2(b) and N1(c) between the 7th and the 11th of March 2016 (the shadowed areas correspond to the occupied periods).4.1.1.IAQ Evaluation by CO 2 ConcentrationThe maximum, minimum and average daily CO 2 concentrations in the three study classrooms are shown in Table5.The variation range of the daily average CO 2 concentration of the three classrooms during the occupied period was between 1510 ppm and 3863 ppm.The highest daily average was about 4 times the recommended value.In the periods of I-T, there was no significant difference in the frequency of window openings between S1 and S2, while the window opening frequency in

Figure 6 (Figure 6 .
Figure 6.CO2 concentration distributions: (a) initial CO2 concentration and (b) average CO2 concentration.During the C-T periods, the CO2 concentration in the three study classrooms generally showed an upward trend.Statistical results indicate that the higher the initial value of CO2 concentration in the classrooms, the higher the average value (see Figure6b).The initial CO2 concentration of 94.3% in the three classrooms was higher than 1000 ppm; these values were 696-3420 ppm (S1); 957-3514 ppm (S2); and 1452-5000 ppm (N1).The maximum values of the average CO2 concentration in the three study classrooms were 3384 ppm (S1), 4229 ppm (S2) and 5000 ppm (N1); these values are 3-5 times the reference value.The mean median CO2 concentration values were 1843 ppm (S1), 1955 ppm (S2) and 2808 ppm (N1).We sought to further analyze the speed by which classroom CO2 concentration exceeded the standard during the C-T periods, and to determine the effect of window opening on reducing CO2 concentrations during the I-T periods.According to Equation (4), the variation rates of CO2 concentration in the three classrooms were calculated.The fastest variation rate and corresponding CO2 values, ventilation rates and air exchange rates are shown in Table7.During the C-T periods,

Figure 7 .
Figure 7.The air freshness votes in the three study classrooms.

Figure 7 .
Figure 7.The air freshness votes in the three study classrooms.

Figure 8 .
Figure 8.The satisfied votes of boys and girls in the three classrooms.

Figure 8 .
Figure 8.The satisfied votes of boys and girls in the three classrooms.

19 Figure 10 .
Figure 10.Votes regarding pupils' comfort level with opening windows.Most pupils accepted the window opening behavior, which should be helpful for improving indoor air quality.However, the actual ventilation rate was very low (calculated in Section 4.1.2).No relationship was found between the observed frequency of, and comfort with, opening windows.Pupils in the three classrooms almost never opened windows in the C-T periods, and the frequency of window openings in S1 and S2 was significantly higher than in N1 during the I-T periods.

Figure 10 .
Figure 10.Votes regarding pupils' comfort level with opening windows.
* Time for lunch and outdoor activities.
Formulas for Calculating Ventilation Rate and Variation Rate of CO 2 Concentration

Table 4 .
The survey questionnaire.

Table 5 .
Indoor CO 2 concentration (ppm) in the three study classrooms.
Note: * For both occupation and unoccupied periods.** Value recorded at 07:00 on Monday.

Table 6 .
Ventilation rates and air exchange rate (AER) in C-T.

Table 7 .
During the C-T periods, the fastest rate of increase in the CO2 concentration was 46.3 ppm/min in S1, meaning that CO2 concentration increased by 1621 ppm in 35 minutes.The corresponding ventilation rate and air exchange rate were only 0.439 L/s per person and 0.358/h, respectively.The corresponding values for

Table 7 .
The fastest rate of variation in CO 2 concentrations and corresponding parameters.
* Dring the I-T periods: S = short intervals, L = short intervals; ** The fastest rate of increase during C-T periods and the fastest rate of decline during I-T periods; *** Diff.= difference value between the initial value and ending value of each C-T and I-T period.