The study was carried out in three educational centers located in Madrid whose characteristics are shown in table 1. The intention is to analyze buildings prior to 1979, as they are a cluster of inefficiency in Spain, since they lack thermal insulation (25). However, a recently constructed building was included in the sample, to analyze the possible influence of the newest regulations. None of the educational centers have mechanical ventilation. In all cases ventilation is done manually by opening windows.

Three spaces were selected in each educational center. Different orientations were taken in order to include possible variations in wind and solar incidence.

Data of CO₂ concentration and temperature are registered in the selected buildings as case studies. Data is obtained with Wöhler CDL 210 data recorders (27). They offer a CO₂ measurement range between 0 and 2,000 parts per million (ppm), with an accuracy of 50 ppm ± 5% and a resolution of 1 ppm. They use the NDIR procedure (non-dispersive infrared absorption analyzer) as a measurement principle. They offer a range of temperature measurement between -10ºC and + 60ºC with an accuracy of ± 0.6% and a resolution of 0.1ºC.

Data has been taken every 15 minutes during one year, between the months of October 2017 and September 2018.

The recorders are installed in the classrooms on the outside wall, one meter from the ground, with the characteristics shown in table 2.

The aim is to evaluate thermal comfort and air quality, comparing the data obtained with the recommended values. To study the indoor air quality, the CO₂ concentration data is considered. This is one parameter used by the Spanish Building 2017 (Code Document Base HS-3) (28), hereinafter DB HS-3 and by the Regulation of Thermal Installations in Buildings, (29) from now on RITE. This parameter is chosen because the adequate instrumentation to obtain its value is available and because it is a good indicator to observe the amount of air renovation in spaces with human presence. Indirectly, the higher or lower rate of ventilation influences to maintain better or worse indoor air quality.

RITE sets a CO₂ concentration for classrooms of 500 ppm above the concentration in outside air. Regarding the concentration of CO₂ in outdoor air, the value of 400 ppm is adopted (28, 30), so the value measured inside the classroom should not exceed 900 ppm. This value is similar to the expressed in DB HS-3 “Air Quality”, where it is stated that: “In habitable rooms of the dwellings, a sufficient external air flow must be provided to ensure that in each location the average annual concentration of CO₂ is less than 900 ppm and that the accumulated annual CO₂ that exceeds 1,600 ppm is less than 500,000 ppm · h” (30). Although this limitation only applies to housing, it offers a second reference value to guarantee the health of people. The standard UNE-EN 013779: 2008 “Ventilation for non-residential buildings - Performance requirements for ventilation and room-conditioning systems” (31) also matches this value by marking in its table A.10 the value of 900 ppm to activate ventilation systems that are responsible for ensuring air quality.

To study thermal comfort, temperature data is taken, as contemplated in several studies (32, 33, 34, 35). Thermal comfort depends on various factors such as the characteristics of the users (age, sex, level of clothing...), the shape of the building, the thermal facilities, as well as the ability to be controlled by the users and the external climate, among others. So humans have an adaptive capacity to feel comfort within varied thermal conditions (36).

RITE establishes, as a requirement for thermal quality of the environment, a temperature range of 21ºC to 23ºC in winter and 23ºC to 25ºC in summer, when metabolic activity is maintained in a classroom (1.2 met) and clothing is usual for the season (0.5 clo in summer and 1.0 clo in winter).

For a gym, the comfort temperature drops to 17-19ºC in winter and 19-21ºC in summer, given that the metabolic activity is greater even though the clothing is lighter (37). Nevertheless, Givoni (38) establishes as a comfort zone, the one comprised between values of 21ºC and 26ºC.

To study the quality of the interior environment and its effect on the wellbeing and comfort of the occupants, we proceed to compare the data obtained with the aforementioned reference values.

Mean monthly data is represented in figure 1.

The highest monthly average values are obtained in the month of January, while the lowest values are obtained in May. These values do not take into account the months of July and August that are not academic and the months of June and September in which the academic activity starts and ends.

In Figure 2, the evolution of CO₂ concentrations during the worst and best week in each building is shown. The week from 22nd Jan. to 28th Jan. stands for the worst performance, while the week of May 20th to May 26th stands for the best performance.

Table 3 shows the annual indicators for CO₂ concentration.

The classrooms are occupied 6 hours a day, and the annual occupation period is 175 days according to the school calendar. This implies that the classrooms remain occupied 12% of the hours throughout the year. Therefore, Table 4 shows the CO₂ concentration data obtained only during school hours.

The obtained average monthly values are shown in Figure 3. These values are compared to the average monthly outside temperature (39).

The highest average monthly values are obtained in the month of October, while the lowest values are obtained in January, if we do not take into account the months of June, July, August and September. Figure 4 shows the temperature evolution in each teaching center during the week of January 22 to January 28, the coldest month.

Figure 5 shows the temperature evolution in each teaching center during the week October 23 to 29 are represented, the hottest month.

Figure 6 shows the percentage of hours in which each classroom is below/above or within the comfort temperature marked by RITE. Winter season and summer season are differentiated since the RITE identifies different comfort temperature values for each season. It is considered the winter season when the heating system is operating from November to April. The one in which the heating system is not working, from May to October, is considered the summer season. The comfort temperature in winter is 21-23ºC for the classrooms and 17-19ºC for the gym, and in summer is 23-25ºC for the classrooms and 19-21ºC for the gym.

As previously mentioned, the classrooms remain occupied for 12% of the hours throughout the year. Therefore, the temperature data is obtained only during the moments in which the classroom remained occupied and shown in figure 7.

The limit set by the current housing regulations in Spain for the annual cumulative CO₂ that exceeds 1,600 ppm is 500,000 ppm h. To set this limit, a scenario of occupation of the dwelling in which users remain outside for 8 hours on weekdays and 2 hours on weekends is established. This scenario implies that users stay in the house 6,464 annual hours.

In the case of schools, the time spent in a classroom is 6 hours for 175 school days, which is 1,050 hours per year. The equivalent limit for the annual accumulated CO₂ that exceeds 1,600 ppm in classrooms, to maintain the same level of health, would be set at 81,219 ppm · h. This value represents 16.2% of the limit marked on housing, since the annual time spent in the classroom is 16.2% of the time spent in the home.

Table 5 shows that classrooms A1 and C2 exceed the equivalent limit value for the annual cumulative CO₂ that exceeds 1,600 ppm between 3.5 and 4 times, which means that students are exposed to high concentrations of CO₂ accumulated during the period of occupation of classrooms. While the next two classrooms, C3 and C1 have values 0.7 times below the limit value, class B3 has a value 0.35 times below the limit value and the rest of the classrooms, values below 0.05.

Regarding the percentage of hours in which the classrooms have a CO₂ concentration above 900 ppm, Table 5 shows that classrooms A1, C2 and B3 exceed this concentration between 46% and 64% of the time in which they remain busy. The same classrooms have an average CO₂ concentration value above 900 ppm, exceeding this value 1.6 times in the case of classroom A1. Table 5 shows the different classrooms ordered according to how they exceed the limit values of the different CO₂ concentration indicators.

First of all, the good behavior that the gym offers in comparison of the classrooms should be highlighted. The value in which it shows the best performance is in Percentage of hours with CO₂> 900 ppm (%), whose results are between 5 and 11 times lower than the classrooms. Despite the fact that in the gym, due to more intense metabolic activity than in the classroom, the emission of CO₂ by the students increases, its greater volume makes the CO₂ concentration lower than that achieved in the classrooms.

It can be observed that the same educational center has some classrooms with very unfavourable air quality conditions, and other classrooms with very favourable conditions regarding the concentration of CO₂.

However, school C obtains the worst results, while the other two schools obtain slightly better results. This could be due to two reasons. School C was built in 2010 and has window framing which present more tightness than other schools, this circumstance limits air infiltration. On the other hand, the classrooms studied in case study C have an average occupation of 26.3 students compared to 22 and 14.3 students in cases B and A respectively. Taking into account that the generation of CO₂ is 19 l / h · occupant, the generation of CO₂ in school C is 20% higher than in school B and 80% higher than in school A. So, since more CO₂ is generated and less dissipated, school C presents the worst results as a whole. The SINPHONIE report: Schools Indoor Pollution & Health Observatory Network in Europe (40) studies air quality in 23 European countries among which Spain is not included. The average value of CO₂ concentration in the 114 schools studied in this report is 1,433 ppm, while the average value obtained in the 3 Spanish schools studied in this article is 916 ppm. In spite of being a value higher than 900 ppm, only Norway with 686 ppm and Sweden with 657 ppm have lower average concentrations.

Finally, it should be noted how the studied gym achieves very low results. This is due to the fact that, despite that the metabolic activity that takes place in it implies a greater emission of CO₂ by the students, as it has a larger volume than a classroom, the CO₂ concentration does not reach the values obtained in smaller spaces.

Table 6 shows different classroom analysis according to the comfort temperature for the winter season with space heating, and summer season, during the annual occupation period of the classrooms.

It can be observed that the highest temperature during the heating period is school B (table 6), while school C has the highest percentage of hours inside the comfort temperature, and school A is the one that presents cooler temperatures.

However, in the periods when heating installations are not working, the warmest case is School C, which is the school with the longest period of comfort, and the coldest is case A. In this last period, thermal behaviour of the building is analysed in free oscillation (without an external incorporation of energy), since it is not affected by the operation of heating appliances.

None of the centers have shading devices such as blinds or horizontal overhangs that allow the entry of natural light while protecting windows from direct solar radiation. In all cases, the entry of direct solar radiation can be blocked deploying the blinds, but this forces the use of artificial lighting when protection from direct solar radiation is needed.

Table 11 shows schools ordered from best to worst behaviour with respect to the different parameters studied.

It can be observed that case C, despite being in the neighbourhood with the highest average income and being of more recent construction, shows the worst behaviour.

As previously indicated, greater air tightness of windows in case C and the high occupancy of its classrooms are causing high concentrations of CO₂. Also the fact of having opaque enclosures and glazing, both with lower thermal transmittances, together with their higher occupancy, are causing overheating in summer. Although, due to these circumstances, it would be foreseeable that school C presented the best temperature comfort in winter, this is not the case. This may be due to the fact that the heating installation in School B, which presents better comfort in winter, contributes with much more energy consumption, despite the fact that it has greater losses.

The results of concentration of CO₂ in the different classrooms lead us to the conclusion that the determining factor is not so much the constructive characteristics of the educational center or the number of students present in each classroom, which undoubtedly have an influence, but the use of natural ventilation by opening the windows which is done in each classroom. Since the CO₂ concentration of the outdoor air is lower than the CO₂ concentration of the indoor air, when the windows are opened, the CO₂ concentration is reduced.

The worst values are obtained in a classroom from each school, while other classrooms in the same center obtain better results.However, school C built in 2010 whose window framing has greater tightness and whose occupancy by classroom is higher, gets worse results against CO₂ concentrations, while the other two schools whose window framing is less tight and their classrooms are less busy, get better results.

It is concluded, therefore, that natural ventilation by opening the windows in the classrooms in the breaks between classes can be a sufficient measure to guarantee air quality in classrooms, but this is not carried out adequately in most cases. To avoid this situation and ensure adequate air quality, it is advisable to carry out a protocol in which someone is responsible for opening and closing the windows during the changing of class. The air quality delegate could be entrusted to supervise, for the good of all, that the classroom is adequately ventilated. Since opening windows in winter improves indoor air quality, but reduces thermal confort, it would be desirable for classrooms to have CO₂ and temperature meters to manage window opening times so that neither of the two parameters were not compromised.

However, this form of ventilation is less energy efficient than double flow mechanical ventilation with heat recovery. Furthermore, with mechanical ventilation, adequate air renewal flow rates are guaranteed regardless of whether correct manual ventilation management is carried out.

Regarding the seasons, the highest concentrations of CO₂ are reached in winter, which is when the windows are least opened due to the low outside temperatures.

The temperature results obtained show us that the classrooms meet comfort temperature only for 30% of teaching hours.

The bad situation of air quality and thermal comfort in the public schools analysed in Madrid is noted. Adequate air quality is achieved for 64% of teaching time. Appropriate temperature conditions are achieved during 30% of teaching time. It is necessary to carry out an investigation of thermal comfort and air quality on a larger sample of schools in Madrid and other regions in Spain, in order to obtain a more accurate diagnosis, depending on the year of construction, the climatic zone in which it is located and other variables that can be identified, as building materials, location of the most polluted school zones, treatment of CO₂ as a tracer gas, etc.