*4.2. Relationship between Physical Parameters and Symptoms Described*

Some symptoms were more predominant when outdoor temperature was lower, although no clear linear relationship could be established. These symptoms were more frequent in winter when the thermal differential is at its highest, usually linked to a lack of ventilation at the time, as supported by the high CO2 indices as a general air quality indicator.

Although the symptoms often appeared to be more evident when the windows were open, this should be seen as a consequence, not a cause, as user perception of the symptoms was generally clearer when opening the windows. This is interesting to note, as it could be due to a situation which exceeded the perception threshold. In the winter, it is more common to observe symptoms such as difficulty concentrating, dry throat, and tiredness. These are very closely linked to poor hygrothermal control, even with windows open, where temperature and relative humidity are far more important, especially with open windows, as well as increased indoor CO2 linked to poor ventilation. In contrast, itchiness and chest tightness were barely noticeable.

The situation changed in midseason and symptoms, such as difficulty concentrating, tiredness, and nasal congestion, were less widely reported. However, symptoms less connected with the absence of hygrothermal regulation increased, while there was a greater presence of symptoms that may be linked to outdoor exposure.

When both lighting parameters, illuminance (E) and illuminance uniformity (Uo), were analyzed and referred to symptomatology, no clear correlation was obtained, as other previous studies showed for educational buildings [80]. This may be because illuminance values in the classrooms under study were generally over 350–400 lx with a uniformity of 0.40–0.50, so they were values good enough to not influence students at a symptomatic level.

The infiltration rate (n50) and the symptoms related by occupants showed a very tenuous connection, with some weak trends in the case of tiredness, as well as difficulty on concentrating, dry throat, and headache. Given that the airtightness of the classrooms was, in general, adequate or even good, with an average value of 6.97 h−<sup>1</sup> with a maximum value of 10 h−1, its influence can be moderate due to its low impact on air renewal. It also indicates that other variables, like time spent inside the classroom or windows' and doors' operation, can have more importance than the airtightness of the room.

There was no clear linear correlation between the students' clothing insulation and the symptoms described during measurements. The possibility of freely varying the level of clothing insulation by the students, according to their individual thermal needs, may be a factor that influenced this lack of relationship between clothing and symptomatology, besides psychological factors linked to clothing.

When symptomatology was assessed as global, there were some trends that could be identified. If CO2 was assumed as an overall indicator of indoor air renovation (not as a contaminant itself), the worsening of indoor environment linked with the increase of symptoms related. It can be approximated to a logarithmic regression relation (Figure 14), although a wide spread of values must be assumed. Different patterns for winter and midseason were described due to adaptation of users and the influence of outdoor species. Although this model presents some uncertainty for its use as a prediction tool, it does have the capacity to act as a qualitative indicator.

**Figure 14.** General average related symptoms' scores relation with indoor CO2 grouped by measured classrooms (green for middle season and blue for winter).

A linear trend model was calculated for the average symptom score and given a record of the average indoor CO2 (logarithmic fit). The model was statistically significant at *p* < 0.05, having a high correlation coefficient (R<sup>2</sup> = 0.8833) and a mean square error (MSE) of 0.6160.

A somewhat weaker linear relationship (logarithmic fit also) was seen (R<sup>2</sup> = 0.509 for midseason and R<sup>2</sup> = 0.143 for winter) but with statistical signification (*p*-value < 0.05 in both cases) and an error of MSE 0.425. Although dispersion was high, it was also a useful qualitative trend indicator, and was found between the overall perception of symptoms and the indoor operative temperature (Figure 15). In this case, it can be established that the symptoms tended to be more frequent when indoor temperatures increased, also with specific patterns for winter and middle season.

**Figure 15.** General average related symptoms' scores relation with indoor operative temperature grouped by measured classrooms (green for middle season and blue for winter).

### **5. Conclusions**

A wide study sample of 47 naturally ventilated multipurpose classrooms of the most representative climate zones of southern Spain was characterized and analyzed through field measurements and surveys distributions, in order to contrast environmental sensation votes, perception, and indoor-related symptoms described by 977 students during lessons with physical and environmental parameters, as well as operational scenarios.

The main operational case to be analyzed, according to votes and symptoms, was the windows' operation. In this sense, the 61% of the case studies during winter season had the windows open, which can be related both to a bad regulation of the heating system (the slight heat excess had to be dissipated) as well as to a poor indoor environment perception. In this way, the case studies with open windows in winter had a higher mean indoor air temperature value (21.5 ◦C versus 21.0 ◦C) and higher standard deviation of the mean radiant temperature (2.6 ◦C versus 1.6 ◦C). The mean thermal perception of students in winter season with open windows reinforced this slight heat excess, given that it was in a comfort range but 0.15 points warmer than in the case of closed windows, also expressing a thermal preference of thermal neutrality-mild cold (−0.06 on the ASHRAE scale) with open windows in contrast to the preference for a warmer environment when the windows were closed (+0.13). The thermal assessment of the environment through the thermal comfort vote (TCV) also had a poorer value with open windows (−0.44 versus −0.35 from 0 to −4), also showing a higher deviation in the votes (0.75 versus 0.54) and a somewhat higher linear correlation with CO2 concentration. Therefore, the architectural design should take into account to guarantee the air quality of the venue, as well as a comfortable heating system, in order to lead students to not open the windows uncontrollably, which produces, as explained above, a noticeable energy consumption and distorts interior comfort control.

The operation of windows during winter helps to decrease the mean value of CO2 concentration, with 1537 ppm versus 2164 ppm with windows closed; but, in most of cases, this decrease was

insufficient both to be within the standard recommendations for healthy environments and to reach threshold values of perceptions of the users. Given that the mean CO2 concentration level was still high even when windows were open, the mean environmental perception of the students (EPV) was not strongly influenced by the opening of windows, with almost 30% of students expressing a certain level of annoying odor in both cases, but also having a moderate correlation between poor environment perception and CO2 concentration just when windows were closed. Therefore, it can be stated that there was not a high correlation between the CO2 value and the students' perceptions, mainly due to the olfactory adaptation phenomenon, irrespective of the need to provide a suitable air quality for healthiness purpose. In this way, when symptoms reported were added to this analysis, they presented a not-direct relationship with EPV, with the higher complaint values when windows were open. This odor perception was also somehow related with tiredness, difficulty on concentrating, eye irritation, headache, and dry throat.

In midseasons, windows' operation led to a greater variation of indoor thermal values, both air and radiant, also maintaining in general CO2 levels over the WHO recommendations (mean vale of 1537 ppm). In addition, students' TSVs were higher with open windows, close to the thermal comfort limit by warmth. Furthermore, the odor perception (EPV) was also poorer (0.63 value versus 0.52) when windows were open in midseasons, reinforcing the finding that windows alone are not able to provide an adequate renewal capacity for the indoor environment.

The study of the symptoms reported during measurements showed that they were largely expressed by students, both for windows open and closed, particularly in the case of difficulty of concentrating (52%), headache and tiredness (46%), followed by dry throat and nasal congestion (39%), which also were the symptoms most frequently combined with the other symptoms. According to the studied scenario, without a mechanically controlled ventilation system, complaints were more often found during winter, especially when windows were closed. In midseason conditions, symptoms were somewhat less common, but students expressed more acute symptomatology when windows were open, especially for dry throat, itchiness, nasal congestion, and headache, which are symptoms that can be related to hypersensitivity to external agents such as allergies and other respiratory conditions. This conclusion states the clear need to provide a ventilation system with a suitable filtering.

Regarding the relationship with indoor temperature, it can also be established that the symptoms tended to be more frequent when indoor temperatures increased, also with specific patterns for winter and middle season, also related to the occupants' thermal perceptions.

Other operation factors, like illuminance and illuminance uniformity, as well as students' clothing insulation, were analyzed referred to this symptomatology, but no clear correlation was obtained. In the case of lighting parameters, almost all the classrooms under study were generally over 350–400 lx with a uniformity of 0.40–0.50, so they were values good enough to not influence students at a symptomatic level. The correlated color temperature was similar in all cases, varying from 3500 to 5500 K; hence, it can be considered that both the amount of light and hue did not affect the thermal perception of the participants. On the other hand, students had the possibility of freely varying the level of clothing insulation, according to their individual thermal needs, so its impact on symptomatology was diminished.

In conclusion, the findings of this study show that effectively controlled ventilation systems are needed to assure an actual indoor ambient renovation and clean air supply. The special sensibility to external species make it advisable to incorporate filtering and cleaning systems for outdoor air beyond the impact on investment costs and energy use that this may entail. In addition, the study of symptomatology suggests that CO2 indicator should be complemented by other pollutants' measurements to assure a proper interpretation of data, given that they could not be correctly identified exclusively using this single CO2 control parameter. As explained above, CO2 levels have a fuzzy influence in the students' symptomatology; hence, the air quality should be complementarily assessed through other parameters, such as particle or VOCs' levels.

The following points can be established as key aspects:

The use of CO2 as a standalone indicator of environmental quality, especially for the management of ventilation systems or driving the windows' opening, may be insufficient and can derivate in situations of increased user discomfort, alongside thermal-ambient disturbance. Although there was evidence that there is a relationship with the indoor CO2 levels growing (assumed as general index) and the increase in reported global symptoms, this was not a direct link and tended to be asymptotic from certain threshold levels (around 2000 ppm).

In most cases, natural ventilation systems are not able to solve properly the removal of pollutants, generating situations with high rates of complaints even when windows are open, although they can mitigate the situations during indoor peak situations (such as produced in winter season). In many cases, windows' opening can be counterproductive, given that, although the classic indicators of the indoor environment valuation improve, the perception of the users was negative or, at least, worse than in situations with closed windows.

Assuming that indoor ambient is a complex and multifactor model, in the current state of the art of school buildings, the use of natural ventilation by itself (with the typical configuration of classrooms and enclosures of the buildings in the region) does not guarantee adequate control of the indoor environment, against popular assumption in the area, both by users and administrators. This aspect, although it was previously included in the text, has been emphasized.

This fact may be related to the need to review the classic indicators and parameters commonly used in the environmental management of these spaces. This research found situations of discomfort even within the ranges generally assumed as comfortable by the standards and design guides. Thus, it is necessary to develop complementary indicators based on the perception and the probability of developing symptoms that allow contributing to the correct valorization of the indoor environments from the users' points of view.

In this way, this analysis should also be complemented with corresponding measurements and surveys distributions in classrooms with mechanical ventilation systems in order to develop a comparison of results with adequate CO2 levels, so further research on this field is required.

**Author Contributions:** Conceptualization, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; methodology, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; software, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; validation, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; formal analysis, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; investigation, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; resources, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; data curation, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; writing—original draft preparation, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; writing—review and editing, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; visualization, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; supervision, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; project administration, M.Á.C.-L., S.D.-A., J.F.-A., and I.A.; funding acquisition, M.Á.C.-L., S.D.-A., J.F.-A., and I.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partially funded by the PIF Program of the Universidad de Sevilla (V Plan Propio).

**Acknowledgments:** The authors wish to express their gratitude to Jaime Costa-Luque for reviewing the manuscript and helping with several of the graphics, to Blas Lezo for the encouragement of this article, as well as to the students, teachers, and management team of the secondary schools which were part of this study. Finally, to the Public Entity "Agencia Pública Andaluza de Educación" from the Regional Government of Andalucía.

**Conflicts of Interest:** The authors declare no conflicts of interest.
