3.1. Occupant Thermal Comfort according to the Difference between Indoor Air Temperature and Radiant Temperature
Figure 3 shows the changes in indoor air temperature (T
A) and radiant temperature (T
r) over time. Both T
A and T
r were high near the windows in the north and in the afternoon. During the experiment, T
A ranged from a minimum of 23.15 °C at 10:00 to a maximum of 28.93 °C at 17:00, while T
r ranged from a minimum of 23.25 °C at 10:00 to a maximum of 33.35 °C at 17:00. Though the minima and maxima of both temperatures each occurred at the same time of day, T
A increased from north to south whereas T
r increased from northwest to southeast.
The distribution of Tr showed notable characteristics that were not observed in that of TA; Tr increased in the central part of the indoor space over time and was exceptionally high in the southeast direction. These differences resulted from the presence of furniture (a partition) installed in the office space, suggesting that the radiant temperature was exceptionally high because the heat could not be dissipated in the air.
Figure 4a,b show the TSV and CSV responses of the subjects, respectively, according to T
A. The blue area indicates that the indoor air temperature is the same as the radiant temperature, whereas the red area indicates that these temperatures are different. In
Figure 4a, when T
A = 28 °C (ii), the responses exhibit similar distributions regardless of T
r, with the largest percentage of responses (~35%) indicating the slightly warm TSV (5) when T
r = 28 °C (blue), whereas when T
r = 30 °C (red), the largest percentage of responses (~43%) indicated the neutral TSV (4). However, when T
A = 26 °C (i), a wider distribution of responses can be observed for T
r = 28 °C (red) than for T
r = 26 °C (blue), with ~44% indicating the neutral TSV (4) and 20% each indicating the cold (1) cool (2), and slightly warm (5) TSVs in the former case. Thus, the distribution of TSV responses was wider when
Tr was not the same as
TA. Therefore, TSV results responded the most to the neutral when the indoor temperature was 26 degrees (i). In addition, when the indoor temperature was 28 degrees (ii), the response was highest to neutral and slimly warm.
In
Figure 4b, when T
A = 26 °C (i), the largest percentage of responses indicated the neutral CSV (4) whether T
r = 26 or 28 °C. However, when T
r = 26 °C (blue), more responses were distributed to the two uncomfortable CSVs (2 and 3), whereas when T
r = 28 °C (red), more responses were distributed to the two comfortable CSVs (5 and 6). When T
A = 28 °C (ii), the same percentage of responses (28%) indicated the comfortable CSV (6) for T
r = 28 and 30 °C. However, 43% of responses indicated the neutral CSV (4) when T
r = 30 °C (red), whereas only ~20% of responses indicated the neutral CSV (4) when T
r = 28 °C, with 28% indicating the slightly uncomfortable CSV (3). Thus, the distribution of CSV responses was wider when T
r = T
A. Therefore, most of the occupants responded to the neutral when the indoor temperature was 26 degrees (i). In addition, when the indoor temperature was 28 °C (ii), the response was high to neutral and comfortable.
The analysis results presented in
Figure 4 indicate a clear difference between the TSV and CSV results. In addition, considerably different thermal comfort results were observed according to the indoor air temperature and radiant temperature. In particular, when the indoor air temperature was different from the radiant temperature, the TSVs of the occupants were distributed among several different and often opposing scores. In other words the blue areas in
Figure 4 indicate an even distribution across the five vertices, whereas the red areas indicate an uneven distribution with a high response at a few specific vertices. It was also found that the comfort of the occupants varied depending on both the radiant temperature and the indoor air temperature. This confirms the influence of the building characteristics (curtain walls); it appears that the radiant temperature showed significant changes under the influence of the solar radiation introduced through the curtain walls, which also had a significant influence on thermal comfort. Thus, a simple comparison between the indoor air temperature and radiant temperature is only of limited utility when analyzing the thermal comfort of office occupants. To provide a more detailed analysis, thermal comfort was estimated in
Section 3.2 based on the accumulated mean thermal sensation vote (mTSV) according to the temperature change.
3.2. Thermal Sensation Changes and Comfort Temperature over Time
The changes in thermal sensations of the subjects were analyzed using the TSV reported at each location. In previous studies, TSV has been used as a representative index for general thermal comfort evaluation [
53]. The changes in TSV over time are shown in
Figure 5, in which the greater the value, the larger the change. The average change in TSV at all ten subject locations was 0.55, with an average maximum of 1.80 and an average minimum of 0, indicating that there were subjects who experienced no thermal change. A σ of 0.45 was calculated for all data. At 10:00, 14:00, and 17:00, the average change in TSV was 1.22, 1.12, and 1.23, respectively, at point A; 0.58, 1.08, and 0.71, respectively, at point B; 0.02, 0.35, and 0.40, respectively, at point C; 0.90, 1.80, and 1.05, respectively, at point D; 0.13, 0.17, and 0.32, respectively, at point E; 0.34, 0.69, and 0.50, respectively, at point F; 0.31, 0.61, and 0.67, respectively, at point G; 0.52, 0.92, and 0.25, respectively, at point H; 0.00, 0.06, and 0.05, respectively, at point I; and 0.21, 0.20, and 0.11, respectively, at point J. Thus, the changes in TSV at point D were the largest, followed by those at A, B, H, G, F, C, E, J, and I.
The changes in TSV at each point were divided into two groups according to their proximity to the windows. Points A, B, D, F, G, and H, which showed changes exceeding the average, were combined together in Group A (Male 50: Female 50); all of these points were close to the windows, where they received high solar radiation. Points C, E, I, and J, which showed changes smaller than the average, were combined together in Group B (Male 15: Female 75); all of these points were relatively far from the windows and closer to the interior corridor. The average and σ of the change in the TSV of Group A were 0.81 and 0.13, respectively, whereas those of Group B were 0.17 and 0.07, respectively, confirming a significant difference between the thermal comfort experienced by the two groups over the course of the day.
The overall σ of the changes in TSV was 1.31. The mTSV was calculated to quantify the change in TSV over the course of the day, with a positive value indicating a change to a warmer sensation, zero indicating no change, and a negative value indicating a change to a cooler sensation. For Group A, mTSV was 5 (slightly warmer) and the corresponding σ was 1.25. For Group B, mTSV was 3 (slightly cooler) and the corresponding σ was 1.19.
The mTSV results of each group were analyzed using a Probit regression analysis to estimate their respective levels of thermal comfort [
54]; the results are shown in
Figure 6. In
Figure 6a, the highest curve represents the warm TSV (6) and the lowest curve represents the cold TSV (1). As the temperature increases from 23 to 28 °C, the curves representing the warm (6) and slightly warm (5) TSVs increase faster than the others. In
Figure 6b, Group B only reported only five TSVs from cold (1) to slightly warm (5); the highest curve represents the slightly warm TSV (5) and the lowest curve represents the cold TSV (1). As the temperature increases from 22 to 26 °C, the curve representing the slightly warm TSV (5) increases faster than the others.
For both groups, the lowest curve represents the cold TSV (1), but a generally warmer thermal sensation scale is shown in
Figure 6a than in
Figure 6b because the highest curve in the former represents the warm TSV (6) whereas that in the latter represents the slightly warm TSV (5). In addition, the proportion of cold TSV (1) decreases with increasing indoor air temperature in
Figure 6a but increases with increasing indoor air temperature in in
Figure 6b. The comprehensive analysis results thus confirm that Group A felt warmer than Group B, indicating that the vicinity of Group A to the windows resulted in a higher temperature in the same office space. Based on these results, it can be determined that a clear difference in comfort temperature exists between the two groups and thus in the thermal environments of their respective spaces.
The comfort temperature results for each group, derived by regression analysis of the neutral score in
Figure 6a,b, are shown in
Table 6. The comfort temperature was derived from the T
op and the mean of TSV. It was found to be at a significant level through the derived R
2 and
p values. The average comfort temperature for all groups was found to be 26.6 °C. Goto et al. [
55], who used the same indoor set temperature as this study (26 °C), stated that the comfort temperature preferred in office buildings was approximately 26 °C. Furthermore, Madhavi et al. [
56] identified 27.1 °C as the comfort temperature, which is also similar to the average comfort temperature obtained in this study.
The comfort temperature for Group A was found to be 24.7 °C for whereas that for Group B was 28.4 °C. Thus, the respondents in Group A felt comfortable at a temperature approximately 2 °C lower than the current indoor set temperature of 26 °C, whereas those in Group B felt comfortable at a temperature approximately 2 °C higher than the set temperature. Consequently, the comfort temperatures in different spaces differed by approximately 3.7 °C, indicating that the comfort temperature is clearly dependent on the occupant location, even in the same space.
Tanabe et al. [
57] performed experiments in six different office buildings, finding that the occupant thermal satisfaction level was 75% at an indoor air temperature of 25 °C, but dropped to 40% at 28 °C. However, a comfort temperature of 28.4 °C was obtained for Group B in this study. Thus, despite a consistent temperature of 28°C, the analysis results were different depending on the presence of solar radiation and occupancy environment.
3.3. Analysis of Sensitivity to the Mean Thermal Sensation and Indoor Operative Temperature
Figure 7 shows the sensitivity analysis of the mean thermal sensation (MTS) to the indoor operative temperature (T
op) for Groups A and B, in which a strong positive correlation can be observed between MTS and T
op [
39]. In the figure, the size of the bubble is proportional to the number of responses to the temperature change in the TSV results, and the slope of the regression equation thus represents the thermal sensitivity of the group to changes in T
op. A low sensitivity indicates that the temperature change is not felt, whereas a high sensitivity indicates that the temperature change is directly felt. The overall average sensitivity was found to be 0.58/°C. The slope for Group A (0.67/°C) is higher than that for Group B (0.49/°C), indicating that Group A was more sensitive to changes in the indoor air temperature, and that Group B could accept a higher temperature. Therefore, Group B can set the temperature higher than 26 °C. Group B feels that they are comfortable with the temperature higher than average. This control can save cooling energy and at the same time, can keep the comfort of occupants having the characteristics of Group B. This will be a positive control.
Group B was likely less sensitive than Group A because the increase in solar radiation admitted through the curtain wall over the course of the day primarily affected Group A. In addition, the amount of change in MTS with temperature confirmed that Group A could not adapt to the temperature change and reacted more sensitively because the large temperature change in the area near the windows owing to the influx of solar radiation. Therefore, a temperature lower than the setting temperature is needed to Group A. Going further, temperature control by hour for reacting to the influence of solar radiation is needed. Thus, the results of this study indicate that sensitivity differs significantly according to the indoor thermal environment at the occupant location in an office space.
A previous study on thermal comfort in various building types [
58] found that the universally adopted sensitivity (Griffiths constant) was close to 0.5/°C by Michael Humphreys. This value is not significantly different from the sensitivity results derived in this study; it is particularly similar to the sensitivity result obtained for Group B (0.49/°C). Furthermore, the comfort temperature (28.7 °C in [
59] compared to 28.4 °C for Group B in this study) of the occupants with low sensitivity (0.255/°C in [
59] compared to 0.49/°C for Group B in this study) was found to be relatively high. These occupants can clearly accept higher temperatures in summer.
Indeed, this study showed similar estimates for sensitivity as previous studies. Rupp et al. [
58] estimated that the thermal sensitivity of office buildings in a subtropical climate, which is similar to the climate of Korea, was 0.568/°C. They also derived a sensitivity of approximately 0.440/°C in air-conditioned offices. These circumstances are comparable to those obtained through field research in summer in this study, but the derived results are different. Additional data are therefore required to confirm the validity of these results. Rupp et al. [
58] mentioned that it remained necessary to avoid reliance on universal thermal sensitivity results by conducting further research because thermal sensitivity is not constant. They also emphasized the necessity of deriving the comfort temperature based on field research data. The field research conducted in this study was accordingly used to derive the comfort temperatures. However, research including more variables remains required to compensate for the limitations of such field research at present.
Thermal comfort is significantly affected by the exterior, type, and geographical location of a building along with situational factors and the human body’s thermal balance [
39]. In this study, field measurements were performed, and subjective sensations collected in an office space experiencing a large influx of solar radiation in a building with a transparent envelope. The analysis confirmed that the solar radiation introduced into an indoor space has a direct impact on the indoor air temperature and the thermal comfort of the occupants. Similarly, Moon [
15] conducted research on the relationship between solar radiation and indoor thermal comfort and emphasized that solar radiation must be considered during the design of HVAC systems, as solar radiation increases body temperature [
11,
17]. Since numerical data describing the influx of solar radiation were insufficient in this study, it remains necessary to collect additional relevant data. The inclusion of such data should be further discussed in future research to obtain definitive results describing thermal comfort in buildings. Thus, in future work, solar radiation data will be measured, and revised regression coefficients will be estimated using all relevant data.