The role of clothing in thermal comfort : how people dress in a temperate and humid climate in Brazil

hermal insulation from clothing is one of the most important input variables used to predict the thermal comfort of a building’s occupants. This paper investigates the clothing pattern in buildings with different configurations located in a temperate and humid climate in Brazil. Occupants of two kinds of buildings (three offices and two university classrooms) assessed their thermal environment through ‘right-here-right-now’ questionnaires, while at the same time indoor climatic measurements were carried out in situ (air temperature and radiant temperature, air speed and humidity). A total of 5,036 votes from 1,161 occupants were collected. Results suggest that the clothing values adopted by occupants inside buildings were influenced by: 1) climate and seasons of the year; 2) different configurations and indoor thermal conditions; and 3) occupants’ age and gender. Significant intergenerational and gender differences were found, which might be explained by differences in metabolic rates and fashion. The results also indicate that there is a great opportunity to exceed the clothing interval of the thermal comfort zones proposed by international standards such as ASHRAE 55 (2013) 0.5 to 1.0 clo and thereby save energy from cooling and heating systems, without compromising the occupants’ indoor thermal comfort.


Introduction
Thermal insulation from clothing is one of the most important input variables used to predict occupants' indoor thermal comfort (AMERICAN…, 2013a).In both of the most common and current thermal comfort standards, clothing is an imperative protagonist.It represents one of the six variables used to calculated PMV/PPD in the static approach (FANGER, 1970), and for the adaptive method, clothing insulation worn inside buildings is considered as a key feature of the occupants adaptation, strongly dependent on both mean indoor and outdoor temperatures (DE DEAR;BRAGER;COOPER, 1997).
As a personal factor, the amount of clothing that people wear inside buildings is highly bound to human subjectivity, also involving important factors such as climate, culture and local fashion (HAVENITH;HOLMÉR;PARSONS, 2002;MORGAN;DE DEAR, 2003;CHUN et al., 2008;KWON;CHOI, 2012;SCHIAVON;LEE, 2013).Within the range of opportunities that occupants make use of strategies to adapt and optimize their thermal sensation, the first and most immediate one -perhaps even an unconscious one -is the clothing adjustment by removing a garment when they feel warm, or adding another layer when feeling cool.
Even though the importance of clothing on thermal comfort evaluation is recognized, not many papers presented results focusing on the subjectivity of clothing insulation.Nevertheless, some studies discussed the neglected factors on clothing calculation (HAVENITH; HOLMÉR; PARSONS, 2002), the variance of clothing insulation across the seasons in Korean (KWON; CHOI, 2012) and the impact of cooling strategies on thermal comfort factors from air-conditioning and naturally ventilated office populations in Thailand.Based in ASHRAE RP-884 database, two studies were performed connecting dynamic models to predict clothing insulation -which was subsequently adopted by ASHRAE 55 Standard (AMERICAN…, 2013a) -and also people's clothing behavior as a function of external and internal temperature (outside, mean weekly outside and internal temperatures); further including Singapore and Indonesia data (DE CARLI et al., 2007;SCHIAVON;LEE, 2013).Type of industry, gender and indoor settings have also been investigated by Morgan and de Dear (2003), and the energy conservation potential due to flexible dress-code such as the Super Cool Biz initiative in Japan was discussed by Indraganti, Ooka and Rijal (2013) and Tanabe, Iwahashi and Tsushima (2012).
Florianópolis is a capital city located in the coast of Santa Catarina state, south area of Brazil (latitude 27°40'S).According to Köppen's classification, the city is represented by a temperate and humid climate, with external temperatures varying from 17 to 29°C during the summer and spring time, 13 to 22°C during the autumn and winter (GOULART; LAMBERTS; FIRMINO, 1998).Relative humidity is high throughout the year (minimum monthly average is 80% in November and maximum monthly average is 84% in July) and there is no dry season.The highest rainfall occurs from January to March, and the lowest from July to August (mean annual precipitation is 1,521mm).There is a higher prevalence in winds in the northern direction, followed by the southern direction.
The field studies were conducted during all four different seasons.Details about the external air temperatures during the measurements, the related seasons and building type are summarized on Table 1 and 2  Simultaneous air temperature, humidity, globe temperature and air speed measurements were carried-out with laboratory precision using a microclimatic station named SENSU1 located in the center of the room (0.60m from the floor).
Measurements were taken continuously while subjects answered a thermal comfort questionnaire (paper questionnaires in classrooms and via electronic software in office environment).Individualized air speed readings were also taken from the occupied zone for each subject, using a handheld hot-wire anemometer sensor2 .Questionnaire items asked subjects to assess their thermal environment on right-here-right-now basis and it featured classic thermal sensation, preference and acceptability questions, air movement acceptability and preference, prior exposure to air-conditioning and cooling preference.For the purpose of this paper, only the question regarding thermal comfort was used for further analysis (right now, is this environment thermally comfortable for you?).Anthropometric characteristics (gender and age) were also collected by means of questionnaires fully filled by the occupants.Each subject stayed inside the building and answered the questionnaire approximately five times throughout a 140-minute survey period (the participation was completely voluntary, and the occupants were aware of the entire protocol before starting the questionnaire).
Full details about the whole study and the complete version of the questionnaire can be found in De Vecchi (2011, 2015).
Subjects were also asked to select the garment items they were wearing from a standardized checklist (Table 3) based in ASHRAE 55 (AMERICAN…, 2013a) just before the beginning on experiments.It is important to highlight that according with the ASHRAE Handbook Fundamentals (AMERICAN…, 2013b)), the clochecklist method can represent a ±25% error in the clothing insulation value, which is a common consequence of most real field studies in all over the world.Thereat, it was assumed that if any error exists in the sample, they are uniformly distributed and represents a low impact in the final comparisons.The changes in the occupants' clothing layers were monitored through questionnaires and site observations registered by the researchers.
The clothing ranges found in each type of buildings were depicted on Table 2 above.It was automatically assumed that all participants were using underwear (briefs for men and panties and bra for women, corresponding to 0.04 clo for both of them).Also, clothing estimates included the incremental effect of the upholstered chair upon which the subject was sitting at the time of the questionnaire -0.10 clo for a standard office chair in classrooms settings and 0.15 clo representing an executive chair in the office settings.
Operative temperature was calculated in accordance with normative appendix A from ASHRAE 55 (AMERICAN…, 2013a), and statistical analysis was performed using IBM SPSS software (Statistical Package for the Social Sciences -SPSS Inc., Chicago, IL, USA; 22.0 version).

Results and discussion
For the purpose of this paper, analysis focused on subjects' clothing insulation and its variations observed during seasons, operative temperature values recorded indoors, type of buildings and also subjects' age and gender.Figure 1 shows a distribution of the calculated operative temperature during the site observations and Figure 2 shows the frequency of collected data according to the seasons.In Figure 1, the calculated range of operative temperature varies between 21°C and 28°C.The main sample is slightly concentrated in autumn (47%) followed by winter (27%), spring (17%) and summer (9%).Results in Figure 3 show that thermal comfort votes indicated a 80% satisfaction reported by building occupants when exposed to operative temperatures spanning from 21°C to 26°C in all building types investigated here, suggesting a worthy ratio between occupants clothing and experienced thermal conditions.In type 1 building, operative temperature has not exceeded the range between 22°C and 24°C due to the tight temperature control observed in the HVAC system.Based on these results, comfort interval was assumed between 21°C and 26°C, and data beyond this range were rejected on further analysis.

Seasonal and ventilation strategy impact on clo values
Figure 4 shows the frequency of clo values observed during different seasons of the year.Not surprisingly, occupants' clo values were strongly influenced by season -values higher than 0.9 clo occurred most frequently in winter, while 0.3 -0.5 clo were registered during summer.Apart from the two well-defined seasons (summer and winter), results from autumn and spring show a strong fluctuation in clo values (0.3 -1.0 clo), which matches the characteristically variability in the outdoor weather observed during these seasons.Similar results were reported by Morgan and de Dear (2003), and Kwon and Choi (2012).These observed differences on occupants' clothing, during the four seasons, could be directly related to the thermal conditions delivered indoor by each building type investigated here.The same way that outdoor weather influences our choice of clothing, the thermal conditions people expect to be exposed indoors may also play a role here which in turn may or may not be linked to the outdoor weatherfor instance subjects may be wearing a cardigan inside of an office building in the middle of summer.Figure 6 shows the outcomes for all buildings type: (a) a fully air-conditioned commercial building (type 1); (b) two mixed-mode government buildings (type 2); and (c) university classroom with a mixed-mode system (type 3) binned by operative temperature values.
The air-conditioning was used in 28% of the measured period in mixed-mode buildings (type 2 buildings) while in classrooms (type 3 buildings) the same usage achieved 52%.Type 1 and 2 buildingsboth office environments, the clo values varied very little while in Type 3 buildings a bigger variation was observed.This can be explained by the dress code observed in Type 1 and 2 buildings and also the age profile of the sample.The mild variation in clo values in type 2 and the higher variability in type 3 suggest that occupants may be taking advantage of the adaptive opportunities available to them in these mixedmode environments (changing the clothing, opening/closing windows or even turning on/off AC units).Figure 7 shows the average values of clothing insulation for each type of building binned by operative temperature values.In type 3 building, where occupants wore more casual garments and no dress code is required, clothing insulation is substantially smaller than others buildings.Comparing the results from the office buildings (1 and 2) it is possible to observe that clothing average values were statistically different and slightly smaller (≅ 0.1 clo) in the fully airconditioning building.Even though small, this difference shouldn't be completely ignoredresults reported by De Carli et al. (2007) indicate that even small difference in clo value of 0.1 in air-conditioning buildings may be sufficient to impact on occupants' comfort (and the PMV/PPD calculations).

Gender and age groups differences in clo values
The frequency of each clo value bin ranging from 0.3 to 1.5 for male and female occupants were illustrated in Figure 8.It is worth mentioning the difference of clo value distribution between male and females.For males, there is a much higher frequency of clo value in the 0.6 bin, which represents a typical ensemble of long trousers and a long (short)-sleeve shirt in the work environment.As for females, the frequencies are more spread out between different bins, meaning that there are more variations in terms of what they are wearing for work.These differences were also observed when each type of building was analysed separately.Females presented a higher value of clo in all occasions (0.04 higher in HVAC buildings and 0.06 higher in mixed-mode buildings and classrooms).
Although the dressing code in the work environment is similar between different genders, Figure 9 does show a much higher variation of women's clo values than men's.Figure 10 also reports a higher SD value for females than males, which corroborate with the findings in Morgan and de Dear (2003).
Figure 10 presents the average clo value along with its 95% confidence intervals across a range of operative temperatures for males and females.Except for the operative temperatures of 21°C and 24°C where the two confidence intervals overlap, other operative temperatures have observed significantly different clo values (p < 0.05) between males and females, with females generally 0.1 to 0.2 higher than males.Theses results may be driven by the higher cold sensitivity that women usually demonstrate, as reported in several comfort studies (NAKANO; TANABE; KIMURA, 2002;MORGAN;DE DEAR, 2003;KINGMA; FRIJNS; VAN MARKEN LICHTENBELT, 2012;SCHELLEN et al., 2012).The higher variability in women's dresses that are discussed previously could also contribute to this difference.According to Morgan and de Dear (2003), men tend to follow a more fixed dressing code than women do, regardless of outdoor temperatures.Another aspect regarding clo values is the intergenerational differences.Figure 11    For classrooms and office work environments, there are different clo values observed even at the same temperature.As much as both of analyzed kinds of buildings can be classified as prolonged exposure environments, a large difference between student's expectations about their daily activities could have an impact on the choices of clothing, as well as the nonexistence of a corporate dress code.These assumptions possibly led them to use lower levels of clothing than office workers.The clothing behavior was also compared across the commercial buildings with different cooling systems, and it was observed a slightly higher variability on clothing insulation in the mixed-mode building (a 0.04 difference on standard deviation), which also represented the model with higher Pearson's coefficient (35% compared to 26%).These results highlight the impact of indoor climate on clothing insulation levels and possible implications for the energy consumption.Wearing heavier clothes during winter and lighter clothes during summer can reduce the heating and cooling loads; mixedmode buildings can serve as an example of more energy efficient buildings and more comfortable environments -especially when controlled by occupants.
Back to the dress code issue, differences between "casual clothing" and "business clothing" environments represented in this study by students and office workers raise another discussion about initiatives such as Cool Biz in Japan and the "notflexibilities" of the corporate dress code applied in large and hot urban centers of Brazil such as Florianópolis.On these assumptions, female occupants may be privileged, given the flexibility they can have on day-to-day corporate clothing verified in this study.However, due to their higher sensitivity to cold environments and differences on metabolic rate discussed in previous papers, women might be dealing with the over cooling inside buildings (or, just with the unnecessary use of air conditioning in mixed-mode environments) through warmer clothes, while men have relatively fixed dress code.
Whether the intergenerational difference represents an important factor to be considered in dress patterns is still under question; Yet, the results from this paper call for a further examination on the thermal comfort impact of fixed and liberal dress codes during the hottest days.

Figure 1 -Figure 2 -
Figure 1 -Distribution of the operative temperature during the field experiments

Figure 3 -Figure 4 -Figure 6 -
Figure 3 -Frequency of thermal comfort votes according with the three types of building and binned operative temperature (°C)

Figure 7 -Figure 8 -
Figure 7 -Clothing average associated with the type of buildings and the operative temperature (°C)

Figure 11 -
Figure 11 -Clothing distribution according with the occupants' age

Table 4
summarizes the subjects' age profile broken down per building type investigated.Type 1 and 2 buildings show a slightly higher concentration of occupants in the middle age group range (31 ≤ age ≤ 50).Type 3 building hosts a majority of young occupants (age ≤ 30) and a reasonable balance between females and males is observed.