Perceived thermal acceptability and behavioural adjustment for Indonesian workers

. Behavioural adaptation is a fundamental factor in thermal comfort. Clothing insulation, opening windows, and integrating cooling systems within buildings are key to ensuring thermal comfort while conserving energy. This study collected behavioural adaptation data from 3000 Indonesian adults who regularly worked indoors. The results indicate that the thermal acceptability is in a similar range for all groups separated by residence space type despite the dominant thermal conditioning measure in each group is different. Based on ISO 9920, the typical clothing ensembles were determined for each group, and their clo-values were estimated. Office workers tended to wear more garments than factory workers, while at home, people wear light clothes. Therefore, the estimated clo-value in home are much lower compare to the workplaces clo-value. Clo-value in the workplaces with AC tends to show a higher value compared to naturally ventilated building. Clo-value for female is also higher than male. In general, AC usage in the office was higher than in factories and homes. Females tends to use more AC than Males. These differences reflect variety of socioeconomic and cultural aspects in public and private in Indonesia.


Introduction
Behavioural adjustment is a fundamental factor in thermal comfort. This behavioural adjustment includes clothing adjustment, ventilation adjustment, room temperature adjustment, etc. Within ASHRAE 55, thermal comfort is a "condition of mind that expresses satisfaction with the thermal environment and is assessed by subjective evaluation". It is affected by six main factors, namely metabolic rate, clothing insulation, air temperature, radiant temperature, air speed, and humidity. However, as the assessment uses subjective evaluation, it is difficult to collectively satisfy the thermal comfort of a group of people in the same room due to various types of behavioural adjustment [1].
Researchers have found that thermal comfort can also depend on much more complex factors such as socioeconomic and cultural aspects [2][3][4][5]. As an example, in Muslim-majority countries, clothing insulation is not a flexible temperature control option, as women and men have specific attires. Indonesia, as the fourth most populous country, has a wide variety of socio-economic and cultural conditions among its people. It influences adaptive thermal comfort behaviours and thermal perception in Indonesians [6]. ________________________________ * Corresponding author: sr.apriliyanthi@gmail.com A common misconception is that in hot, humid tropical climates like Indonesia, people tend to set their air conditioners to the lowest possible setting to achieve a comfortable thermal environment. This causes a heavy burden on the cooling energy consumption of the building [4][5][6][7]. Some researchers found that instead of focusing on the room temperature adjustment using the AC, thermal comfort for hot-humid climates can be adjusted by manipulating other means such as clothing or wind speed [8][9][10]. Therefore, one possible method to ensure thermal comfort while ensuring lesser energy consumption is integration of clothing adding/removal, window opening/closing, operation of other cooling devices such as fans into thermal environmental adjustment in building.
The present work analysed consciousness of Indonesian behavioural thermal condition adjustment and their thermal acceptability based on a large set of online data on Indonesian workers. This data includes their perceived, thermal environment condition, clothing ensembles, and adaptive thermal adjustment behaviours in the workplace and home.

Data collection
Up to January 2022, a large-scale online survey was conducted using a web-based questionnaire platform developed by an international research company branched in Indonesia. The questionnaire contains three sections on personal characteristics, building characteristics, and clothing preferences.

Personal characteristics
This section includes the personal information of the respondent, such as occupation, field of industry, preferences for thermal adjustment behaviour, etc.
The web-based questionnaire was distributed to the enrolled population database through a research company. As a result, 3000 samples were collected from 31 provinces across Indonesia. These samples consist of 1541 male respondents and 1459 female respondents within the working age. The overview of the respondents' age and workplaces are listed in Table 1. Meanwhile, Figure 1 compares the distribution of the sample province and the actual distribution of population in percentage. The distribution of collected data in the middle part of Indonesia are mostly dominated by South Sulawesi and East Kalimantan, meanwhile only a small number of samples manage to be collected from eastern part of Indonesia.

Building characteristics
This section includes the characteristics of building where the respondent is working usually and those of respondent's home. They include the information on building type: Air Conditioned (AC) building and Naturally Ventilated (NV) building. They also include the respondent's perceived adaptive behaviour in each of the building, such as the intensity (ratio of time in a day) of AC usage, fan usage, window opening time and portable fan usage. Respondents also asked to state their perceived thermal acceptability for their main working room and their main room in home.

Clothing ensembles
This section includes the information of the most frequently worn clothing ensembles for each respondent Fig. 1 The online questionnaire data population compared to Indonesian actual population in each province in the workplace and home. In this section, detailed clothing garments list were divided into four groups of respondents based on the respondent gender and working places: men in workplace (office/factory), men in home, women in workplace (office/factory), and women in home. In each group, respondents select their most frequently worm clothing ensemble by combining the type of garments.
The ensembles for men in workplaces compose of item categories such as outerwear, sweaters, shirts, undershirts, trousers, underpants, coveralls, socks, footwear, and other accessories. Men  These items were carefully selected based on clothing insulation databases [11], [12] and pilot study of commonly used garments in Indonesia. To visualize the garments, the questionnaire displayed a small example image of each garment. A total of 446 garments selections were listed in the questionnaire.

Clothing ensembles analysis
Due to a large number of types of garments, the garments type in each category were classified into several sets based on the ratio between covered area by the garments and human body surface area. These sets for garments then referred to as the binary components for the following hierarchical cluster analysis. The classification was done to simplify the definition of clothing garments to make the clusters. The clothing ensemble were then validated using hierarchical cluster analysis (Ward's method), as the collected binary data for each garment set. Unrealistic samples were regarded as invalid samples and excluded for the following study. An example of these unrealistic samples was the sample that selected at least one garment from each category. The estimated clo-value of these excluded ensembles are considered too high for the hot-humid climate of Indonesia. Further data collection, especially through actual field studies, may be required to validate these types of clothing ensembles.
Furthermore, hierarchical cluster analysis divided the clothing ensemble of each samples' group into several distinguished clusters, for example, the male office workers group is divided into 11 clusters, the male factory group is divided into 7 clusters of clothing ensemble, and the male home group is divided into another 7 clusters with a considerably smaller number of garments in one set of clothing ensembles. The clustered clothing ensembles are defined as the typical clothing ensemble for the respondent group. The typical clothing ensemble corresponding to each cluster was given using the following rules: When more than 50% respondents in the cluster selected one or more garments from one garment set, the garment set became the components of the typical clothing ensemble of the cluster. This garment set is represented as 1. If selected by less than 50% of respondents, the garment set is represented as 0. Aside for validating the ensembles, these clusters might be useful in the future study.

Estimation on clo-value
Clo-values for each garment are estimated based on ISO 9920. Indonesian garments for which clo-values are not indicated are estimated from the values of garments with similar properties: body coverage, fabric material, or clothing thickness. The clothing thickness used as a reference for the local clothing was obtained from a clothing thickness experiment conducted before this online survey.
The clothing ensemble intrinsic insulation (Icl) can be calculated by Eq. (1) or Eq. (2). Both equations are using the total of each garment effective clothing insulation (Iclu).

Icl = ΣIclu
(1) Icl = 0.161 + 0.835 ΣIclu (2) In general, the accuracy of estimated clo-value for clothing ensemble by using Eq. (2) is higher compared to Eq. (1) [11,12]. However, clo-value from Eq. (2) is too high when it is applied for extremely light clothing ensemble, because Eq. (2) has positive intercept of 0.161. Therefore, clo-value were estimated using Eq. (2) for male in workplace, female in workplace and female in home. Meanwhile, male in home groups estimation used Eq. (1). This is due to the number of garments for each ensemble of male in home might be very small, which naturally resulting in a small clo-value as well.

Result and discussion
The data overview is shown in Table 2. The obtained samples show that the average clo-value of AC buildings is higher than that of NV buildings in almost all the datasets. On the other hand, the female clo-value is also higher than the male clo-value. As the building type is separated into office, factory, and home, it is also found that both, male and female, percentages of the office building with AC are 88.3% and 87.3% respectively. Meanwhile, AC installation ratio in factory and home is lesser than office. The AC usage intensity perceived in the office is also higher compared to the other type of buildings. Interestingly, the perceived AC usage intensity in the home for the female is higher than for the male. Despite that, the ratio of respondents' thermal acceptability stated that thermal environment is generally acceptable for all types of buildings. It is also did not highly affected by the AC installation.

Clo-value and thermal acceptability
The overview of the clo-value can be identified from Figure 2. Overall, the estimated clo-value for office workers tend to be higher than those of factory and home. Differences in the range of clo-values for AC and NV buildings are visible in both office and factory, but not visible in the home clo-value range. This could be explained by socioeconomic and cultural aspects that are applied differently in public areas such as workplaces and in private areas such as homes. There may be a certain limitation of freedom in which someone could control the thermal environment. In this case, alternative countermeasures, such as reducing clothing insulation, are done to keep the thermal acceptability. However, these differences appeared to be more subtle in the female cases for most types of buildings. It is due to other socioeconomic and cultural aspects. Some researchers have also found that the case of female clo-value is higher than male, and it is most likely related to the socioeconomic and cultural aspects [9,10]. Despite Indonesian people consist of various kinds of ethnicities and religious backgrounds, it is still a Muslim-majority country. Therefore, female is expected to dress in a certain manner that covers most of their body in public, such as wearing a hijab, long robe (also called gamis), long dress, etc. However, during their stay at home, they may dress as they prefer.
Although the female clo-value is higher compared to the male, the thermal acceptability trend between male and female are the same. More than 70% male and female reported that their thermal environment are acceptable or slightly acceptable, as shown in Figure 3.   Figure 4 shows AC usage intensity of the room based on their reported perceived thermal acceptability level. AC usage intensity means the perceived percentage of time ratio for the respondent in using AC in a certain room. There is a subtle hint that thermal adaptive behaviour are related to social status or socioeconomic and cultural aspects, such as unable to adjust the AC freely. On the average, female who stated that the thermal environment are acceptable tend to have higher AC usage intensity. On the other hand, males that have a lower average clovalue than female tend to have smaller AC usage intensity. In the unacceptable samples side, females tend to have lower AC usage intensity. In this case, insufficient cooling might be the cause of their perceived unacceptable thermal sensation. Meanwhile, for the male samples, the AC usage intensity range is scattered wider from the lowest to the highest intensity. The small sample number (see Figure 3) and large data scattering of the unacceptable sensation makes it difficult to clarify the relation between AC usage and acceptability. Further investigation is required.

Adaptive behaviour: AC
The samples distribution of AC usage preferences is shown in Figure 5. Although the majority of Indonesian people prefer to use AC, there are also quite some people who choose to be neutral about their AC preferences. Therefore, there is still a huge potential to persuade them to use other means of cooling, such as applicating a new way to dress like the Japanese cool biz or by encourage the building management to open the window [8][9][10]. Figure 6 explores the socioeconomic and social status aspects based on gender difference. The comparison of the relationship between AC preferences and actual AC usage intensity are shown separately for males and females. For females perceived AC usage intensity did not show a clear relation with their preferences in any kind of building. It shows that they are lacking control over the room thermal environment adjustment compared to males, both in the home and workplace.  For males, preferences of AC tend to match perceived adaptation behaviour in the home. However, in the workplace their control over the AC might be limited, therefore their AC perceived usage did not show a clear relation with their preferences. Just as found in female samples, the effect of socioeconomic is also observed for male samples in a slightly different way. There might be other reasons aside from thermal comfort when the males are adjusting AC, especially in workplaces. This reason might be related to social status, socioeconomic and cultural aspects. For example, as they are lower rank workers in the workplace, they cannot access the building HVAC system, which is adjusted by the building management of their company. These situations are especially expected in the factory. Other possibilities are such as they share their room with other co-workers. Therefore, further investigation is necessary.

Clo-value and AC preferences
In Figure 7, the male clo-value is not affected by their AC preferences. However, when the number of people Male samples Fig. 7 Relation between clo-value and AC preferences for male with perceived AC usage intensity above 50% is selected, the clo-value slightly increases. This indicates that people with lower clo-value tend to use the AC only for a short time. These people with higher clo-value and higher perceived AC usage intensity might not be able to control the AC settings. Therefore, to keep a comfortable heat balance in their body they are wearing a heavier set of clothing.
Meanwhile, as indicated in Figure 8, the female clovalue is not affected by both AC preferences and AC usage intensity. This strengthens the hypothesis that Indonesian female clo-value do not clearly show a relation to thermal adjustment factor.
According to the previous data shown in Figure 7 and Figure 8, AC preferences do not show a strong relation with clo-value for both males and females, although the AC usage intensity had a certain effect on male clo-value. The different trend of the AC usage intensity and clo-value between males and females can be explained by the socioeconomic and cultural aspects which tend to restrict female clothing options.

Conclusion
The study found that office workers tended to wear thicker clothing than factory workers both male and female. However, at home Indonesian people tend to wear lighter clothes. The estimated clothing insulation also show a higher value for the female samples compared to male. This is due Indonesia, as a Muslimmajority country, has a certain expectation on the female dresses.
Perceived AC usage in the office was higher than in factories and homes. AC preferences tended to be consistent with perceived adaptive behaviours at home for males but showed no clear relation with adaptive Female samples Fig. 8 Relation between clo-value and AC preferences for females behaviours at workplaces. Meanwhile, female AC preferences and perceived AC usage intensity did not show any strong relationship for any places. Overall, females' perceived thermal environment adjustments in Indonesia are strongly influenced by socioeconomic and cultural aspects compared to males despite the similar trend of thermal acceptability. Further study is necessary to find out the relationship with the other adaptive thermal behaviour such as fans, window openings, and portable fans.
The results obtained in this study can be used for future studies of adaptive thermal comfort models. This study also contributes to the development of new thermal comfort policies that promote behavioural adaptations such as clothing adaptation and window adaptation to countries with hot and humid climates like Indonesia.