1. Introduction
In countries in the Association of Southeast Asian Nations (ASEAN), rapid population growth is underway, and the energy demand in urban areas is increasing owing to urbanization [
1]. Indonesia has a population of 276 million as of 2021 and is expected to reach 300 million by 2031 [
2]. The urban population accounted for 49.8% of the total population in 2010, and the urban population is expected to continuously increase. Consequently, energy consumption in Indonesia continues to increase [
1,
3]. In particular, the household sector has the highest electricity consumption compared to other sectors, including the commercial, public, and industrial sectors; it was 49,790 GWh in 2008, increased to 95,329 GWh in 2016, and is expected to double to 183,600 GWh by 2027 [
4]. Moreover, 58% of the primary energy used for power generation in Indonesia is coal, and 27% is natural gas [
4]. Therefore, the further increase in household-sector energy demand is expected to increase GHG emissions, which would contribute to climate change.
Furthermore, the increase in the urban population is causing shortages in housing stocks. In 2023, 12.7 million units were estimated to be insufficient [
5,
6]. Consequently, the construction of collective housing is expected to rapidly increase against the backlog of future urban population growth. Since 2015, The Ministry of Public Works and Housing (PUPR) has been supplying low-cost public collective housing, also known as
Rusunawa, to replenish the housing stock and improve the living standards of low-income communities. In 2022, 7024 units of
Rusunawa were built, and a construction target of 13,500 units has been set for 2024 [
5,
7,
8,
9]. Collective housing, such as
Rusunawa, is being developed to support current and future urban populations in Indonesia. Thus, the implementation of energy conservation measures in collective housing from the development stage is considered to be effective to suppress future energy demand in the housing sector.
Regarding household energy consumption, previous studies have shown that the use of home appliances, especially the use of ACs, significantly contributes to the increase in household energy consumption in tropical regions [
10,
11,
12,
13,
14,
15]. For instance, Batih et al. [
10] investigated household electric consumption in the urban households of Indonesia through a bottom–up estimation. The results indicated that ACs, lighting, and televisions (TVs) were the appliances with the largest electric consumption. Furthermore, Usep et al. [
11] and Novianto et al. [
12] studied the effect of lifestyle changes by the COVID-19 pandemic on energy consumption of urban housing of Indonesia. It was confirmed in both studies that the increase in the use of ACs caused by the increase in time spent at home significantly increased household energy consumption.
Moreover, the socio-demographic characteristics of the household are also known to be a key factor in household energy consumption [
16,
17,
18,
19,
20,
21]. Ali et al. [
16] examined the determinants of household energy consumption in Malaysia through a multiple regression analysis. The results showed that income had the largest correlation with energy consumption, followed by education level and household size. Other studies investigating the relationship between household socio-demographic characteristics and energy consumption, such as those by Chen et al. [
17] and Kim [
18], concluded that income and family size largely impacted household energy consumption.
The prevalence of ACs in Indonesia is expected to increase due to economic growth even among low-income communities [
4]. Consequently, reducing the reliance on ACs is considered effective to suppress increasing household energy consumption. As a countermeasure, the implementation of NV is considered to have the potential to reduce dependance on ACs in tropical regions.
Several studies have investigated the effect of NV on thermal comfort [
22,
23,
24,
25,
26,
27], of which, Gou et al. [
22] examined the occupant perception of the thermal environment in naturally ventilated residences in Singapore. The results showed that occupants showed a higher heat tolerance than that outlined by the specifications of the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE). In addition, increasing the air velocity was found to have positive impacts on thermal comfort. Furthermore, according to the findings of Gamero-Salinas [
23], NV, wall absorption, solar protection, and having a semi-outdoor space had the strongest impact on lowering the overheating risks in tropical apartment buildings.
Building upon previous research, Mori et al. [
28] focused on the behavioral aspect of NV by extracting existing daily patterns of AC usage, fan usage, and window opening in terrace houses in Malaysia and collective houses in Indonesia. In addition, Mori et al. [
28,
29] determined the factors that inhibited window openings using a logistic regression analysis. The results suggested that ACs were mostly used in bedrooms during sleep, while non-AC users mostly opened windows during the day. In particular, the reasons commonly found for closing windows during the night for non-AC users were “Insect intrusion”, “Theft”, and “Privacy”.
Overall, there are a few reports that have investigated the effectiveness of energy conservation measures in the residential buildings of tropical regions [
10,
11,
12,
13,
14,
19,
20,
30,
31,
32]. However, very few have focused on collective housing exclusively. Therefore, to investigate lifestyle habits and preferences related to energy consumption in collective housing in the urban areas of Indonesia, the authors of this paper conducted a face-to-face questionnaire survey on energy consumption and space cooling methods targeting approximately 3200 households living in collective housing in the urban areas of Indonesia.
Given that the use of ACs has a strong impact on household energy consumption and NV has the potential to mitigate future reliance on ACs in tropical residences, this study aimed to clarify the impact of the use of ACs and the implementation of NV on energy consumption and thermal comfort by identifying existing space-cooling methods in collective houses of different income levels in Indonesia. Based on the results of a questionnaire survey conducted in 2022, the authors estimated household energy consumption and classified the patterns of three cooling methods (ACs, fans, and window opening) using a cluster analysis. The findings of this study serve as a starting point for verifying the energy saving effects of air conditioning habits for the purpose of decarbonizing collective housing, including future predictions and energy simulations. In addition, it contributes to promoting air-cooling strategies that achieve both the enhancement of thermal satisfaction and the reduction of energy consumption.
2. Methodology
The authors classified collective housing in Indonesia into the following three types (
Figure 1):
Rusunawa: public rental housing for low-income earners constructed and managed by the Indonesian government.
Rusunami: apartments for low- to middle-income earners managed by the services of the Indonesian government.
Condominium: apartments managed by the private sector without government assistance.
In 2022, the authors conducted a survey on energy consumption and lifestyle habits, targeting households living in these three types of collective housing located in the urban areas of Indonesia. Based on this survey, the authors estimated the amount of energy consumed by the utility costs for each building type. Furthermore, through a cluster analysis, the authors clarified the cooling patterns of three methods, ACs, fans, and window opening, and the respective cooling patterns for each building type to investigate the relationship between the cooling patterns, energy consumption, and thermal comfort.
2.1. Survey on Energy Consumption and Lifestyle Habits
A survey on household energy consumption and individual lifestyle habits was conducted from July to November 2022, targeting collective housing in major Indonesian cities. The survey consisted of two parts. Part 1 was answered by the head of the household regarding household energy consumption and appliance usage, whereas Part 2 was answered by individuals regarding lifestyle and preferences. An overview of the survey is presented in
Table 1. The survey items are listed in
Table 2.
2.2. Energy Consumption Calculation
The electricity consumption was calculated by considering different billing systems depending on the payment method. The formulas for calculating the electricity consumption for post- and pre-payment methods are shown in Equations (1) and (2), respectively.
Here, MECH represents the energy consumption per household [kWh/(month · hh)], x is the respondent’s monthly electricity bill [IDR/month], UC is the electricity usage fee [IDR/kWh], PPJ is the public street light tax rate, and SC is the stamp fee [IDR].
Table 3,
Table 4, and
Table 5 show the contract power, public street light tax, and stamp fees, respectively. In this study, the power consumption was calculated based on answers from memory. Therefore, the authors derived a correction coefficient based on the least-squares method with a fixed origin in the distribution of electricity consumption by a memory-based response and the actual value pair to correct the consumption by the memory-based response.
In the case of city gas consumption, the usage fee for collective housing was set at 4250 [IDR/m
3] [
33]. LPG consumption was calculated based on regional sales prices [
33]. Furthermore, for households that responded that they used an electric stove for cooking, samples without answers regarding gas (LPG and city gas) consumption were also recognized as valid. The monthly electricity and gas consumption calculated in the above process were converted to annual consumption by multiplying by 12, and a further secondary energy conversion was performed.
Table 6 presents the secondary conversion coefficients. Finally, outlier tests were conducted for electricity and gas consumption.
Table 7 shows the sample size after excluding the outliers.
2.3. Cooling Pattern Derivation by Cluster Analysis
The data used were binary codes for whether cooling methods were used or not for each cooling method, every 15 min for 24 h. Due to the large sample size and high dimensions of the binary data set, a non-hierarchical cluster analysis (k-means method) of the Euclidean distance was conducted using R Studio. The cluster analysis was performed with the number of clusters set to 3 until 8. Finaly, for each cooling pattern, the cluster number with the highest silhouette width was adopted as the final classification.
3. Results
3.1. Respondent Information
Table 8 presents the basic information on the survey respondents. The distribution of monthly income was the lowest in the order of
Rusunawa,
Rusunami, and condominiums, as well as the average floor area. On the other hand, the family size was largest in
Rusunawa and lowest in condominiums, indicating a decline in birth rate occurring in higher income families within collective houses.
Furthermore, this survey required detailed information on energy use at home. Therefore, it is important to note that the respondent was required to be the family member who was most knowledgeable about household finances. From this, it can be inferred that the respondents were heavily skewed towards women. Specifically, 82.2%, 64.2%, and 59.5% of the household managers were women in Rusunawa, Rusunami, and in condominiums, respectively. In other words, housewives in low-income households were commonly responsible for household management.
3.2. Energy Consumption in Collective Houses of INDONESIA
Figure 2 shows the annual household energy consumption. As a result, the average energy consumption of
Rusunawa, 9.2 [GJ/(year · hh)], was the lowest, followed by
Rusunami with 13.0 [GJ/(year · hh)], and condominiums with 20.4 [GJ/(year · hh)], indicating that households with lower incomes tended to have lower energy consumption.
3.3. Cooling Patterns
3.3.1. AC Usage Patterns
AC usage patterns were classified into four categories, as shown in
Figure 3. Among AC users, ACs were mainly used in bedrooms during sleep at night (A3, 21.5%). On the other hand, it can be seen that ACs were still not common in collective houses (A4, 69.8%).
Figure 4 shows the breakdown of AC usage patterns by building type. In condominiums, AC users (A1, A2, and A3) accounted for 76.1% of the total. However, in
Rusunawa, where the income level was the lowest of the three, more than 92.5% of the households did not use ACs. Meanwhile,
Rusunami had an intermediate relationship with condominiums and
Rusunawa, with approximately half of them using ACs, suggesting that the use of ACs increases with an increase in income level.
3.3.2. Fan Usage Patterns
Fan usage patterns were classified into four categories, as shown in
Figure 5. Similar to AC usage patterns, the majority did not use fans (B4, 69.7%). However, among the fan users, “Daytime” users (B3, 18.9%) were found to be the most common pattern.
Figure 6 shows a breakdown of fan usage patterns by building type. For all building types, the proportion of non-fan users (B4) was more than 60%. In other words, fans were not the mainstream method of cooling for collective housing in Indonesia. Meanwhile, it can be seen that the proportion of fan users increased as the income level decreased.
3.3.3. Window Opening Patterns
Window opening patterns were classified into four types, as shown in
Figure 7. Among the groups that opened windows, “Bed & living room” (C1, 19.1%) and “Living room” (C3, 17.2%) were most common, indicating that in collective houses, windows were mostly opened in living rooms during the daytime. However, households that did not practice window opening were found to be most common of all patterns (C4, 54.5%).
Figure 8 shows the breakdown of window opening patterns according to the building type. In condominiums and
Rusunami, C4, which did not use windows, accounted for 70%. In contrast, in
Rusunawa, 60% of the households practiced window opening.
3.3.4. Comprehensive Cooling Patterns
Figure 9 shows the classification of the comprehensive cooling patterns. This showed that the most common pattern was D8, which did not use any form of cooling method. However, among the groups that used either ACs, fans, or window opening, “Bed & living room window” (D5, 15.4%), “Bed room AC” (D2, 14.5%) and “Living room window” users (D7, 14.5%) were found to be common.
Furthermore, households were mostly found to use only one method of cooling with the exception of D3. In other words, the complex use of multiple cooling methods was rarely seen in collective houses.
Figure 10 shows the breakdown of cooling patterns according to the building type. In condominiums, ACs were the main method of cooling, and 62.2% of condominiums exhibited patterns of using ACs. On the other hand, in
Rusunawa, window opening was the most common method of cooling. Furthermore, in
Rusunawa and
Rusunami, households that did not use any cooling (D8) accounted for approximately 30% of the total.
3.4. Cooling Patterns and Energy Consumption
Figure 11 shows the annual energy consumption for each cooling method and pattern. Regarding the use of ACs, there was a significant difference in energy consumption between AC users (A1, A2, and A3) and non-AC users (A4). In particular, A1 and A3, which used ACs in the bedroom, had higher energy consumption rates. This suggested that there may have been factors that increased energy consumption in the use of ACs in bedrooms, such as the number of rooms in which ACs were installed.
In contrast to AC use, fan users (B1 and B3) and window openers (C1 and C2) did not show significant differences from non-users (B4 and C4). However, nighttime fan users (B2) and living room window openers (C3) showed significantly lower energy consumption rates than the other patterns. B2 was the only pattern with nighttime cooling other than AC users, and C3 had a shorter window opening practice compared to C1 and C2. It is possible that both B2 and C3 spent less time at home than the other groups, causing B2 and C3 to have lower energy consumption rates.
Figure 12 shows the energy consumption for each comprehensive cooling pattern and building type. When the energy consumption is compared by building type, consumption was the highest in condominiums, followed by
Rusunami and then
Rusunawa within the same cooling pattern, such as D2, D5, and D8. Therefore, it can be considered that there were factors contributing to the increase in energy consumption other than the use of ACs corresponding with the rise in income.
Furthermore, AC users had significantly higher energy consumption rates. Especially, groups that used ACs in bedrooms (D2) had the highest energy consumption rate of 19.0 [GJ/year · hh]. Meanwhile, fan users (D3 and D4) and window openers (D5, D6, and D7) had lower energy consumption rates from 8.7 [GJ/year · hh] (D4) to 12.3 [GJ/year · hh] (D3), due to the absence of AC use. Therefore, it was proved that the use of fans and window opening reduce household energy consumption.
Regarding window openers, a gradual increase in energy consumption was observed in the order of D7, D6, and D5. However, the window opening time was the longest in D5 and shortest in D7. This suggested that groups with longer window opening times spent more time at home, causing a slight increase in energy consumption.
Finally, D4 and D8 were found to have the lowest energy consumption rates among all the building types. Regarding D4, it was the only pattern that practiced cooling during the night other than AC users; it was possible for this group to spend less time at home compared to other groups, causing D4 to have a lower energy consumption rate.
3.5. Cooling Patterns and Thermal Comfort Level
Figure 13 and
Figure 14 show the distribution of thermal comfort and satisfaction assessment, respectively. In general, more than 80% of the respondents felt comfortable with their thermal environment. Meanwhile, there were some tendencies observed within cooling patterns and thermal comfort/satisfaction. For example, comfort and satisfaction tended to be higher at night than during the day. Furthermore, D4, D7, and D8 had a larger proportion of respondents who were dissatisfied. In particular, during the day, 61.5% of D4; 75.5% of D7; and 71.1% of D8 respondents felt satisfied and during the night and 77.6% of D4; 83.4% of D7; and 79.2% of D8 respondents felt satisfied, whereas in other cooling patterns, more than 80% during the day and more than 90% during the night felt satisfied with their thermal environments. Therefore, satisfaction could be improved by introducing appropriate air conditioning habits for D4, D7, and D8. In addition, there was a tendency for thermal comfort and satisfaction to be high in the order of condominiums,
Rusunami, and
Rusunawa, regardless of cooling patterns, indicating that higher-income families tended to be more satisfied with their thermal environment.
In terms of condominiums, no significant differences in thermal comfort and sensation were observed between cooling patterns. More than 90% responded that they felt comfortable and satisfied both day and night, even in D1 and D2, which did not conduct daytime cooling. Therefore, it can be considered that the indoor thermal environment in condominiums was maintained at a high level, regardless of the cooling methods.
For Rusunami, thermal comfort was at a high standard as well. More than 90% felt comfortable, except for D4 and D8, which had 15.5% and 19.9% of respondents feeling discomfort, respectively, during the daytime. Meanwhile, non-AC users, such as in D5 and D7, showed an equally high level of satisfaction as AC users. Less than 10% of D1, D2, D5, and D7 respondents were dissatisfied both day and night, indicating that thermal satisfaction can be achieved without the use of ACs.
Although, in other building types, more than 90% of the AC users felt comfortable and satisfied. In Rusunawa, AC users showed low satisfaction levels during the daytime, with 19.4% feeing dissatisfied. It was likely that cooling patterns with no daytime cooling in Rusunawa resulted in poor daytime satisfaction. Specifically, 19.4% of D2; 37.6% of D4; and 23.2% of D8 respondents were more than slightly dissatisfied during the day. In contrast, the patterns for Rusunawa that had more than 80% of respondents satisfied for both day and night were Groups D3, D5, and D6, with daytime ventilation.
5. Conclusions
This study clarifies the impact of the use of ACs and NV on energy consumption and thermal comfort assessments by identifying the cooling methods of existing spaces in collective houses in Indonesia. The findings were as follows:
Regarding energy consumption, the results indicated that energy consumption increased along with an increase in household income (
Figure 2). Furthermore, energy consumption by cooling methods indicated that the use of ACs had a significant impact on energy consumption (
Figure 11). The relationship between energy consumption and comprehensive cooling patterns indicated that the implementation of NV was an effective measure to reduce energy consumption regardless of income level.
Regarding thermal comfort, the results showed that the majority of residents felt comfortable and satisfied with their thermal environment at home. However, as a tendency, daytime satisfaction showed to be lower than nighttime satisfaction, particularly in
Rusunawa. Cooling patterns with no daytime cooling, such as D2, D4, and D8 in
Rusunawa, showed poor satisfaction levels during the day (
Figure 14). On the other hand, daytime window opening patterns, such as D3, D5, and D6, showed higher satisfaction. Therefore, practicing daytime ventilation is likely to improve the thermal comfort of D2, D4, and D8.
Through a comparison between the study by Mori et al. [
28] and this study, socio-demographic changes were observed in low-income households since 2016. The changes in household attributes have led cooling habits to change as well. In particular, a decline in window opening habits was observed. Furthermore, the relationship between cooling habits in
Rusunawa,
Rusunami, and condominiums (
Figure 10) suggested that a rise in household income will lead to a further decline in window opening habits and increase in the use of ACs.
The findings above indicate that the implementation of NV has the ability to reduce energy consumption by the use of ACs and improve the thermal satisfaction of residents. On the other hand, the results suggested that NV habits will fade as living standards improve. In order to reduce the future reliance on ACs and promote NV, housing designers are recommended to remove potential factors that would inhibit residents from opening their windows. In addition, collective housing should be provided with features that enable effective NV. Furthermore, residents should recognize the positive effect of NV on thermal comfort and indoor air quality. Meanwhile, policy makers should support the above recommendation through policies that would motivate housing designers to adopt designs that promote NV.
It is important to note that the findings from this study were limited to collective housing in tropical regions of Southeast Asia and do not pertain to landed houses. Moreover, due to the nature of the survey, the statistical values pertaining to personal attributes of the respondents used in this study were a compilation of information about the household’s financial managers. This may have caused biases in the findings derived from data regarding personal attributes.
The findings of this study verified the energy saving effects of NV habits for the purpose of decarbonizing collective housing in tropical regions. In addition, they contribute to promote cooling strategies that achieve both the improvement of thermal satisfaction and energy conservation with the consideration of the rapidly changing socio-demographic characteristics of collective housing residents in a developing nation. Finally, as a prospect, the findings of this study will be implicated in future studies that will contribute to optimize operational energy consumption in collective housing in tropical regions.