Proposed Model of Neutral Temperature Equation for Adaptive Thermal Comfort in Student Flats Units in the Tropics

Thermal comfort of residential rental flats is one of the problems from the ventilation aspect considering that the planning of flats is based on natural ventilation. This article is intended to formulate an alternative to the neutral temperature equation from the equation proposed by Humphrey, Auliciem, de Dear, Nicol, and Roaf with day and night measurement times. Measurements were carried out for three weeks in a student rental apartment building, Hasanuddin University Faculty of Engineering in Bontomarannu, Gowa Regency, which has 5 floors with 104 rooms. Measurements were made with 2,577 respondents during the day, and 2,043 at night. During the day, with an operating temperature between 29.2°C to 31.0°C, the neutral temperature (Tn) obtained from the PMV model is 27.6°C (T o ). This figure is lower than the actual choice of the TSV scale and the TCV scale, which were 30.3°C (T o ) and 29.7°C (T o ), respectively. At night, with an operating temperature between 27.7°C to 29.5°C, the neutral temperature (T n ) obtained from the PMV model is 26.2°C (T o ). Meanwhile, the neutral temperature obtained from the respondents' actual choices (TSV and TCV) were 28.8°C (T o ) and 28.2°C (T o ), respectively. The neutral temperature equations presented in the Humphrey, Auliciem, de Dear, Nicol, and Roaf equations show a lower thermal coefficient than the corrected thermal coefficient and the elaboration of the equations carried out in this study, both from daytime measurements and from night measurements.


Introduction
Thermal comfort is defined as a "state of mind expressing satisfaction with the thermal environment" as described in ASHRAE [1]. Thermal comfort plays a very strategic role in building design, because most or about 80% of the time people's activities are done indoors [2]. From the aspect of energy use, Perez-Lombard stated that 70% of energy use in buildings is allocated for air conditioning and artificial lighting. Therefore, buildings need to be designed to respond to climatic conditions while maintaining the comfort of the occupants [3].
The level of productivity and physical endurance (health) of humans is in a certain zone, while outside this zone human productivity and physical endurance decrease. This zone is related to comfort, where according to reports from various studies it is different between groups of people who live in different climatic locations. The British have a comfort zone of 15°C to 20°C while the humans who live in the United States achieve comfort levels between 19.5°C and 26.7°C, while those who live in the tropics achieve a comfortable level in the temperature. 23.4°C to 29.4°C, all at humidity conditions between 30 to 70% [4].
Thermal comfort research in the tropics is important given that there is a significant increase in the mean annual temperature. Harimi [5] explained that when the climate data for the Kota Kinabalu area were analyzed, it was found that the trend of maximum outdoor temperature increase per decade was around 0.4°C with an average outdoor temperature increase of 0.3°C. The maximum outdoor temperature appears to be increasing faster than the average value.
Human thermal comfort sensation depends partly on his activity and clothing and partly on four environmental variables: air temperature, average emission temperature, air velocity, and humidity. Fanger [6] introduces a comfort equation forming all the combinations of the six factors mentioned above that will provide thermal comfort to humans. The convenience equation does not include time of day as a variable, but assumes that the comfort criterion is constant throughout the day. Fanger [7] in the study of thermal comfort conditions during simulated normal working days at the office and found no significant difference in the preferred ambient temperature during the 'morning and evening hours. Likewise with research conducted in 1974 by Fanger [8] who carried out investigations into thermal comfort in the morning (09.00-12.00) and in the evening (19.00-22.00) which did not found significant differences in the preferred ambient temperature at these two times.
Various studies on thermal comfort in the tropics have been carried out, including in Indonesia [9,10], Malaysia [11], and Singapore [12]. Most of the thermal comfort studies in the tropics are carried out with a heat balance model using the Predicted Mean Vote (PMV). In 1970 Fanger developed the PMV model which was later adopted by ASHRAE as the ASHRAE 55 standard [13]. The PMV model is calculated using six variables, namely air temperature, relative humidity, mean radian temperature (MRT), air velocity, clothing index, and the metabolic rate of the respondent's activity. A number of studies of thermal comfort in the tropics mentioned above found the neutral temperature predicted by PMV to be significantly lower than that of the respondents using the respondent's choice of thermal sensation (TSV). Several studies conducted by Hamzah [10,14] found that the students' neutral temperature was at 23.0°C (PMV) and 29.0°C (TSV). Karyono also found this difference in numbers [15] which revealed that the ISO-7730 and ASHRAE 55-56 standards were about 1-3°C lower than the comfortable temperature required by respondents in tropical regions such as Indonesia. Such conditions will have implications for wasteful energy use and complaints of discomfort due to low room temperature.
The general condition of rental apartment units built in Indonesia, including Makassar City, is a residential building that is mostly subsidized or assisted with technical requirements that direct the use of a building ventilation system with natural ventilation [16]. With thermal conditions tinged with temperatures varying from 28°C in the morning to 33°C in the afternoon [14]. This temperature range is outside the comfort zone by Indonesian local standards [17] and ASHRAE standards [1]. However, a number of studies of thermal comfort in Makassar City found that despite the hot temperature, more than 80% of respondents chose a "comfortable" area and more than 80% of students received a "hot" thermal environment. Adaptation activities that are usually carried out by respondents in order to adjust their thermal comfort needs with the temperature conditions in the room, including the use of fans, changes in activities, clothes, opening and closing windows, changing fans and so on. Roaf et al. [18] suggested that there are three categories of thermal comfort adaptation activities, namely physiological, behavioral, and psychological. Of the three adaptation activities, behavioral adaptation is the most common way for people to adjust their thermal comfort.
Another thermal comfort concept is the adaptive thermal comfort model developed by Humphrey [19] based on thermal comfort data collected from 1935 to 1975. This concept is specially formulated to predict thermal comfort in naturally ventilated buildings located in tropical areas with hot thermal environments. The concept is put forward in the form of linear regression between neutral temperature (Tn) and average outside temperature (Tout) as follows: The equation is then revised to improve the accuracy of the equation as follows [20]: T n = 13,2 + 0,534T out (2) This equation was further refined in 1988 by formulating the following equation [20]: T n = 13.5 + 0.546T out Auliciems [21] develop a neutral temperature formula by stating the following equation: Auliciems dan de Dear [22] propose another equation with the following formula: Nicol dan Roaf [23] suggests a different equation by proposing the following formula: A number of equations mentioned above are based on research on thermal comfort conducted in Indonesia [14], showing a neutral temperature below the neutral temperature figure based on actual measurements with the TSV model, both for general building functions, and for residential buildings with effective day and night activities. Therefore, this study aims to analyze the neutral temperature equation above based on tropical climatic conditions characterized by relatively high temperature and humidity.

Research Method
Measurements were made in student apartment units which are five-story buildings. Measurements were carried out every day in twelve rooms spread over five floors simultaneously. This is intended to obtain thermal conditions which represent the general thermal conditions of buildings. The downside of this measurement is that it doesn't specifically differentiate between thermal conditions on different floors.
The proposed neutral temperature equation is obtained by changing the existing equation constants according to the actual value of the internal temperature measurement results against the thermal conditions at the research location, both during the day and at night.

Data Collection
The data used as the object of study consists of four data components including: outdoor temperature data for the period 2014-2019, outdoor temperature data for the period of October 2020, objective data measurement results, and subjective data measurement results. Data on temperature and relative humidity for outdoor space on a 6-year scale for the period 2014-2019 obtained from the weather monitoring tool Vaisala Type WXT520 which is equipped with data loggers Type CR800 series 2 which automatically stores data for a 24 hour period, RTU (Remote Terminal Unit), which is part of the AWS (Automatic Weather Stations) device on the campus data station of the Faculty of Engineering, Unhas Gowa.

Outdoor Temperature Data for the Period of October 2020
Temperature and relative humidity data for the out space for the period of October 2020 were obtained from the Vaisala Type WXT520 weather monitor. This data is obtained from the same instrument to measure 6-year scale data for the 2014-2019 period.

Objective Data Measurement Results
The results of objective data measurements are carried out to collect thermal environmental conditions and clothing data of respondents. The main instrument used in this study was the Thermal Comfort Multi Logger (LSI TC) from LSI-Lastem, Italy. LSI TC consists of three sensors that collect and record in one data logger. Tiga sensor termasuk psikometrik portabel probe (BSU102), which is used to record temperature and humidity relative humidity data. The second sensor is a Globe Thermometric Probe (BST131) which is used to record the temperature of the globe. The third sensor is a wet-bulb temperature probe (BSU121) for recording the wet-bulb temperature. Apart from LSI TC, another instrument used is Hobo Loggers which records measurements of the thermal environment in space. This tool was developed and manufactured by Onset in the United States. Two types of loggers Hobo are used in this study, namely the standard Hobo which records air temperature and relative humidity, and the Hobo logger with an external sensor that records air velocity in the room. This instrument (LSI TC), 2 hobo loggers with external sensors, and 3 standard hobo units are spread out in 6 rooms simultaneously on the same floor level with a continuous data recording period of 24 hours with the number of days that have been arranged. The sensor is installed at a height of approximately 1 meter above the floor level [12]. Respondent clothing data is an integral part of the questionnaire.

Subjective Data Measurement
Subjective measurements were carried out to record respondents' responses using a questionnaire instrument developed and modified based on the questionnaire used by Wong and Khoo [12]. The questionnaire used contains seven questions. Four questions are related to the thermal aspect, namely thermal sensation, thermal comfort, thermal preference and acceptance, the fifth and sixth questions concern air velocity, and the seventh question concerns indoor humidity. Questions about TSV cover the seven ASHRAE scales consisting of cold (-3), cool (-2), slightly cool (-1), neutral (0), slightly warm (+1), warm (+2), and hot (+3). Questions about the TCV covering the seven Bedford scale consist of: much too cool (-3), too cool (-2), comfortable cool (-1), comfortable (0), comfortable warm (+1), too warm (+2), and much too warm (+3). This subjective measurement was carried out by taking two times scales, namely during the day between 10:00-15:40 and evening between 18:00-21:40 hours. This is done considering the function of the building as a function of housing.
Annual temperature conditions for the period 2014-2019 show the temperature conditions measured by day and night where there is a significant difference between the average daytime and night temperatures of 2.0°C. The maximum daytime temperature was recorded at 29.5°C in 2016 and the minimum temperature was 28.8°C in 2019. The maximum nighttime temperature was recorded at 30.5°C in 2017 and the minimum temperature was 24.9°C in 2015. Average air temperature during the day recorded in the 2014-2019 measurement period it was higher than the average air temperature measured in the 1968-2003 period [5].

Data Processing and Analysis
Data processing and analysis were carried out using spreadsheet software and statistical packages which were used to calculate the maximum value, minimum value, and average value as the basis for describing the PMV, TSV, and TCV charts. The PMV of each respondent was calculated using data from four environmental variables and two personal variables. Four environmental variables include air temperature, average radiant temperature, relative air velocity, and relative humidity. Two personal variables include data on clothing and metabolic rate. The PMV value of respondents was determined using the PMV calculator developed by Farina [24] based on ASHRAE standard 55 [1]. Linear regression of PMV, TSV, and TCV was performed using SPSS statistical analysis (statistical package for social science) to show the distribution of respondents' actual voices to operating temperature (To). Data validity and reliability tests are carried out prior to data processing in order to ensure data normality.

Research Samples
The survey and objective data measurement were carried out for three weeks (21 days) in the Rusunawa building, Bontomarannu Engineering Faculty, Gowa Regency. This building is a 5-storey building with 104 rooms divided into 16 rooms on the ground floor and 22 rooms on the 2,3,4, and 5 floors. Measurements are carried out 24 hours a day to obtain temperature data that represents daytime and night considering the function of the building as a residential unit. The characteristics of the day and night measurement samples are described in Table 1 and Table 2 below.

Outdoor and Indoor Microclimate Conditions
Outdoor microclimate condition data is data obtained from the weather monitoring tool Vaisala which is located at the top of the campus building of the Bontomarannu Faculty of Engineering, Gowa Regency. Average outdoor temperature the room recorded in October during the day is 29.6°C with a maximum temperature of 30.9°C and a minimum temperature of 27.3°C. While the average temperature at night is 26.4°C, the maximum temperature is 28.0°C and the minimum temperature is 22.7°C.
Data on indoor microclimate conditions from day and night measurements that are recorded during the measurement can be seen in Table 3. This table shows that the average indoor air temperature during the day is 31.6°C with the maximum and minimum temperatures respectively 32.3°C and 30.6°C. Air temperature conditions indicate relatively high temperature conditions in the room. The average temperature recorded during the day and night is higher than the local Indonesian thermal comfort standards [17] and ASHRAE international standards [1]. The average relative humidity during the day and night was 71.78% and 74.05%, respectively. As with temperature, the relative humidity conditions are well above Indonesia's thermal comfort standards [17].
The data presented in Figure 1 contain the percentage of respondents 'choice of thermal sensation, thermal comfort and prediction of the average choice of respondents during the day, which can be explained that the average prediction of respondents' choices is relatively different from the TSV scale and the TCV scale. In the prediction model (PMV), there are 75.28% of respondents predicted to choose hot areas (+2 and +3), 24.72% of respondents predicted to choose the middle area (-1, 0, and +1). If you compare the predictions of respondents who choose in the middle area (-1, 0, and +1) with the respondents' choices on the TSV and TCV scales figures 77.22% and 85.06%, It can be explained that the predicted figures in the PMV model against the options in the middle region (-1, 0, and +1) are much lower than on the TSV and TCV scales. Likewise, the predictions of respondents who chose a slightly cool area (-1) on the TSV scale were 27.01% and a comfortable cool area (-1) on the TCV scale was 12.03% compared to 0% in the PMV model, and respondents who chose a slightly warm area (+1) on the TSV scale as much as 19.17% and 32.48% of respondents chose a warm comfortable area (+1) on the TCV scale which is also compared to 24.72% in the PMV model.   0  7  36  30  15  3  6  0  6  6  34  40  8  3   9 October 2020  7  28  43  33  16  3  3  0  3  23  62  33  11  2   10 October 2020  6  25  42  33  15  4  1  0  2  26  63  26  11  0   11 October 2020  0  17  46  42  18  8  3  0  3  31  56  34  8  2   12 October 2020  0  14  37  42  31  8  2  0  7  26  54  38  6  2   13 October 2020  0  6  28  45  34  13  8  0  7  5  58  52  7  4   14 October 2020  0  7  29  43  32  17  6  0  8  5  60  52  7  2   15 October 2020  0  9  41  45  20  8  4  0  7  5  55  50  8  3   16 October 2020  0  8  35  50  23  8  3  0  2  5  50  57  9  3   17 October 2020  0  12  32  37  18  6  2  0  10  22  41  25  7  2   18 October 2020  0  8  27  40  32  18  8  0  7  5  53  60  9  1   19 October 2020  5  22  33  25  12  3  0  8  22  42  22  6  0  0   20 October 2020  17  35  37  23  8  2  2  0  9 Figure 2 contains data on the percentage of respondents' choice of thermal sensation, thermal comfort and the predicted average choice of respondents at night. This data shows that in the PMV model, there are 75.38% of respondents predicted to choose hot areas (+2 and +3), 24.62% of respondents predicted to choose the middle area (-1, 0, and +1). When comparing the predictions of respondents who chose in the middle area (-1, 0, and +1) with the choice of respondents on the TSV and TCV scales at 77.00% and 84.87%, it can be explained that the numbers predicted in the PMV model to the choices in the middle area ( -1, 0, and +1) are much lower than the TSV and TCV scales. Likewise, the predictions of respondents who chose a slightly cool area (-1) on the TSV scale were 26.97% and a comfortable cool area (-1) on the TCV scale was 11.89% versus 0% on the PMV model, and respondents who chose a slightly warm area (+1) on the TSV scale were 19.24% and 32.35% of respondents chose a warm comfortable area (+1) on the TCV scale which was also compared to 24.62% in the PMV model. Based on the data elaborated in Figure 2 and Figure 3, it can then be explained that the results of research on thermal comfort were carried out in several regions such as Indonesia [9,10], Malaysia [11], and Singapore [12] find relatively the same phenomenon. Therefore, the PMV model can be said to be an irrelevant model to be applied in thermal comfort research, especially in naturally ventilated buildings in tropical areas such as Indonesia.   Figures 3,4, and 5 show the scatter plots and linear regression of the operating temperature (To) against the predicted PMV, TSV, and TCV during the day. With operating temperatures ranging from 29.2°C to 31.0°C, the neutral temperature (Tn) obtained from the PMV model is 27.6°C (To). This predicted neutral temperature is lower than the actual choice of the TSV scale and the TCV scale which are 30.3°C (To) and 29.7°C (To), respectively. Figures 6, 7, and 8 are scatter plots and linear regression of operating temperature (To) against predicted PMV, TSV, and TCV at night. With operating temperatures ranging from 27.7°C to 29.5°C, the neutral temperature (T n ) obtained from the PMV model is 26.2°C (T o ). As with the linear regression results during the day, the predicted neutral temperature is also lower than the actual choice of the TSV scale and the TCV scale which are 28.8°C (T o ) and 28.2°C (T o ), respectively.
The neutral temperature (T n ) obtained from the relationship between the actual choice (TSV and TCV) and the operational temperature (T o ) between day and night shows a significantly different number. On the TSV scale there is a neutral temperature difference between day and night of 1.5°C. On the TCV scale there is a neutral temperature difference of 3.5°C. Meanwhile, there is a neutral temperature difference predicted on the PMV scale between day and night of 1.4°C.

Respondent/Occupant Adaptive Activities
Adaptive activities carried out by respondents / residents in responding to thermal conditions in the room during the day or at night are by setting windows to a wide open position, opening doors, using paper fans, using battery-powered mechanical fans, and some rooms using electric fans. The use of air conditioning in the room was not found because the flats management policy did not allow the use of air conditioning due to considerations of limited electrical resources on campus. Other adaptive activities carried out by residents are leaving the room and joining a study room where the occupants are relatively better because of the sufficient number of window openings.

Elaboration of Equations on Neutral Temperature Measurement Results
The neutral temperature equations which are the object of study in this study are the six neutral temperature equations proposed by Humphrey, Auliciem, Auliciem and de Dear, Nicol and Roaf, respectively, as described in equation (1-6) above. With an actual neutral temperature of 30.3°C (TSV), an average outdoor temperature of 29.6 °C, an average indoor temperature of 31.6 °C, then the average neutral temperature of the 6 equations with respect to daylight measurement results the day is as follows: The data presented in table 6 above shows that the constant value of equation (3) from the daytime measurement results is the value closest to the thermal comfort figure based on the TSV scale measurement (30.3 o C). Likewise, the results of night measurements where equation (3) is the closest value to the TSV scale neutral temperature (28.8 o C). Furthermore, the proposed neutral temperature equation from daytime calculations is shown in Figure 9, while the proposed neutral temperature equation from nighttime calculations is shown in Figure  10.

Conclusions
Based on the analysis of the research, it can be concluded that the PMV model used to measure the level of thermal comfort in the tropics, especially in the research location, cannot produce real thermal comfort figures which are marked by a significant difference between the PMV measurement results and the respondents' actual choices (TSV and TCV). There is a difference significant between the neutral temperature predicted by PMV both in the daytime measurement and the nighttime measurement.
In the daytime measurement, with operating temperatures ranging from 29.2°C to 31.0°C, the neutral temperature (T n ) obtained from the PMV model is 27.6°C (T o ). This predicted neutral temperature is lower than the actual choice of the TSV scale and the TCV scale which are 30.3°C (T o ) and 29.7°C (T o ), respectively. At night measurements, with an operating temperature ranging from 27.7°C to 29.5°C, the neutral temperature (Tn) obtained from the PMV model is 26.2°C (To). As with the daytime measurement, the neutral temperature obtained from the respondents' actual choice (TSV and TCV) was 28.8°C (T o ) and 28.2°C (T o ), respectively. The neutral temperature equations presented in the Humphrey, Auliciem, de Dear, Nicol, and Roaf equations (equations 1-6) have a thermal constant that is relatively lower than the thermal constant of the actual measurement results carried out in this study, both from the results of daytime measurements, as well as from the results of night measurements. Therefore, this study proposes an alternative neutral equation suitable for tropical areas.