Available online at www.sciencedirect.com ScienceDirect Procedia Engineering 169 (2016) 158 – 165 4th International Conference on Countermeasures to Urban Heat Island (UHI) 2016 Indoor Thermal Comfort Assessment of Industrial Buildings in Singapore N.H Wonga, E Tana, O Gabrielaa,*, S.K Jusufb a Department of Building, School of Design and Environment, National University of Singapore, Architecture Drive, Singapore 117566, Singapore b Engineering Cluster Sustainable Infrastructural Engineering, Singapore Institute of Technology, 10 Dover Drive, Singapore 138683, Singapore Abstract In a tropical climate such as Singapore, as compared to fully air-conditioned buildings, naturally ventilated (NV) buildings tend to have lower Thermal Comfort (TC) condition This research investigates and develops a thermal comfort assessment method for naturally ventilated industrial buildings Data was collected through field survey which consisted of environmental measurement and questionnaire of the thermal perception of several industrial buildings’ occupants From statistical analysis of the field survey data, thermal comfort prediction model was developed The acceptable range to achieve thermal comfort for industrial buildings was also analyzed © 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license © 2016 The Authors Published by Elsevier Ltd (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-reviewunder under responsibility oforganizing the organizing committee of IC2UHI2016 the 4th IC2UHI2016 Peer-review responsibility of the committee of the 4th Keywords: Thermal Comfort; Thermal Sensation; Naturally Ventilated; Industrial Building Introduction Building and Construction Authority (BCA) sets the Green Mark (GM) scheme standard in 2005 Throughout the years, the criteria used for Residential Buildings (RB) have been adopted for Non-Residential Buildings (NRB), and the passing criteria of area weighted wind velocity of 0.6 m/s [1] for developments with higher GM rating is deemed to be not sufficient to assess the natural ventilation (NV) design for NRB * Corresponding author Tel.: +65-8657-6580; +62-811-799-882 E-mail address: bdgog@nus.edu.sg 1877-7058 © 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the organizing committee of the 4th IC2UHI2016 doi:10.1016/j.proeng.2016.10.019 N.H Wong et al / Procedia Engineering 169 (2016) 158 – 165 In Singapore, as compared to fully air-conditioned buildings, NV buildings tend to have better Indoor Air Quality (IAQ), but worse Thermal Comfort (TC) condition So far, there is no research study that specifically focus on industrial buildings, therefore, this study looks into the TC of industrial buildings This study will help BCA to further ensure sustained building performance, which key strategy would be to minimize energy consumption through the optimization of design The objectives of this study are as follows: To develop thermal comfort model, analysis and validation; and To recommend the optimum requirement and assessment method to achieve thermal comfort in industrial buildings Methodology 2.1 Objective and subjective measurement Data was collected through field survey, and used to develop and validate the TC model Field survey allows “firsthand” data that help to capture occupants’ TC perception in their actual daily environment The in-situ environmental measurement also records the characteristics of NV buildings, which are dynamic and unpredictable These NV characteristics such as intermittent wind, solar radiation and high humidity cannot be simulated easily by mechanical means in a chamber, thus field survey is considered as the best method for the data collection The measurement protocol for the field survey followed Class II protocol of thermal comfort field research [2] There are two groups of data required, i.e objective and subjective measurement data Since the human perception is not as simple as “stimulus-response” (cause-effect) phenomenon, the field survey attempted to observe and collect data to comprehend better the complex human perception, behavior and background The objective measurement measured the air temperature, wind speed, relative humidity and globe temperature near each occupant (respondent), and noted the activity and the clothing level of the respondent while the respondent did the subjective assessment The objective measurement was conducted at around 0.8 – 1m high from the floor The measurement was conducted using handheld equipment of Testo 445 as shown in Fig 1(a) During the indoor field survey, a weather station was installed on the roof of the building to measure the micro-meteorological condition Fig 1(b) shows the HOBO weather station The monitored environmental parameters included ambient temperature, wind speed and wind direction, relative humidity, and solar radiation The subjective assessment was formulated into questionnaire form Some standard questionnaires (response scales) for TC studies such as ASHRAE scales for thermal sensation vote and Bedford scales for thermal comfort vote [3, 4] are used By using both scales, the consistency of response between thermal sensation (hot to cold) and perception (uncomfortable-comfortable warm or cold) can be further verified 2.2 Thermal comfort survey For each building, the survey was conducted in two sessions daily, i.e in between 10AM to 12PM (morning session) and in between to 4PM (afternoon session), in order to capture the thermal perception of different parts of the day The data collected from the field survey was compared between external and internal environment The surveys were on three different types of industrial buildings in Singapore, and were conducted on the occupant/worker inside the buildings Fig shows the floor plans of the buildings surveyed Fig 3(a) shows the weather station installation on the rooftop, while Fig 3(b) and Fig 3(c) show the TC survey conducted within the units 159 160 N.H Wong et al / Procedia Engineering 169 (2016) 158 – 165 a b Fig (a) Test 445; (b) HOBO weather station a b c Fig (a) Building A; (b) Building B; (c) Building C a b c Fig (a) HOBO weather station on the rooftop; (b) and (c) Field measurement in progress Result and Discussion 3.1 Field measurement data The TC survey at building A, which is classified as standard factory, had perception of 35 and 20 respondents collected during morning and afternoon sessions respectively Total number of respondents was 55 Building B and C are classified as flatted factory These two buildings are located next to each other Building B is a 4-story building with ramp-up to the units and it has a corridor at its perimeter In each story, there were 40-60 respondents participated in the survey Total number of respondents was 188 Building C is a 9-story building which units have doublecorridors Total number of respondents was 145 In overall, there were 388 responses collected, of which 204 and 186 respondents were from morning and afternoon session respectively The survey in Building A was on a very cloudy day when the solar radiation was less than 400 W/m2 and the air temperature was less than 30°C The survey in Building B was during the hottest weather condition among all sites when the solar radiation was reaching 800 W/m2 and the air temperature ranged between 30 and 34°C The first day N.H Wong et al / Procedia Engineering 169 (2016) 158 – 165 of survey in Building C was partially cloudy, while the second day was clearer than the first day Fig shows the measured outdoor weather condition during the time of the TC survey Fig Outdoor weather condition (10AM – PM daily) Fig shows the measured indoor data and the summary of air temperature (°C), mean radiant temperature (°C) and wind speed (m/s) during the collection of survey responses respectively Generally, the trend of the measured indoor air temperature follows the trend of the outdoor air temperature Indoor air temperature increases when outdoor air temperature increases The indoor air temperature in Building C was generally higher than Building B although the outdoor air temperature was lower during the survey in Building C This may be due to the heat build-up in Building C because of its double-corridor design The mean radiant temperature did not differ much from the air temperature due to less high-heat-generating equipment used in the facility Most of the wind speed were in between range of to 1.1 m/s although the maximum measured one reached up to 5.4 m/s Fig Respondents’ environment measurement (10AM – PM daily) 161 162 N.H Wong et al / Procedia Engineering 169 (2016) 158 – 165 Fig shows the thermal acceptability of both Thermal Comfort Vote (TCV) and Thermal Sensation Vote (TSV) The acceptance level of TSV is always lower than that of TCV TCV results show that the respondents have higher tolerance towards hot sensation This shows that in NV spaces, people have higher comfort tolerance towards warm sensation Comparing all sites, Building A shows the highest percentage of thermal acceptability, while Building B shows the lowest percentage This may be due to the outdoor weather condition which was cloudy during the survey in Building A, and clear hot day condition during the survey in Building B The weighted overall percentages of thermal acceptability for TSV and TCV show that only 52% and 57% are acceptable respectively Fig Thermal acceptability 3.2 Thermal comfort model development The TC model development used analytical method, which relied on the statistical method to analyze primary data and to model mathematically the thermal comfort prediction TC model development usually looks into the relationship of TSV with the environmental variables, but in this study, analysis of relationship between TCV with the environmental variables was discussed instead The main reason is because people tend to have higher acceptability of TCV in NV spaces (as shown in Fig 6) Data from the thermal comfort survey was grouped into parts One part of data was used to develop the model, while a small part of the data came from random selection from each site was used to validate the model Weighted estimation was made to determine the number of respondents selected from each site and each session for validation purpose Computation of the statistical analysis used SPSS® Version 23 software Multiple Regression Analysis is the most common statistical tools used by many TC researchers to develop TC indices [5, 6] 3.3 Model development and validation The statistical analysis analyzed the TCV as dependent variables, while measured air temperature (DBT), relative humidity (RH) and wind speed (WIND), calculated mean radiant temperature (MRT) from measured globe temperature, and calculated clothing level (CLO) as the independent variables The direct correlations between the dependent and the independent variables are depicted in Table Strong correlations between variables (Pearson correlation > 0.7) show that these variables should not be grouped in one mathematical model It is apparent that DBT and RH is strongly correlated and only one of them can therefore be taken in the final equation This table also shows the significance level of the correlation for each independent variable 163 N.H Wong et al / Procedia Engineering 169 (2016) 158 – 165 with the dependent variable Table shows that each RH, CLO and WIND correlation with TCV are not significant (Sig value (two-tailed) or p > 0.05) Multiple regression was then conducted to develop the thermal comfort model Table Inter-correlation between independent variables and TCV TCV TCV DBT RH MRT CLO WIND Pearson Correlation Sig (2-tailed) DBT RH MRT CLO WIND 183 -.072 162 -.109 -.060 001 214 005 059 298 -.736 247 009 160 000 000 874 005 -.347 -.109 -.148 000 060 010 -.347 012 035 840 549 012 130 Pearson Correlation 183 Sig (2-tailed) 001 Pearson Correlation -.072 -.736 Sig (2-tailed) 214 000 Pearson Correlation 162 247 Sig (2-tailed) 005 000 000 Pearson Correlation -.109 009 -.109 Sig (2-tailed) 059 874 060 840 Pearson Correlation -.060 160 -.148 035 130 Sig (2-tailed) 298 005 010 549 024 024 During the statistical analysis process, it was found that DBT was the only significant independent variable For MRT, the insignificance was due to less high-heat-generating equipment used in the facility For CLO, it was due to the similar clothing level of the respondents For WIND, it was due to the large data that had small range of wind speed value Past studies found that in the tropics, uncomfortable hot and sticky conditions required higher speed of wind flow over the human body to increase efficiency of sweat evaporation [7, 8, 9, 10] Although in this study it was found that wind speed did not significantly affect the thermal comfort perception, wind speed was still considered in the thermal comfort model development Table shows the combinations of TCV with DBT and TCV with DBT and WIND Table Combinations of DBT and WIND Variables in equation Combinations DBT MRT WIND Constant R2 Adj R2 Residual Mean Sqr DBT 0.187 - - -4.603 0.033 0.030 1.439 DBT WIND 0.202 - -0.181 -4.974 0.042 0.035 1.432 The equation for the thermal comfort perception, which considers wind speed, can be written as equation (1) TCV = 0.202 DBT – 0.181 WIND – 4.974 (1) Fig shows the boxplot comparison of measured TCV data (Measured_TCV) with calculated TCV using Equation in Predicted_TCV The median value for both predicted and measured is similar, though the prediction model seems to over-predict towards warmer perception Equation is then considered as the Predicted Mean Vote (PMV) equation to predict the thermal comfort in industrial buildings 164 N.H Wong et al / Procedia Engineering 169 (2016) 158 – 165 Fig Comparison of measured TCV and predicted TCV for model validation 3.4 Boundary condition The thermal comfort prediction model as shown in Equation has two components, i.e DBT and WIND For WIND, indoor wind speed with or without fans shall be derived from Computation Fluid Dynamic (CFD) simulation results For DBT, Singapore’s Changi meteorological weather station air temperature (dry bulb temperature) of 32 years was analyzed Since industrial buildings generally operate at around 9AM to 6PM daily, the analysis was also conducted for this period There were two approaches being looked into The first approach was by daily averaging the DBT at 9AM to 6PM This data with one decimal point was grouped into DBT bins with 1°C interval to simplify the analysis The frequency of occurrence of each DBT bins was calculated The most frequently occurred DBT is 30°C The second approach was by grouping the hourly DBT from the weather station data directly into the DBT bins without daily averaging The frequency of occurrence of each DBT bins was calculated The most frequently occurred DBT is also 30°C From this analysis, the most frequently occurred DBT from both approaches is 30°C Therefore, the air temperature / DBT for calculating the thermal comfort equation is recommended at 30°C Based on the boundary condition of DBT 30°C, Table shows the range of WIND for PMV -1 to PMV -1 to is chosen to represent the acceptable thermal comfort range At wind speed of 0.6 m/s based on current GM criteria, PMV is calculated to be almost 1, i.e the highest acceptable range Table Range of WIND for PMV -1 to DBT (°C) WIND (m/s) PMV 30 11.5 -1 30 6.0 30 3.2 0.5 30 2.7 0.6 30 2.1 0.7 30 1.6 0.8 30 0.6 0.98 ~ N.H Wong et al / Procedia Engineering 169 (2016) 158 – 165 GM scheme has different levels of passing criteria PMV -1 to is proposed as the minimum passing criteria Since the wind speed required to achieve PMV 0.5 at 3.2 m/s might cause discomfort, PMV 0.8 with wind speed of 1.6 m/s is proposed for the higher level’s passing criteria Conclusion Thermal comfort assessment of industrial buildings shall be based on PMV equation as shown in Equation DBT is indoor air temperature (°C) Baseline of DBT is 30°C WIND is indoor wind speed (m/s) The value shall be derived from the result of indoor ventilation simulation The recommended PMV level as the minimum passing criteria is PMV -1 to PMV 0.8 is proposed for the higher level’s passing criteria Building industry consultants can propose innovative methods to lower the DBT without additional energy consumption Acknowledgements This paper is part of the research project “Development of Computation Fluid Dynamic (CFD) Simulation Methodology and Evaluation parameters, Thermal Comfort Model & Simulation Methodology for Wind Driven Rain in Natural Ventilated Building for Non-Residential Buildings (NRB) BCA Green Mark Criteria” funded by Building and Construction Authority (BCA) Research & Innovation Fund grant number 1.51.602.22153.00, and managed by Institute of High Performance Computing (IHPC) References [1] Building and Construction Authority, BCA Green Mark for New Non-Residential Buildings Version NRB/4.1, Singapore, 2013 [2] Brager, G.S., and De Dear, R.J., Thermal Adaptation in the Built Environment: A Literature Review, Energy and Buildings, 27 (1998), 83-96 [3] Chrenko, F.A., and Bedford, Bedford’s Basic Principles of Ventilation and Heating, HK Lewis and co, London, 1974 [4] McIntyre, D.A., Indoor Climate, Applied Science Publishers, London, 1980 [5] Webb, C.G., An Analysis of Some Observations of Thermal Comfort in an Equatorial Climate, British Journal of Industrial Medicine, 16 (1959), 297-310 [6] Sharma, M.R., and Ali, S., Tropical Summer Index – A Study of Thermal Comfort of Indian Subjects, Building and Environment, Pergamo, 21 (1986), 11-24 [7] Chandra, S., Fairey, P.W., and Houston, M.W., Cooling with Ventilation, Solar Energy Research Institute, USA, 1987 [8] Ernest, D.R., Bauman, F.S and Arens, E.A., The Prediction of Indoor Air Motion for Occupant Cooling in Naturally Ventilated Buildings, ASHRAE Technical Data Vol No 1, 1991 [9] Jozwiak, R., Kacprzyk, J., and Zuranski, J.A., Influence of Wind Direction on Natural Ventilation of Apartment Buildings, Journal of Wind Engineering and Industrial Aerodynamics, 60 (1996), 167-176 [10] Aynsley, R., Estimating Summer Wind Driven Natural Ventilation Potential for Indoor Thermal Comfort Journal of Wind Engineering and Industrial Aerodynamics, 83 (1999), 515-525 165