Workplace productivity and individual thermal ...

Building and Environment 91 (2015) 42e50

Contents lists available at ScienceDirect

Building and Environment journal homepage: www.elsevier.com/locate/buildenv

Workplace productivity and individual thermal satisfaction Shin-ichi Tanabe a, *, Masaoki Haneda b, c, Naoe Nishihara d, 1 a

Department of Architecture, Waseda University, 3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan Research Institute for Science and Engineering, Waseda University, 3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan c Toda Corporation, 1-7-1 Kyobashi, Chuo-ku, Tokyo 104-8388, Japan d Department of Education, University of the Sacred Heart, 4-3-1 Hiroo, Shibuya-ku, Tokyo 150-8938, Japan b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 30 December 2014 Received in revised form 23 February 2015 Accepted 24 February 2015 Available online 5 March 2015

This study examines the relationship between individual thermal satisfaction and worker performance. Field measurements and a questionnaire survey were conducted within an organization participating in the COOL BIZ energy conservation campaign. A subjective experiment was also conducted in a climate chamber with eleven Japanese male subjects, testing five scenarios combining operative temperature (25.5 C and 28.5 C), clothing (with and without suits), and cooling items (desk fan, air-conditioned shirt, mesh office chair). From the individual analysis, actual air temperature in the COOL BIZ office was poorly correlated with self-estimated performance, whereas perceived thermal satisfaction correlated well with self-estimated performance (R2 ¼ 0.944, p < 0.001). The results of the subjective experiment indicate that performance during simulated office work (i.e. multiplication and proof reading tasks) increased with greater individual thermal satisfaction (R2 ¼ 0.403 and 0.464, p < 0.001). The finding that perceived thermal satisfaction of occupants is reflected in objective measurement of office work performance has practical implications for the evaluation of thermal satisfaction in real offices as a means to boost workplace productivity. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Thermal satisfaction Office work performance Workplace productivity Summer season

1. Introduction There is an extensive body of research on comfort and satisfaction in indoor environments, which includes a strong focus on directly relating elements of indoor environments with workplace productivity. However, there is no clear evidence that employee satisfaction with the indoor environment is associated with improved productivity. The main aim of this study is to examine the relationship between individual thermal satisfaction and worker performance. In Japan, building-related carbon dioxide emissions, or the sum of the “commercial” and “residential” sectors, account for nearly one-third of the total emissions, and show the greatest increase. In order to reduce emissions from the “commercial” sector, the COOL BIZ campaign [1] has been promoted by the Japanese government since the summer of 2005, which involves raising the preset temperature for cooling, and modifying the business dress code in

* Corresponding author. Tel.: þ81 3 5292 5083; fax: þ81 3 5292 5084. E-mail addresses: [email protected] (S.-i. Tanabe), [email protected]. jp (N. Nishihara). 1 Tel.: þ81 3 3407 5976. http://dx.doi.org/10.1016/j.buildenv.2015.02.032 0360-1323/© 2015 Elsevier Ltd. All rights reserved.

offices during summer. The issue of energy conservation has been more actively addressed in Japanese offices after the Great East Japan Earthquake of 2011 [2]. The most well-known catchphrase of the COOL BIZ campaign is “28 C,” which is the upper limit to invoke cooling, set by the “Act on Maintenance of Sanitation in Buildings” [3]. The effect of removing jackets and ties was reported to be equivalent to lowering air temperature by 2 K [4]. The Energy Conservation Centre, Japan (ECCJ) reported that 1.2% of the annual energy consumption of an HVAC (heating, ventilation, and air conditioning) system can be saved by raising the temperature set point during summer from 26 C to 28 C. Since the annual primary energy consumption of a typical office building in Japan is 2225 MJ/ (m2$yr), the reduction under the COOL BIZ campaign was estimated to be 26.7 MJ/(m2$yr) [5]. Although the effect was estimated in terms of energy conservation, there was no recognition of the effect of raising air temperature by 2 K on worker performance. Several studies have reported the effects on work performance of a moderately warm environment, corresponding to a COOL BIZ office. Tanabe et al. [6] tested the effect of moderately hot environments at 25 C, 28 C, and 33 C on office work performance via a subjective experiment. The effect differed between task types and was inconsistent, while there was a clear trend of greater mental

S.-i. Tanabe et al. / Building and Environment 91 (2015) 42e50

fatigue with higher temperature. Witterseh et al. [7] conducted a subjective experiment to test the combined effect of thermal (22 C corresponding to thermally neutral, 26 C corresponding to slightly warm, 30 C corresponding to warm) and acoustic environments (35 dBA and 55 dBA with extra recorded noise) over an exposure period of 180 min. In the conditions without noise, there was no significant difference in the performance of simulated office work at different temperatures. In terms of the effect of thermal discomfort, the subjects who felt too warm (pooled results at both 26 C and 30 C) made 56% more errors during a mathematical addition task. Lan et al. [8] investigated the effects of thermal discomfort on health and human performance by testing two conditions at 22 C and 30 C with twelve subjects. Task performance decreased when the subjects felt warm at 30 C than when they felt thermally neutral at 22 C. Tanabe et al. [9] performed a field survey at a call center in Japan to investigate the effect of indoor air temperature on call response rate. The results showed that raising indoor air temperature by 1.0 K from 25.0 to 26.0 C was associated with a 1.9% reduction in call response performance. €nen et al. [10] Following a detailed review of the literature, Seppa proposed an equation relating indoor air temperature and relative performance of office work. In this equation, performance decreases with increasing indoor temperature higher than 21.8 C. In addition to indoor air temperature, many of previous laboratory studies that focused on the effect of thermal comfort on worker performance considered other elements of thermal environment. Wyon et al. [11] conducted a subjective experiment that provided two ensembles of clothing (1.15 clo and 0.60 clo) that allowed subjects to adjust the indoor temperature according to their preference in each condition. Several types of simulated office tasks were performed during 2.5 h of exposure; however, no significant difference was observed between the conditions. Witterseh [12] exposed subjects to test conditions (duration 173 min); clothing was adjustable to maintain thermal neutrality at 22 C and 25 C, and clothing was fixed to feel slightly warm at 22 C. Subjects made significantly fewer errors during a mathematical addition task at 25 C than at 22 C. Fang et al. [13] found no significant difference in performance when subjects were exposed to conditions of 20 C/40%RH, 23 C/50%RH and 26 C/60%RH for 280 min. The subjects were allowed to adjust their clothing so that they were thermally neutral (0 ± 0.5) in these conditions. The decline in performance when exposed to different temperatures was not significant when the thermal perception of the subjects was within the comfort zone. Willem [14] exposed 96 Singaporean subjects to temperatures of 20 C, 23 C, and 26 C for 245 min. Average thermal sensations reported by the subjects at these conditions were 1.8, 0.9, and þ0.5, and were comparatively colder than votes by European subjects reported by Witterseh et al. [7]. This might indicate that the origin or the habitual climate of subjects can influence their thermal sensation. Performance during a proof reading task was significantly better at 20 C and 26 C than at 23 C; and typing speed was faster at 20 C than at 23 C and 26 C [14]. From the review of the previous literature, it is clear that there are no effects on performance at neutral thermal conditions, whereas the effects are inconclusive at non-neutral conditions. As de Dear et al. [15] concluded from an extensive literature review, the effects of thermal comfort on task performance and productivity remain ambiguous due to diverse definitions of the productivity metric. Huizenga et al. [16] conducted a questionnaire survey at 215 buildings in the US, Canada, and Finland, and reported that satisfaction with workstation temperature was strongly correlated with self-assessed productivity. Unfortunately, no further attempt has been made to relate satisfaction with thermal environment to office work performance.

43

In this study, a field survey was conducted to investigate thermal environment in an office at which the workers were responsible for promoting the COOL BIZ campaign. A questionnaire survey was conducted to evaluate the effect of thermal satisfaction on productivity. Then, a subjective experiment was conducted in a climate chamber to evaluate the effect of improving thermal satisfaction (by introducing cooling items) on the performance of simulated office work. 2. Field survey in COOL BIZ office 2.1. Methods 2.1.1. Investigated office The field survey was conducted in an office on the 23rd floor of a 26-story office building that was constructed in 1983 in Tokyo, Japan. The office, as shown in the floor plan in Fig. 1, was split into two rooms on the east and west sides of the building, and covered a floor area of 530 m2. Forty-eight workers are based in the east room and sixty-eight in the west. During the COOL BIZ campaign from June 1 to September 30, the cooling temperature in the office was set at 28 C and the workers were encouraged not to wear suit jackets and ties. Operation of the HVAC system started at 09:30 and lasted until 19:00 during this survey. 2.1.2. Measurement of the thermal environment of the office The thermal environment of the office was measured from July 23 to September 30, 2007. The measured points are indicated in Fig. 1. Air temperature and relative humidity were measured at 36 points in the office with RSW-20S (Espec) thermo-recorders positioned 0.6 m above the floor. Vertical air temperature profile was measured at the perimeter and the interior in the east room with an RTW-30S (Espec) thermo-recorder positioned at 0.1 m, 0.6 m, 1.1 m, 1.7 m, and 2.1 m above the floor. Supply air temperature of the HVAC system was measured at 8 points in the office (RTW-30S). Supply air temperature of the fan-coil unit (FCU) was measured at 6 points (RTW-30S). All measurement intervals were 10 min. 2.1.3. Questionnaire survey A questionnaire survey was conducted in three periods: July, August, and September. In each period, three pairs of questionnaires for the “Beginning” and “End” of the working day were distributed to and collected from 105 workers. The “Beginning” questionnaire consisted of present health status, assessment of indoor environment, and subjective fatigue symptoms [17]. The “End” questionnaire assessed indoor environment during the working hours, self-estimated performance, mental workload, concentration on work, motivation for work, usage of cooling items, and subjective fatigue symptoms. The scoring schemes for thermal satisfaction and self-estimated performance are shown in Fig. 2. The cooling items used in the office included a personal paper fan, a personal electric fan on the desk, a shared electric fan on the floor, and/or a dehumidifier. 2.2. Results 2.2.1. Measurement of thermal environment in the office The mean indoor air temperature of the office during operation of the HVAC system was approximately 28 C, which was the set temperature point for cooling; however, the temperature varied with location and time. As an example of a typical summer day, the planar temperature distribution at 12:00 on August 8 was obtained by bilinear interpolation, as shown in Fig. 3. The indoor air temperature tended to be higher in areas with high density of heat sources, such as workers and electrical appliances. Fig. 4

44


Fig. 1. Office floor plan and measuring points.

Fig. 2. Scoring schemes for thermal satisfaction and self-estimated performance.

Fig. 3. Planar temperature distribution at 12:00 on August 8, 2007.

shows changes in indoor air temperature and Fig. 5 shows changes in indoor relative humidity on August 8, representative of a typical summer day. The indoor air temperature could exceed 30 C by the time the HVAC system started operation in the morning. On August 8 and also on days in July and August when

the outdoor temperature was high, it usually took more than 90 min to cool the indoor air temperature to 28 C. As in Fig. 5, the air was dehumidified while cooling; the relative humidity was kept between 50%RH and 60%RH during the operation of the HVAC system.


45

Paper fans

65

Electric fans on the desk

17

Electric fans on the floor

32

Dehumidifiers

0

Nothing

13 0

10

20

30

40

50

60

70

Response rate [%]

Fig. 6. Usage of cooling items in COOL BIZ office (n ¼ 78).

Fig. 4. Changes in indoor air temperature on August 8, 2007.

Perimeter (West)

of using cooling items are shown in Fig. 6. Most office workers used fans to alleviate their heat stress.

Perimeter (East)

Interior (West)

Interior (East)

Outdoor relative humidity

Downtime of HVAC system

Relative humidity [%RH]

80 70

60 50 40 0:00

3:00

6:00

9:00

12:00

15:00

18:00

21:00

24:00

Fig. 5. Changes in indoor relative humidity on August 8, 2007.

2.2.2. Perception of thermal environment The perception of thermal environment is summarized in Table 1. The percentage of dissatisfied workers was calculated as the ratio of dissatisfied votes relative to the number of valid responses. From the questionnaire, many workers felt slightly warm to warm and more than 70% were thermally dissatisfied in July and August. 2.2.3. Self-estimated performance As shown in Table 1, self-estimated performance averaged between 60 and 68, and was lowest in the August period and highest in September. Self-estimated performance was higher when thermal sensation was lower or closer to thermally neutral, or when thermal satisfaction was higher. 2.2.4. Clothing of occupant and use of cooling items In a prior survey in this office about clothing during the COOL BIZ period, the average value of the basic thermal insulation of clothing (Icl) was calculated according to ISO 9920 [18]. The Icl values were 0.54 clo for males and 0.52 clo for females. The results

2.2.5. Analyses of thermal environment and performance vote The relationship between air temperature and self-estimated performance is shown in Fig. 7, while Fig. 8 shows that between thermal satisfaction and self-estimated performance. In Fig. 7, the data set for temperature of surrounding air (0.5 K intervals) was averaged and compared with corresponding self-estimated performance (n ¼ 335). In Fig. 8, the data set for thermal satisfaction (0.2 interval) was averaged and compared with corresponding selfestimated performance (n ¼ 335). In Figs. 7 and 8, the size of the circle represents the number of valid responses. The relationship between surrounding air temperature and selfestimated performance was inconclusive (R2 ¼ 0.045). On the other hand, a much stronger relationship (R2 ¼ 0.944) was found between greater thermal satisfaction and higher self-estimated performance.

3. Subjective experiment 3.1. Purpose Provision of thermal satisfaction can be important for achieving optimal performance in offices with moderately high temperatures. The evidence in the previous section was based on subjectively estimated performance; the subjective experiment was conducted to support it with objectively measured performance. Reduction of clothing insulation and control of air movement around the occupants are examples of methods to improve satisfaction with thermal environment. To reduce clothing insulation, both reducing clothing items and changing the type of office chair [19] can be considered. Providing personal control of air velocity [20] was reported to be useful for improving work performance and reducing fatigue.

Table 1 Perception of thermal environment and self-estimated performance (beginning/end of the working day). July

Perception of thermal environment Number of valid responses Thermal sensation Thermal satisfaction Percentage dissatisfied Self-estimated performance Number of valid responses Self-estimated performance

August

September

Beginning

End

Beginning

End

Beginning

End

171 1.2 0.30 73%

170 1.0 0.29 74%

142 1.8 0.50 80%

141 1.1 0.35 70%

59 0.2 0.04 37%

59 0.8 0.14 64%

137 65

141 60

60 68

46


Fig. 9(c), the back and seat areas comprise mesh material that allows air movement and the removal of heat and moisture. Electricity consumption was measured with a small electrical power meter (NTT-AT, SHW3A); consumption by the desk fan was 26.1 W, and that of the shirt was 2.8 W.

Fig. 7. Relationship between surrounding air temperature and self-estimated performance (y ¼ 1.19x þ 96.5, R2 ¼ 0.045, p < 0.001, n ¼ 335).

Self-estimated performance

100 21

80 60

5

74

22

55

73 32

40

9

7

37

3.2.2. Experimental setup of subjective experiment The experiment was conducted in a climate chamber on Tuesdays to Fridays from September 18 to November 4, 2007. The experiment lasted from 12:00 to 18:00. The chamber was airconditioned during the experiment. The layout of the chamber and a picture taken during the experiment are shown in Fig. 10. Desk fans were placed on the left corner on the workstation. 3.2.3. Subjects Eleven Japanese male students (age: 22.4 ± 1.6; mean height: 171.9 ± 4.7 cm; mean weight: 60.6 ± 4.8 kg) were recruited and divided into four groups, each of two or three subjects. Each group participated in the experiment once a week, repeated at the same time and day. To keep the subjects motivated, they were told that they would receive a bonus depending on their performance.

20 0 -1 Dissatisfied

-0 Just Dissatisfied

+0 Just satisfied

+1 Satisfied

Thermal satisfaction

Fig. 8. Relationship between thermal satisfaction and self-estimated performance (y ¼ 21.1x þ 69.3, R2 ¼ 0.944, p < 0.001, n ¼ 335).

3.2. Methods 3.2.1. Cooling items Three cooling items were used in this study, namely: desk fan (Denkyosha, Tower-fan, DAT-2171R/N), a type of work shirt with integrated fans (Kuchofuku, K-200), and office chair with mesh fabric (Itoki, VESSREAL, KV2-SL-EL). The desk fan, shown in Fig. 9(a), can be turned off, adjusted to three levels of air volume, and allows oscillating the direction of air movement. The shirt with integrated fans, shown in Fig. 9(b), can be adjusted to off, or operated at two air supply rates. For the office chair, shown in

3.2.4. Experimental conditions Conditions were set according to a combination of operative temperature, clothing insulation and use of cooling items. In the “25.5 C-Suit” scenario, the operative temperature was set at 25.5 C with clothing ensemble of suit jacket, tie, long-sleeved business shirt, short-sleeved T-shirt, thin trousers, underwear, socks, and leather shoes, i.e. 0.96 clo [21]. The “28.5 C-Suit” scenario repeated the same clothing ensemble with operative temperature of 28.5 C, simulating an increased temperature set point while not modifying dress code. In “28.5 C-CB,” the operative temperature was 28.5 C with lightweight business ensemble of short-sleeved business shirt, short-sleeved T-shirt, thin trousers, underwear, socks, and leather shoes, i.e. 0.57 clo. In “28.5 C-DF,” the desk fan was used with an operative temperature of 28.5 C and lightweight clothing ensemble. In “28.5 C-ALL,” the desk fan was used and the air-conditioned shirt was worn instead of a standard short-sleeved business shirt, while other clothing items matched those in scenario 28.5 C-CB. The mesh office chair was only used in 28.5 C-ALL; all other scenarios used a chair with normal cushions. In scenarios 28.5 C-DF and 28.5 C-ALL, the subjects were allowed to adjust the speed of the desk fan, the direction of air movement,

Fig. 9. Cooling items.


Desk fan

47

2,700mm

3,600mm

Measurement instruments: Thermo-recorder and IAQ monitor

Thermocouples, Globe thermometer

Fig. 10. The layout of the chamber and picture taken during the subjective experiment.

and could select whether the fan should operate in oscillating mode; and could adjust the speed of the fan incorporated into the shirt. Relative humidity (RH) was set at 50%. Fluorescent lights on the ceiling were controlled to provide illumination of 750 lx at the desk level. The main noise sources during the experiment consisted of the fan of the air-conditioning system, the desk fans, and the airconditioned shirts; noise from operation of personal computers; and any noise made by the subjects. Eating was not allowed during the experiment. Subjects were provided with bottles of plain mineral water and were allowed to drink whenever necessary. Prior to exposure, the subjects practiced the experimental procedures. The condition during this session was the same as in the 28.5 C-Suit scenario. The experimental conditions were balanced for order of presentation.

3.2.5. Measurements 3.2.5.1. Physical measurements of indoor environment. The physical parameters measured in the chamber were air temperature, mean radiant temperature, relative humidity, carbon dioxide concentration, and desktop illumination. Air temperature was measured every minute using copper-constantan thermocouples at 1.1 m above the floor. The mean radiant temperature was measured using a globe thermometer; relative humidity using the RSW-20S (ESPEC) thermo recorder; and carbon dioxide concentration using the IAQ monitor (Model2332, KANOMAX) every minute at 1.1 m above the floor. The instantaneous value of desktop illuminance was measured at the center of every desk using a photo-recorder TRL-10 (ESPEC) before the subjects entered the chamber. The noise level of A-weighting was measured using an NL-31 noise meter (RION) every second for 30 min at 1.1 m on another day following each experimental scenario. The equivalent noise level was calculated.

3.2.5.2. Perception of indoor environment. Subjects were asked a series of questions on their perceptions of the indoor environment, including: thermal environment, humidity, indoor air quality, lighting, noise, and working environment. The voting schemes were presented on a PC screen, and subjects clicked on each of the continuous scales to indicate their perception.

3.2.5.3. Performance of simulated office work. The subjects completed three types of simulated office work: three-digit multiplication, proof reading, and creative thinking. 3.2.6. Experimental procedure The experimental procedure is shown in Fig. 11. In the waiting room, the subjects changed their clothing to that of the experimental scenario, received instructions, and rested while seated for approximately 10 min. The subjects then entered the climate chamber and sat at their workstation. They then judged the indoor environment, and their levels of fatigue and sleepiness. Their finger pulse wave was measured, and they worked on the P-Tool performance evaluation tool [22] for 12 min. They then rested while seated for 30 min to adapt to the conditions in the chamber. Subsequently, the subjects reevaluated the indoor environment, fatigue, and level of sleepiness. Their finger pulse wave was measured and they also worked on the P-Tool. Then they performed three sessions consisting of the multiplication task for 30 min, proof reading for 25 min, and creative thinking task for 20 min. Following each task session, the subjects completed the NASA-TLX [23] questionnaire, indicated their level of concentration on the task, and judged the indoor environment, their fatigue, and level of sleepiness. Their finger pulse wave was measured and they worked on the P-Tool. In the 28.5 C-DF and 28.5 C-ALL scenarios, the subjects were allowed to start using and adjusting the desk fan and air-conditioned shirt from the initial rest/adaptation period in the climate chamber (shown as D in Fig. 11). Exposure to the experimental conditions lasted 338 min, of which 225 min involved simulated office tasks. 3.2.7. Statistical analysis The Friedman test and two-way analysis of variance by ranks were used to test for differences between the scenarios (p ¼ 0.05). The Wilcoxon signed-rank test was used when a significant difference was observed. The risk of family-wise Type-I error was controlled by the HolmeBonferroni method [24]. The Wilcoxon signed-rank test was used to compare whether the perceptions of indoor environment changed before and after the adaptation period, during which subjects were allowed free use of the cooling items in the 28.5 C-DF and 28.5 C-ALL conditions. The results of multiple comparisons with significant difference were presented in

48


Fig. 11. Experimental procedure.

3.3.1. Physical measurements The physical parameters of the indoor environment are summarized in Table 2. The operative temperature was controlled as intended. The average carbon dioxide concentration indicated that the ventilation rate was kept almost constant throughout the experiment.

than with 28.5 C-Suit (p < 0.05). After Task Session 1, they were significantly more satisfied with the 28.5 C-CB, 28.5 C-DF, and 28.5 C-ALL conditions compared with 28.5 C-Suit (p < 0.05). After Task Session 2, they were significantly more thermally satisfied with 28.5 C-DF and 28.5 C-ALL than with 28.5 C-CB (p < 0.05). After Task Session 3, they were significantly more thermally satisfied with 28.5 C-ALL than with 28.5 C-Suit and 28.5 C-CB (p < 0.05). In conditions 28.5 C-DF and 28.5 C-ALL, votes cast towards the end of the adaptation period showed significantly greater thermal satisfaction than those before the period (p < 0.05). This evidenced the effect of allowing the use of the desk fan and airconditioned shirt after adaptation had started.

3.3.2. Thermal satisfaction In this section, comparisons between the conditions are discussed in terms of 28.5 C operative temperature, although statistical analyses were conducted for all five conditions. The result for thermal satisfaction is shown in Table 2. After the adaptation period, the subjects expressed significantly greater thermal satisfaction with the 28.5 C-DF and 28.5 C-ALL scenarios

3.3.3. Performance of simulated office work For the multiplication task, performance indices consisted of the number of correct answers per hour and the accuracy of answers. For proof reading task, the accuracy of detecting errors, the number of errors detected and the number of characters read per hour were calculated as performance indices. For the creative thinking task, the number of responses per hour was calculated. There were no

the text. Linear regression analysis was used to study the relationship between thermal satisfaction and normalized performance of tasks. 3.3. Results

Table 2 Measured conditions in the climate chamber during task sessions.

Physical measurement Operative temperature [ C] Relative humidity [%RH] CO2 conc. [ppm] Desktop illuminance [lx] Noise level, LAeq [dB] Thermal satisfaction Before adaptation After adaptation** After task session 1*** After task session 2* After task session 3* Multiplication task Accuracy Number of correct answers (answers/hr) Proof reading task Accuracy Number of errors detected (detected/hr) Reading speed (characters/hr) Creative thinking task Number of ideas (ideas/hr)

25.5 C-Suit

28.5 C-Suit

28.5 C-CB

28.5 C-DF

28.5 C-ALL

25.5 (3) 49 (2) 735 (73) 743 (18) 51

28.6 (0.4) 46 (3) 680 (83) 726 (13) 51

28.6 (0.3) 46 (3) 734 (60) 736 (15) 51

28.5 (0.2) 48 (1) 686 (44) 736 (21) 54

28.6 (0.1) 47 (1) 707 (63) 735 (22) 55

0.23 0.47 0.38 0.64 0.51

0.16 (0.50) 0.02 (0.52) 0.40 (0.37) 0.02 (0.60) 0.18 (0.68)

0.00 0.34 0.33 0.38 0.42

0.10 0.57 0.48 0.60 0.74

0.11 0.56 0.57 0.71 0.76

92% 124

92% 116

89% 118

92% 123

91% 124

74% 185 23,384

73% 177 22,482

72% 175 22,785

71% 185 23,836

74% 185 23,269

28

30

27

28

28

(0.50) (0.44) (0.42) (0.33) (0.39)

(0.50) (0.41) (0.47) (0.43) (0.35)

The results of Friedman test were noted in the table as *: p < 0.05, **: p < 0.01, ***: p < 0.001 (Standard deviation).

(0.53) (0.35) (0.35) (0.32) (0.29)

(0.60) (0.47) (0.36) (0.35) (0.31)


49

significant differences in performance between the conditions. The indices were averaged over all task sessions, and the results are summarized in Table 2. 3.3.4. Thermal satisfaction and performance of simulated office work To determine the effect of thermal satisfaction on the performance of simulated office work, the thermal satisfaction votes after each task session were correlated with performance in the multiplication (number of correct answers) and proof reading (number of errors detected) tasks. To exclude personal variability from the performance assessments, the normalized performance s of the number of correct answers xi,j,k in session k of condition j by subject i was obtained by Equation (1). In the equation, xi is the average number of correct answers per session of subject i across all conditions, and si is the standard deviation of the average.

s xi;j;k ¼

xi;j;k xi si

(1)

The data set for thermal satisfaction vote (intervals of 0.2) was averaged, and compared with the corresponding normalized performance. The relationship between thermal satisfaction and normalized performance in the multiplication task is shown in Fig. 12; and that between thermal satisfaction and normalized performance in the proof reading task is shown in Fig. 13. The size of the plots in the figures represents the number of votes used for calculating the averages. The coefficients of determination (R2 value) of the linear regression, as weighted by the number of corresponding votes, were 0.403 for the multiplication task and 0.464 for proof reading. From the results, performance in the multiplication and proof reading tasks was higher when subjects were more thermally satisfied. 4. Discussion 4.1. Thermal satisfaction and performance Perceived thermal satisfaction or personal assessment of thermal environment would provide practical indications of the effect of thermal environment on productivity especially in the situation that the modification of thermal environment was allowed personally. In the surveyed office, self-estimated performance has a stronger correlation with perceived thermal satisfaction than with actual air temperature. This finding is inconsistent with the relationship obtained in the field study at a call center [9]. A possible reason could be that the situation of the office promoting COOL BIZ was slightly different from the call center; most workers in the COOL BIZ office had to use cooling items such as personal fan to

Fig. 13. Thermal satisfaction and standard score (y ¼ 0.254x 0.107, R2 ¼ 0.464, p < 0.001, n ¼ 165).

of

proof

reading

task

alleviate their heat stress, by which they actively controlled their thermal environment. From the results of the COOL BIZ office survey, workers indicate that greater thermal satisfaction facilitates higher workplace performance. Similar results were obtained from the subjective experiment in the climate chamber. Performance in multiplication and proof reading tasks was higher in indoor environments with which the subjects were more thermally satisfied. The results provide experimental evidence to support the hypothesis that thermal satisfaction affects workplace performance. 4.2. Improvement of thermal satisfaction by introducing cooling items The experimental results show significant improvement in thermal satisfaction following the use of cooling items, i.e. desk fan and air-conditioned shirt, in the 28.5 C-DF and 28.5 C-ALL scenarios. Besides reducing clo value, the introduction of cooling items can improve thermal satisfaction of occupants in COOL BIZ office environments. As individuals have different preferences, it seems reasonable to improve thermal satisfaction by providing personal control of the thermal environment. 4.3. Type of task and the effect of thermal environment on performance The experimental results show no significant difference in the performance of simulated office tasks after different cooling items were introduced. Performance in the multiplication and proof reading tasks were lower (non-significant) in the 28.5 C-Suit scenario than in the other conditions, whereas performance in the creative thinking task was highest (non-significant) in the 28.5 CSuit scenario. In previous studies, performance in creative thinking tasks was better at a temperature of 27 C or at lowered level of arousal compared with 20 C and 23.5 C [25,26]. The effect of thermal environment on creative thinking performance may differ from the effect on tasks such as multiplication and proof reading, whose performance is higher when the level of arousal is higher. The differing effect of thermal environment on the performance of simulated office work was also evidenced in this experiment. 5. Conclusions

Fig. 12. Thermal satisfaction and standard score (y ¼ 0.245x 0.103, R2 ¼ 0.403, p < 0.001, n ¼ 165).

of

multiplication

task

Field measurements were conducted from July 23 to September 30, 2007 to investigate the thermal environment in a COOL BIZ office. A questionnaire survey was conducted during the measurement to evaluate the effect of thermal environment on productivity. Subsequently, eleven Japanese male subjects were included in a subjective experiment conducted in a climate

50


chamber, to test five different scenarios for operative temperature, clothing, and cooling items. Spatial and temporal variations in air temperature were observed in the surveyed office. The indoor air temperature was overheated and exceeded 30 C in mornings, especially during midsummer. From the individual analysis, actual air temperature was poorly correlated with self-estimated performance; Selfestimated performance correlated well with thermal satisfaction. In the climate chamber experiment, subjects expressed greater thermal satisfaction in the scenario combining operative temperature of 28.5 C with the use of cooling items. The experimental findings demonstrate that the performance of simulated office work, i.e. multiplication and proof reading tasks, improved when individual thermal satisfaction was higher. Based on the evidence that subjective thermal satisfaction among employees influences objective measurements of workplace performance, the evaluation of thermal satisfaction in real offices has clear practical benefits for improving workplace productivity. Acknowledgments The authors are grateful to the workers in the surveyed office and to the experimental subjects. The authors thank then-graduate students for their contributions to this study. The study was partially funded by the Global Environment Research Fund (H-061) by the Ministry of the Environment, Japan; and by the Project Research of the Advanced Research Institute for Science and Engineering, Waseda University. We would like to thank Editage (www. editage.jp) for English language editing. References [1] Doi K. COOL BIZ. J SHASEJ 2006;80(7):5e7 [in Japanese]. [2] Tsushima S, Tanabe S, Utsumi K. Workers' awareness and indoor environmental quality in electricity-saving offices. Build Environ 2014. http:// dx.doi.org/10.1016/j.buildenv.2014.09.022 [available online 2 Oct., 2014]. [3] Ministry of Health, Labour and Welfare, Japan. Act on maintenance of sanitation in buildings (1970). 2006. [4] Okuma R, Ishino H, Nakayama S. Thermal comfort for office occupants in air temperature of 28 C in summer. J Environ Eng AIJ 2007;618:31e6 [in Japanese]. [5] The Energy Conservation Center, Japan. Guidebook for energy conservation in building of FY2005. 2005. [6] Tanabe S, Nishihara N. Productivity and fatigue. Indoor Air 2004;14(s7): 126e33.

[7] Witterseh T, Wyon DP, Clausen G. The effects of moderate heat stress and open-plan office noise distraction on SBS symptoms and on the performance of office work. Indoor Air 2004;14(s8):30e40. [8] Lan L, Wargocki P, Wyon DP, Lian Z. Effects of thermal discomfort in an office on perceived air quality, SBS symptoms, physiological responses, and human performance. Indoor Air 2011;21:376e90. [9] Tanabe S, Kobayashi K, Kiyota O, Nishihara N, Haneda M. The effect of indoor thermal environment on productivity by a year-long survey of a call centre. Intell Build Int 2009;1(3):184e94. €nen O, Fisk WJ, Lei QH. Effect of temperature on task performance in [10] Seppa office environment. In: Proceedings of the 5th international conference on cold climate HVAC (on CD-ROM); 2006. [11] Wyon DP, Fanger PO, Olesen BW, Pederson CJK. The mental performance of subjects clothed for comfort at two different air temperatures. Ergonomics 1975;18(4):359e74. [12] Witterseh T. Environmental perception, SBS symptoms and the performance of office work under combined exposures to temperature, noise and air pollution [Ph.D. thesis]. Technical University of Denmark; 2001. [13] Fang L, Wyon DP, Clausen G, Fanger PO. Impact of indoor air temperature and humidity in an office on perceived air quality, SBS symptoms and performance. Indoor Air 2004;14(s7):74e81. [14] Willem HC. Thermal and indoor air quality effects on physiological responses, perception and performance of tropically acclimatized people. Chapter 5 [Ph.D. thesis]. National University of Singapore; 2006. [15] de Dear RJ, Akimoto T, Arens EA, Brager G, Candido C, Cheong KWD, et al. Progress in thermal comfort research over the last twenty years. Indoor Air 2013;23:442e61. [16] Huizenga C, Abbaszadeh S, Zagreus L, Arens E. Air quality and thermal comfort in office buildings: results of a large indoor environmental quality survey. Proc Healthy Build 2006;III:393e7. [17] Yoshitake H. Occupational fatigue-approach from subjective symptom. Tokyo, Japan: The Institute for Science of Labor; 1973 [in Japanese]. [18] ISO 9920:2009. Ergonomics of the thermal environment. Estimation of thermal insulation and water vapour resistance of a clothing ensemble. International Organization for Standardization; 2009. [19] McCullough EA, Olesen BW, Hong SW. Thermal insulation provided by chairs. ASHRAE Trans 1994;100(1):795e802. [20] Nishihara N, Tanabe S. Effect of individual control of air velocity on office worker's productivity with subjective experiment. J Home Econ Jpn 2005;56(3):153e61. [21] ASHRAE. ASHRAE handbook fundamentals. SI ed. ASHRAE; 2005. [22] Haneda M, Tanabe S, Nishihara N, Ueki M, Kawamura M. Development of survey tools for indoor environmental quality and productivity. Proc Clima 2007;1:43e50. [23] Hart SG, Staveland LE. Development of NASA-TLX (task load index): results of empirical and theoretical research. In: Hancock PA, Meshkati N, editors. Human mental workload. Amsterdam: North Holland; 1988. p. 139e83. [24] Holm S. A simple sequentially rejective multiple test procedure. Scand J Statistics 1979;6:65e70. [25] Wyon DP. The effects of moderate heat stress on the mental performance of children. National Swedish building research document, vol. 8. National Swedish Institute for Building Research; 1969. p. 83e4. [26] Wyon DP. Creative thinking as the dependent variable in six environmental experiments: a review. In: Proceedings of the 7th international conference on indoor air quality and climate, vol. 1; 1996. p. 419e22.