Thermal comfort in office buildings: Findings from a ...

1 downloads 0 Views 2MB Size Report
Thermal comfort in office buildings: Findings from a field study in mixed-mode and fully-air conditioning environments under humid subtropical conditions.
Accepted Manuscript Thermal comfort in office buildings: Findings from a field study in mixed-mode and fully-air conditioning environments under humid subtropical conditions R. De Vecchi, C. Candido, R. de Dear, R. Lamberts PII:

S0360-1323(17)30328-1

DOI:

10.1016/j.buildenv.2017.07.029

Reference:

BAE 5008

To appear in:

Building and Environment

Received Date: 6 April 2017 Revised Date:

8 July 2017

Accepted Date: 21 July 2017

Please cite this article as: De Vecchi R, Candido C, de Dear R, Lamberts R, Thermal comfort in office buildings: Findings from a field study in mixed-mode and fully-air conditioning environments under humid subtropical conditions, Building and Environment (2017), doi: 10.1016/j.buildenv.2017.07.029. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

3

ACCEPTED MANUSCRIPT Thermal comfort in office buildings: findings from a field study in mixed-mode and fully-air conditioning environments under humid subtropical conditions

4

De Vecchi, R.a,*, Candido, C.b, de Dear, R.b and Lamberts, R.a

1 2

a

Department of Civil Engineering, Federal University of Santa Catarina, Brazil.

6

b

Faculty of Architecture, Design and Planning, The University of Sydney, Australia.

7

*Corresponding author: Laboratório de Eficiência Energética em Edificações

8

(LabEEE), ECV /CTC, Campus Universitário, UFSC, Trindade, Caixa Postal 476,

9

Florianópolis-SC, Brazil. Tel.: +55 48 3721 5184.

SC

10

E-mail address: [email protected] (Renata De Vecchi)

M AN U

11 12 13 14

19 20 21 22 23 24 25

EP

18

AC C

17

TE D

15 16

RI PT

5

28

ACCEPTED MANUSCRIPT Thermal comfort in office buildings: findings from a field study in mixed-mode and fully-air conditioning environments under humid subtropical conditions

29

De Vecchi, R.a,*, Candido, C.b, de Dear, R.c and Lamberts, R.a

26 27

30 31

a

32 33

*Corresponding author. E-mail address: [email protected] (Renata De Vecchi)

34

Abstract

35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

This paper addresses thermal comfort conditions in office buildings with rudimentary mixedmode environments controlled by occupants compared to fully air-conditioning in a humid subtropical climate in Brazil. Occupants from three office buildings with two different environmental control strategies (two with mixed-mode ventilation and one with permanent airconditioning) assessed their thermal environment via “right-here-right-now‟ online questionnaires, while indoor climatic measurements were simultaneously carried out in situ. 2,688 questionnaires from 617 occupants were collected. The results indicated that airconditioning in mixed-mode (MM) buildings controlled by occupants was used permanently throughout the year without any season pattern, being specially connected to the peak outdoor air temperature. In addition, there was a strong tendency toward thermal discomfort due to excessive cold in MM buildings at times when air-conditioning mode was under operation, and hot discomfort during the naturally ventilated mode. When compared to fully-air conditioning buildings, there were similarities in terms of occupant thermal sensation and acceptability levels within the same intervals of Standard Effective Temperature (SET). The results obtained in this study could be useful as a framework for future studies, as well as a baseline for a Brazilian thermal comfort standard.

51

Keywords: thermal comfort; mixed-mode buildings; fully air-conditioning environments; humid

52

subtropical climate.

54 55 56 57

EP

TE D

M AN U

SC

RI PT

Department of Civil Engineering, Federal University of Santa Catarina, Brazil. Faculty of Architecture, Design and Planning, The University of Sydney, Australia.

AC C

53

b

ACCEPTED MANUSCRIPT 1. Introduction

59

There is a common agreement around the world that air-conditioning in buildings

60

contributes significantly to energy consumption, and therefore carbon emissions into the

61

atmosphere. In Brazil, the use of air-conditioning systems is responsible for

62

approximately 50% of office buildings’ electric energy consumption, which is broadly

63

consistent with values observed worldwide [1–3]. A trending discussion [4,5] is that

64

world populations are becoming more dependent on these energy-intensive indoor

65

climates inside buildings [6–9], and one of the main consequences being felt is during

66

peak electricity demand episodes, especially during the summer months and heat waves.

67

As an example, in Brazil, a significant change in the energy consumption profile over

68

the past six years has been observed [10], and the peak demand previously pragmatic in

69

late afternoon and early evening periods is now occurring mid-afternoon on summer

70

days [11]. This shift is specially associated with the growth of air conditioning

71

consumption in both residential and commercial building sectors [10].

72

Despite the sheer amount of energy required by Heating, Ventilation and Air

73

Conditioning (HVAC) systems in order to heat and cool office environments, research

74

findings indicate that this will not necessarily result in occupant’s satisfaction, in fact,

75

the opposite seems to be true [12]. According to Zhang, Arens and Zhai [13], there is a

76

strong trend observed in industry to overcool offices during summer and overheat

77

during winter. This practice is not only wasteful in terms of the energy consumed by

78

buildings, but it may also have a significant impact on workers’ productivity and health

79

due to discomfort, especially in the case of over cooling/heating [14–18].

80

As an alternative to avoid the undesirable, and oftentimes unnecessary use of air

81

conditioning, increasing attention is being directed towards mixed-mode ventilation

82

strategies [2,19–21] wherein there is more opportunity for occupants themselves to

83

adapt and control their immediate thermal environments, which, in turn, may result in

84

higher levels of overall occupant satisfaction [21,22]. A rich revision of those system

85

were performed by [23] and [24] focusing on different operation modes of MM

86

buildings and the influence of occupants’ and automatic control. Although there is a

87

growing body of research knowledge coming from mixed-mode buildings, a research

88

gap remains when it comes to studies conducted in different climate zones, and also in

89

buildings where occupants rather than Building Management System (BMS) decide

AC C

EP

TE D

M AN U

SC

RI PT

58

ACCEPTED MANUSCRIPT when to switch the ventilation system’s mode of operation. This is particularly true in

91

Brazil, where a significant portion of the older office building stock has had air

92

conditioning retrofitted into a basic, naturally ventilated design, rendering it mixed-

93

mode ventilation. Very little field research has been conducted in such environments to

94

date, and differently from what is observed in colder countries (where heating is needed

95

during winter seasons), Brazilian buildings use to rely only in the cooling system

96

throughout the year.

97

This paper aims to contribute to this knowledge gap by presenting results from a

98

thermal comfort study conducted in mixed-mode buildings, where occupants in most of

99

cases need find a consensus situation to control their immediate thermal environment

100

without the assistance of a BMS. Such situation somehow allows them to a higher

101

degree of autonomy if compared to a HVAC environment; thus, with comparative

102

purposes, thermal comfort responses were also involved from permanently-air-

103

conditioned office buildings analysis in which occupants had negligible thermal

104

autonomy.

105

2. Methods

106

Measurements of air temperature, air velocity, globe temperature and relative humidity

107

where taken within the occupied zone simultaneously with the administration of online

108

thermal comfort questionnaires. Field studies were carried out during a doctoral

109

research [25] from March (Autumn) through October (Spring) 2015 in three office

110

buildings located Brazil, covering all four seasons. Specifics about the climate,

111

buildings and field studies are detailed in following sections.

112

2.1. Climate description

113

Florianópolis is located in the southern region of Brazil (latitude 27°40”S). According

114

to Köppen’s classification, the city presents a humid subtropical climate, with regular

115

outdoor temperatures varying from 17 to 29 °C during summer (running from

116

December 21th to March 20th) and spring (from march 21th to June 20th); and from 13

117

to 22 °C during winter (June 21th to September 20th) and autumn (September 21th to

118

December 20th) [26]. Relative humidity is high throughout the year (minimum monthly

119

average of 80% in November and maximum monthly average of 84% in July) and there

AC C

EP

TE D

M AN U

SC

RI PT

90

ACCEPTED MANUSCRIPT is no dry season. The highest level of rainfall occurs during summer months from

121

January to March, and the lowest in winter, from July to August (mean annual

122

precipitation is 1,521 mm). The most prevalent wind directions in Florianópolis are

123

from the North and Northeast, all year round. The mean daily total solar radiation

124

incident horizontally is 4.2 kWh/m² [27].

125

Outdoor air temperature and relative humidity measured by the National Meteorological

126

Institute of Brazil (INMET – Instituto Nacional de Meteorologia), which is located

127

close to the downtown of Florianópolis, were used to characterize the outdoor

128

conditions during the field study according with the seasons (Table 1), and also to

129

support the analysis of mixed-mode environments.

130

According to Table 1, the highest value of outdoor temperature was registered during

131

autumn (28.7°C) and not in summer, as expected, which occurred due to the small

132

number of sample days falling within the hottest months. However, the mean values

133

recorded are consistent with the characteristics for each season of the year described

134

above for the climate of Florianópolis.

135 136

Table 1. Outdoor air temperature (°C) and relative humidity (%) during the days of field study in accordance with the seasons

M AN U

SC

RI PT

120

Summer Autumn Winter Spring

Min.

Mean

25.1 28.7 28.1 26.3

21.5 14.9 6.3 12.1

23.3 22.6 17.9 22.3

EP

Season

TE D

Outdoor Air Temperature (°C) Max.

RH (%)

n

81 75 69 70

83 902 1325 378

2.2. Building characteristics1

138

Field studies were conducted in three buildings with two different strategies of

139

ventilation and air conditioning: one fully-air conditioned (FAC) and two mixed-mode

140

buildings (MM). The first (Bd A, fully-air conditioned) is a 5-storey square building,

141

with a total area of 27,735 m² occupied by 1,200 employees who work from 8:00 am to

AC C

137

1

Due to constraints of a confidentiality agreement, all building-related information provided here is

limited, including use of photographs.

ACCEPTED MANUSCRIPT 6:00 pm. The second (Bd B- mixed-mode) is a 2-storey, H-shaped, mixed-mode

143

building with a total area of 5,200 m², occupied by 280 employees working from 1:00

144

pm to 7:00 pm. The third (Bd C – mixed-mode) is a 12-storey, rectangular building with

145

a total area of 4,200 m² occupied by 350 employees working from 1:00 pm to 7:00 pm.

146

All buildings present a mixed workspace layout, including open-plan and private,

147

single-occupant offices. This spatial configuration is representative of Brazil’s office

148

building stock. The study presented here was conducted within the open-plan office

149

area. Details about adaptive features are discussed as follows:

150



RI PT

142

Bd A: fully time-controlled temperature conditions, with static indoor environment set at 24°C ± 2°C and a ventilation air system fed through modular

152

vents (1x1 m) mounted in the ceiling. The windows are sealed and overlaid by

153

brise-soleils that can be adjusted by the occupants during the year. There are no

154

internal shading devices or humidity control. The facility manager just maintain

155

an occupant complaint log or help desk for air-related complaint in their

156

workspace; when it occurs, the maintenance sector installs an adapted blade to

157

redirect the airflow to opposite sides over the head, redistributing the flow to the

158

surrounding environment. Other than that, occupants were not afforded the

159

opportunity to ask for thermal conditions to be changed.

M AN U



TE D

160

SC

151

Bds 2 and 3: in these buildings, there are operable windows on all perimeter offices arranged along the external facades. Air conditioning units are installed

162

in all of the open-plan spaces delimited by movable walls, being dimensioned

163

according to the number of persons. As long as there is a consensus among

164

people sharing the same space with colleagues, they might be completely at

165

liberty to close/open the windows and shading devices (internal blinds), as well

AC C

166

EP

161

as to turn on/off the AC units.

167

2.3. Measurements and occupant questionnaire

168

Air temperature, humidity, globe temperature and air speed were measured with

169

laboratory-precision instruments configured into a microclimatic station that was

170

located at 0.60 m height above floor level at the central point within the occupied zone.

171

Readings were averaged every 1 minute. Additionally, individualized air velocity values

172

were taken with a handheld hot-wire anemometer close to each occupant (0.5 m radius)

ACCEPTED MANUSCRIPT 173

whilst their thermal comfort questionnaire was being completed. Physical parameters

174

and instruments’ accuracy are shown in Table 2.

175

Table 2. Physical measurements and instrument’s accuracy

Microclimatic Station Air Temperature (°C) Globe Temperature (°C) Relative Humidity (%) Air velocity (m/s)

Valid Range

Accuracy

0-60 0-60 5-96 0-3

±0.2°C ±0.2°C ±3.0% ±0.04 + 3%Var

Hot-wire Anemometer Valid Range

Air Temperature (°C) Air velocity (m/s)

0-80 0-20

Accuracy ±0.2°C + 3%Var

SC

Physical Parameter

RI PT

Physical Parameter

Subjective questionnaires asked occupants to assess their thermal environment and

177

register their thermal comfort, sensation, preference, acceptability and air movement

178

acceptability (see Table 3). Subjects answered the questionnaire a total of five times, at

179

20 minutes interval spanning a 100 min period (Figure 1). The online questionnaires

180

were sent directly to each participant via e-mail and completed on their desk computers.

181

Before the beginning of the study, the researchers ensured that all subjects familiarized

182

themselves with the questionnaire; and during the measurements it was always

183

reinforced to the occupants that they were free to adapt their clothing as they wished to.

184

These changes were recorded to be used on the adjustment of the clo values before SET

185

calculation.

AC C

EP

TE D

M AN U

176

Figure 1. Measurement protocol schematic representation

ACCEPTED MANUSCRIPT 186

Anthropometric characteristics (gender, age and height) were also collected along with

187

clothing and metabolic rate estimation in accordance with ASHRAE 55 [28]. A 0.1 clo

188

value was added to the ensemble insulation estimates during post-processing

189

calculations to account for chair insulation [28].

190

Table 3. The right-here-right-now comfort questionnaire used in this field study

1

Thermal Sensation: Right now, how do you feel?

- Hot - Warm - Slightly warm - Neutral - Slightly cool - Cool - Cold

2

Thermal Preference: Would you prefer to be:

- Warmer - No change - Cooler

5

+3 +2 +1 0 -1 -2 -3 +1 0 -1

- Yes - No

0 -1

- Comfortable - Uncomfortable

0 -1

Air Movement Acceptability: Right now, how would you classify the air movement in your space?

- Unacceptable, air movement too slow - Acceptable, air movement slow - Acceptable, appropriate air movement - Acceptable, air movement fast - Unacceptable, air movement too fast

+2 +1 0 -1 -2

Air Movement Preference: Right now, which air movement option would you prefer?

- More air movement - No change - Less air movement

+1 0 -1

EP

6

M AN U

4

Thermal Acceptability: Is the current thermal environment acceptable for you? Thermal Comfort: At this moment, how would you consider this thermal environment?

TE D

3

Code

RI PT

Answer

SC

Question



2.4. Data analysis

192

At first, occupants’ responses and the environmental variables were quantitatively and

193

visually summarized through descriptive statistics. The main analyses were constructed

194

using a cross-information method relating the frequency of votes’ occurrence (thermal

195

sensation, preference, comfort and acceptability, as well as air movement acceptability

196

and preference) with the calculated Standard Effective Temperature (SET). SET was

197

selected because it combines the six key thermal comfort parameters (ta, MRT, rh, v,

198

clo, met) in a unitary temperature index based on a physiologically realistic simulation

199

of the body’s heat and mass exchanges with its environment.

200

simulation model underpinning SET renders it well suited to warm environments such

AC C

191

The physiological

ACCEPTED MANUSCRIPT as those found in the present field study. The SET index was calculated for each set of

202

field measurements using WinComf© software [29]. Second, to compare and determine

203

whether two samples are likely to have come from the same two underlying populations

204

that have the same mean, t-test were performed using IBM SPSS software2. Last, a

205

PROBIT model was used to estimate and compare the neutral temperature between

206

groups based in their thermal preference votes.

207

3. Results and Discussion

208

A total of 87 field measurements were conducted including 617 occupants and 2,688

209

questionnaires completed for analysis. Among them, 1,274 (46.3%) questionnaires were

210

collected from Bd 1 and 1,414 (53.7%) from Bds 2 and 3 combined. Table 4

211

summarizes a breakdown of key indoor and outdoor variables recorded during this

212

study.

213

Table 3. Variability of indoor and outdoor parameters observed during field study

M AN U

SC

RI PT

201

Variable

AC C

EP

TE D

Mean Indoor Operative Temperature (°C) Max. Indoor Operative Temperature (°C) Min. Indoor Operative Temperature (°C) Mean Standard Effective Temperature (°C) Prevailing Mean Outdoor Temperature (°C) Mean Radiant Temperature (°C) Mean Relative Humidity (%) Mean Air Velocity (m/s) Max. Air Velocity (m/s) Min. Air Velocity (m/s) Male occupants Female occupants Mean Clothing Insulation (clo) Mean Metabolic Rate (met)

214

Mixed-mode (Bds 2 and 3 combined)

Fully-AC (Bd 1)

NV Mode n = 1,069

AC Mode n = 345

n = 1,274

23.0 26.9 16.9 23.8 19.3 22.9 62.6 0.16 0.37 0.10 543 531 0.76 1.00

23.6 26.0 20.9 23.4 22.9 23.6 60.1 0.15 0.31 0.02 151 189 0.62 1.00

22.9 24.5 21.7 23.3 21.7 22.8 61.0 0.13 0.27 0.07 796 478 0.68 1.00

215

According to Table 4, all buildings presented moderate air speed values ranging

216

between 0.1 and 0.3 m/s (mean values of 0.13 m/s for FAC building and 0.15 m/s for

217

MM buildings). These low air velocity values were observed in MM buildings even in

218

naturally ventilated mode with windows open as a consequence of the poor cross

2

IBM Corp. Released 2013. IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp.

ACCEPTED MANUSCRIPT ventilation conditions. Clothing and metabolic rate values were similar in all buildings;

220

but the highest mean clo values were observed during the naturally ventilated period in

221

MM buildings, and this condition was commonly observed during colder days. The

222

ratio of male to female respondents in each type of building differed slightly: in MM

223

buildings, the percentage of males was 49% and females 51%, while in FAC buildings

224

these percentages were 63% and 37% respectively.

225

Figure 2 shows the frequency of the Standard Effective Temperature (SET) considering

226

the same values interval for both types of buildings (MM and FAC). A comparison

227

reveals that the SET values registered in the MM buildings presented a wider range of

228

distribution (19-28 °C), while FAC buildings tended to register values close to 22-24 °C

229

(ranging between 21-27 °C). The wider range of temperatures normally distributed

230

around the mode of 24 °C in Figure 2a strongly characterizes the behaviour of the MM

231

buildings, where indoor conditions fluctuate in sympathy outdoor conditions (registered

232

dry-bulb temperature ranged between 17-28 °C), whereas in fully-air conditioned

233

buildings (Figure 2b) there is a tendency to remain in a predominant thermal condition,

234

with no expectation of seasonal nor synoptic variation (registered dry-bulb temperature

235

ranged between 22-25 °C).

a)

AC C

EP

TE D

M AN U

SC

RI PT

219

b)

Figure 2. Comparison between standard effective temperature values observed in FAC building on the right, and MM buildings on the left

236

3.1. Observed mixed-mode features and operation

237

The results of this section focus solely on data collected from Bd 2 and 3 (mixed-mode

238

buildings), pooled according predominant ventilation operation mode at the time of

239

survey, namely: 1) when natural ventilation was in use (NV mode); and 2) with air-

ACCEPTED MANUSCRIPT conditioning in use (AC mode). During the field study, seasons in which natural

241

ventilation was a more likely option for occupants were prioritized in the field campaign

242

schedule. Thus, months with lower external air temperatures were preferred and

243

therefore a higher frequency of data occurred in autumn and winter (93% of votes in

244

those seasons). A small number of the questionnaires was collected during summer and

245

spring days (only 7%), and most of these were during spring – see Figure 3. It is

246

important to highlight that in Brazil, collecting representative data from commercial

247

buildings on summer days is impeded by the festive seasons (Christmas, New Year and

248

Carnival) and the vacation period, since work conditions are typically far from normal

249

and few occupants are present inside the buildings.

M AN U

SC

RI PT

240

TE D

Figure 3. Distribution of data from Bd 2 and 3 filtered by season and operation mode

Figures 4 and 5 provide distribution of SET values in mixed-mode buildings sorted

251

according to ventilation system mode of operation. These figures also show the daily

252

outdoor temperature readings and the maximum outdoor temperature considering only

253

the occupied hours of the buildings (from 8am to 7pm). It is interesting to note that, as

254

observed previously in Figure 2 (comparison between MM and FAC buildings), the

255

SET temperatures presented a wider range during NV mode and a high concentration of

256

values in the range of 23-24 °C in the AC mode;

257

As stated by [24], the present results also suggests that occupants do not always choose

258

the use the artificial conditioning all year round to maintain a constant indoor

259

temperature. It was observed that when outdoor conditions are acceptable, occupants

260

tend to accept them passively adjusting the windows opening and shade devices; this

261

behaviour in mixed-mode buildings is promising in terms of energy savings.

262

AC C

EP

250

ACCEPTED MANUSCRIPT

RI PT

24

M AN U

SC

Figure 4. Relation between outdoor air temperature and the standard effective temperatures (SET) in the naturally ventilated mode

TE D

Figure 5. Relation between outdoor air temperature and the standard effective temperatures (SET) in the air-conditioned mode

A remarkable feature in figures 4 and 5 is the limit of maximum values: 1) when the

264

maximum outdoor temperature exceeds 25 °C, the AC was in use (Figure 5); and when

265

maximum outdoor temperature is below 24°C, the NV was in use (Figure 4). It is

266

interesting to note in Figure 5 that the AC was in use even when the daily mean

267

temperature was close to 20 °C. These result may suggests that during the cold months,

268

air-conditioning use in MM buildings may be a direct consequence of the external peaks

269

in temperature along with the building internal load (which usually occurs during the

270

afternoon as a consequence of solar radiation influence), rather than the daily mean

271

temperature. A daily mean temperature close to 20 °C during winter periods in

272

Florianópolis may be strongly influenced by the beginning and end of day temperatures

273

(especially those between 6pm to 7am) rather than the peak temperatures. A similar

274

condition is observed in Figure 4 (daily mean temperature close to 20 °C), but in this

275

case, the external maximum temperature of 24 °C was not enough to lead the occupants

276

to trigger the AC.

AC C

EP

263

ACCEPTED MANUSCRIPT 3.2. Thermal comfort in mixed-mode buildings

278

Initially, the thermal sensation data were analysed considering the two operation modes

279

in Bd 2 and 3: a) AC and b) NV. Figure 6 summarizes the mean values for the thermal

280

sensation votes with a 95% confidence interval for the sample, represented on a seven-

281

point scale (ranging from cold (-3) through neutral (0) to hot (+3)) and related to SET

282

binned at 1 °C intervals. According to Figure 6, it can be assumed that thermal

283

sensation was strongly dependent on the operation mode. In AC mode thermal

284

sensations were relatively unaffected by standard effective temperature from 21 through

285

to 27°C, whereas the same span of temperatures in NV mode was associated with a

286

steady increase in thermal sensation.

EP

TE D

M AN U

SC

RI PT

277

AC C

Figure 6. Differences in the mean thermal sensation votes for MM buildings under the two different modes of operation

287

Figure 7 summarizes the thermal preference votes of occupants in the mixed-mode

288

buildings sorted according to ventilation mode - NV and AC mode, focusing on a

289

common interval for a SET of 22-26 °C. Requests for cooler temperatures increased

290

significantly once standard effective temperature increased to 25 °C and 26 °C, but

291

frequency of requests for cooler temperatures were more than twice as prevalent in NV

292

mode as they were in AC mode in this same 20~26 oC temperature band. Even when the

293

outdoor mean temperature during the daytime is higher in the AC mode, a substantial

294

minority preference is still for a warmer environment, which could be associated with

ACCEPTED MANUSCRIPT the season and comfort expectations inside buildings. The best SET scenario (highest

296

percentage for ‘no change’) occurred at 24 °C in the NV mode, and 25 °C in the AC

297

mode (80% in both cases).

M AN U

SC

RI PT

295

a)

b)

Figure 7. Thermal preference changes according with binned SET and the operation mode: NV on the left (a) and AC on the right (b)

Based on the thermal preference votes, more specifically the preference for a “cooler

299

environment” or a “warmer environment”, the PROBIT regression model was used to

300

estimate the lowest probability of obtaining thermal preference votes for a cooler or

301

warmer environment and, consequently, the most likely temperatures in which a

302

preference for no temperature change will be expressed. This analytical method has

303

been previously used by Fanger [30] for the PPD index, so the intersection point

304

between the warmer and cooler curves from the probit regression of Figure 8 was

305

assumed to be the representative value of the preferred temperature for both of the

306

operation modes: NV mode on the left and AC mode on the right. According to the

307

results presented in Figure 8, the preferred SET temperature for the NV mode is close to

308

24 °C, with 12% of occupants preferring different temperatures, whereas for the AC

309

mode the preferred standard effective temperature was close to 26 °C with only 10%

310

preferring a change in temperature (for the PROBIT model the significance level for the

311

coefficient and intercept is p < 0.001).

AC C

EP

TE D

298

ACCEPTED MANUSCRIPT Warmer Cooler

SC

RI PT

Warmer Cooler

Figure 8. Preferred standard effective temperatures during the NV mode vs. AC mode

These differences in the thermal sensation and preferred temperature are consistent with

313

studies conducted in Australia [31] and China [20], which supports the distinction of

314

thermal comfort responses of occupants based on contextual factors such space

315

conditioning mode. According to the cited authors, the AC and NV modes prompt

316

slightly different Actual Mean Vote (AMV thermal sensation) to the identical

317

temperatures. In Sydney, the neutral temperature for NV was lower than that for AC,

318

while in Shenzhen occupants tended to accept a wider range of indoor thermal

319

conditions and were more likely to report a neutral thermal sensation. In this study, the

320

frequency of a neutral thermal sensation was slightly higher during the AC mode (55%

321

vs. 52%, respectively) with an AMV of -0.32. On the other hand, votes within the

322

thermal comfort zone (between -1 and +1) were slightly higher during the NV mode

323

compared to AC (94% vs. 91%, respectively), and the AMV during this operation mode

324

was 0.08. The discomfort rate was also higher during the AC mode compared to the NV

325

mode (10% vs. 5%, respectively).

326

Air movement acceptability was also analysed through the questionnaire item - “right

327

now, how do you classify the air movement in your space?” Figure 9 presents the air

328

movement acceptability votes in relation to SET binned values separately for the two

329

ventilation modes in operation, NV on the left (a) and AC on the right (b). On

330

comparing the two graphs, it can be observed that air movement was more unacceptable

331

in the NV mode, especially the 25°C and 26°C SET bins). During the AC mode, air

332

movement acceptability reached 100% for the 25°C and 26°C bars.

AC C

EP

TE D

M AN U

312

b)

SC

a)

RI PT

ACCEPTED MANUSCRIPT

M AN U

Figure 9. Air movement acceptability for the two modes of operation according to the binned values of SET: a) NV mode and b) AC mode

Figure 10 describes the air movement preference through the questionnaire item “Right

334

now, how would you classify the air movement in your space?” The response options

335

were: 1) I would prefer more air movement; 2) I would prefer no change, and 3) I would

336

prefer less air movement. Any option other than “I would prefer no change” was

337

considered as an expression of dissatisfaction. The results showed that the preference

338

votes tended toward “more air movement” in the NV mode, reaching more than 20% of

339

dissatisfaction at all SET values. In the AC mode, the dissatisfaction votes had a

340

counterintuitive relationship with SET in that the percentage of occupants requesting

341

more air movement generally decreased in warmer environments, as expressed in SET.

342

In Figure 10 there is a predominant preference for higher air velocities during the

343

operation of NV, but this preference was lower during AC mode at the same intervals of

344

SET (which uses the air speed as an input parameter), with mean values for the two

345

modes being similar.

EP

AC C

346

TE D

333

b)

SC

a)

RI PT

ACCEPTED MANUSCRIPT

M AN U

Figure 10. Air movement preference for the two modes of operation: a) NV mode and b) AC mode

3.3. General comparison between MM and FAC buildings

348

A general comparison between the mixed-mode buildings and the fully-air conditioned

349

building results was carried out to determine whether there is a difference between the

350

two types of buildings in terms of thermal performance using the perception, comfort

351

and acceptability votes, along with air movement acceptability. Figures 11a and 11b

352

illustrate the relationship between thermal sensation and thermal preference according

353

to ventilation type. It can be observed that the neutral sensation resulted in 95% of “no

354

change” votes in both cases. In the FAC building, none of the occupants considered the

355

thermal environment to be too hot (+3), but in both types of buildings the percentage of

356

dissatisfied occupants wanting a cooler or a warmer environment when thermal

357

sensation reached +3 or -3 was 100%. However, it is interesting to note that -2, -1 and

358

+2 and +1 can generate different preferences when compared. Firstly, the “cold”

359

condition (-2) resulted in a higher preference for a warmer environment in mixed-mode

360

buildings when compared to the FAC building (77% vs. 96%, respectively); and,

361

secondly, the “slightly warm” condition (+1) resulted in a higher preference for a cooler

362

environment in FAC building when compared to the other +1 bars from MM building

363

(72% compared to under 50%). These results suggest that the slightly warm condition

364

may be more acceptable in mixed-mode buildings than in fully-air conditioning

365

buildings (no change preference is 51% in MM building and 26% in FAC building

366

considering the +1 interval). An interesting conclusion from a study conducted in a hot-

AC C

EP

TE D

347

ACCEPTED MANUSCRIPT humid area of China considering naturally ventilated environments [32] is that subjects

368

are more acclimated and tolerable with hot and humid environments and more

369

uncomfortable and intolerable with cold environments. This assumption can also be

370

delineated for the results found in mixed-mode buildings (Figure 11a).

M AN U

SC

RI PT

367

a)

b)

Figure 11. Relationship between thermal sensation and thermal preference in: a) Mixed-mode buildings and b) FAC building

Figure 12 demonstrates the relation between the thermal sensation and acceptability

372

votes in mixed-mode and the fully-air conditioning building. Once again, differences

373

between the two types of buildings can be observed either side of the neutral thermal

374

sensation. In mixed-mode buildings, the “slightly cool” sensation was associated with

375

only ~9% calls for temperature change, while in HVAC buildings requests for different

376

temperatures were twice as frequent at this same slightly cool thermal sensation.. This

377

difference in the preferred change percentages can also be observed for the -2 sensation

378

bar (cool) and the +2 bar (warm), which suggests that in FAC buildings the cold and hot

379

sensations can be considered as uncomfortable in a higher percentage of cases when

380

compared to mixed-buildings, where occupants can control the windows or the air

381

conditioner units.

AC C

EP

TE D

371

a)

b)

RI PT

ACCEPTED MANUSCRIPT

SC

Figure 12. Relationship between thermal sensation and thermal preference in: a) Mixedmode buildings and b) FAC building

The results indicate that there is a high expectation to remain in a neutral condition in

383

FAC buildings, which is not the case in a mixed-mode building. Mixed-mode occupants

384

seem to accept greater fluctuations in their thermal response to the indoor environment,

385

even at ±2, which could be connected to the “forgiveness factor” discussed by Deuble

386

and de Dear [14] or even consensus issues about turn on/off the AC in a heterogeneous

387

group of people sharing the same indoor space.

388

Figure 12 bins the binary discomfort votes into the seven-points of the thermal sensation

389

scale for both MM and FAC buildings, and Figure 13 shows the overall distribution of

390

thermal acceptability votes in these two building types, binned by SET. The results

391

indicate that both MM and FAC buildings achieved relatively high levels of overall

392

thermal acceptability (neither type exceeded 20% of threshold commonly reported in

393

office building field studies – Arens et al [33]). However, the difference between the

394

thermal acceptability for these Brazilian MM and FAC buildings is clear (6~7%) and

395

significant (p < 0.05) in the cooler temperatures of Figure 13 (21°C