Building and Environment 44 (2009) 1399–1408

Contents lists available at ScienceDirect

Building and Environment

journal homepage: www.elsevier.com/locate/buildenv

Thermal comfort in office buildings: Two case studies commented

Hugo S.L.C. Hens
Laboratory of Building Physics, Department of Civil Engineering, K.U. Leuven, Kasteelpark Arenberg 40, B-3001 Leuven (Heverlee), Belgium

a r t i c l e i n f o a b s t r a c t

Article history: Air conditioning in offices has become a current practice in North Western Europe. The main reasons
Received 6 May 2008 for that are high internal loads, solar gains and increased comfort expectations. Hence, the move away
Received in revised form 9 July 2008 from the naturally ventilated cellular office increased thermal comfort complaints.
Accepted 25 July 2008
The paper presents two cases. In both the results of a comfort enquiry are compared with measurements.
The enquiries gave numbers of dissatisfied at a PMV zero that were much higher than the standard PMV/
Keywords:
PPD curve does. Measurements instead showed that in one of the two offices only comfort complaints
Thermal comfort
could be expected in summer. But even then, the enquired severity of complaints could not be related to
Office buildings
Enquiries the measured data.
Comfort measurements Several hypotheses are forwarded to explain the results. Individuals interpret the 3 to þ3 scale for
PPD/PMV methodology thermal sensation differently, which has a direct impact on the number of dissatisfied. The standard
curve further-on is a most significant mean of thousands of steady state comfort votes under well-
controlled conditions while an on site enquiry involves much smaller numbers of people. These have
a clear expectation: an improvement of comfort condition, thanks to the study. For that reason they may
exaggerate their complaints when enquired. And finally, an alternative PMV versus PPD curve, published
in literature, shows more people complaining at a given PMV than the standard curve forwards.
Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction comfort perception. These tests resulted in the definition of

For decades, office buildings in North Western Europe had no air
conditioning. The reference typology was the cellular office with
central heating, maximal usage of daylight, operable windows and
low internal loads. From the 1960s onward, deep landscape offices
with fully glazed facades and one employee per 10–15 m2 floor
surface gained popularity. That increased the need for artificial
lighting, turning visual comfort into an important internal load. In
the 1980s, personal computers replaced pen and typewriter, i.e.
introduced an additional load, rated at one extra employee per
employee. That doubled the heat they release. Even in mild
climates, these extra loads could no longer be removed by venti-
lation only. The answer was full air conditioning. However, water-
based systems also move large amounts of cooled air through these
landscape offices, increasing the risk on overall thermal discomfort
and draft complaints.
That situation surely was a driver behind the intense thermal
comfort research deployed since the 1960s. Anyhow, prior to that,
thermal comfort was already a subject of research in Europe and
USA. In the late 1920s, ASHVE conducted a research project on
equivalent environments, i.e. environments with different dry bulb
and wet bulb temperatures and different air speed but identical
E-mail address: hugo.hens@bwk.kuleuven.ac.be
‘effective’ temperature, i.e. the dry bulb temperature in an equiv-
alent environment with relative humidity 100% and air speed zero
(research referenced in Ref. [1]). In 1931, the ASHVE results were
reinterpreted by the French HVAC-engineer Missenard [2], who
stated that the tests overlooked an important environmental
parameter: the radiant temperature. He therefore advanced a more
general definition of ‘effective’ temperature that included the
radiant temperature, which was assumed to be equally important
as the dry bulb temperature, and called that generalised effective
temperature the resulting temperature. His remarks drove ASHVE
to redo the testing.
In Germany, comfort research was mainly conducted by physi-
ologists, whose interest focussed on skin temperature and heat flux
at the skin surface as indicators for the thermal comfort humans
experience as exothermal and homoeothermic creatures under
steady state and dynamic conditions [3,4]. Their studies underlined
the importance of the heat flux at the skin surface. As that flux
depends on the convective and radiant heat exchanges with the
environment, the results confirmed the correctness of using
resulting temperatures.
At the end of the 1960s Fanger [5] turned the physiological
approach into an engineering tool by linking global satisfaction
with the thermal environment to a steady state balance zero
between the metabolic heat produced and the total heat a clothed
human loses to the environment, using average skin temperature

0360-1323/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.buildenv.2008.07.020
1400 H.S.L.C. Hens / Building and Environment 44 (2009) 1399–1408

100

90

Predicted Percentage of Dissatisfied (PPD)

80

70

60

50

Fig. 1. Case 1: a view of the trading room.
and sweating as control quantities. The degree of discomfort,
expressed in terms of a vote on a scale from 3 to þ3 with 0 as
neutral, 1 as somewhat too cold, 2 as too cold and 3 as much
too cold, 1 as somewhat too warm, 2 as too warm and 3 as much too
warm, was coupled to the imbalance between metabolic heat rate
and heat rate exchanged. The diversity between humans got its
expression in a PMV/PPD approach that linked a predicted
percentage of dissatisfied (PPD) to a predicted mean vote (PMV).
Dissatisfied with the given environmental conditions were those
voting 2 or 3 and 2 or 3.
The PMV/PPD curve given in ISO standard 7730 [6] and ASHRAE
standard 55-2004 [7] is based on numerous comfort evaluations
under controlled steady state conditions using thousands of
randomly chosen subjects. Everyone satisfied came out as being
statistically impossible. Even at a predicted mean vote zero, still 5%
of the tested individuals were dissatisfied. That changed the
comfort approach into a discussion on acceptable percentages of
dissatisfied. ASHRAE went for 80% acceptability while the EN-
standards forward 90% acceptability as the objective.
From the 1980s on the PMV/PPD approach was questioned. One
element of criticism was that it overlooked the adaptive principle.
To some extent, the PMV/PPD approach takes it into account as
diversity between humans is considered. However, based on the
definition of comfort as a state of mind and using field results
mainly collected in warm climates among people working in
naturally ventilated office buildings, Humphreys, Nicol, de Dear,
Brager, Baker and Standeven stated that comfort was much more
adaptive [8–10]. Though the adaptive actions proposed vary from
Constant, PD 40
as enquired
30
Cold, PD as
enquired 20
Warm, PD as 10
enquired
0
-4 -2 0 2 4
Predicted Mean Vote (PMV)
Fig. 3. Case 1: comparing the percentages of dissatisfied for the mean votes enquired
with the standard PMV/PPD curve.
quite easy (turning up or down the thermostat, more or less
activity, changing posture [11] and clothing, opening windows) to
possible in theory but not doable in practice (building a new house
or emigrating), the adaptive temperature model proposed gave
better fits than Fanger’s PMV/PPD model in naturally ventilated
buildings. It, however, hardly differed from the PMV/PPD predic-
tions in air conditioned environments.
Another element of discussion was the split Fanger introduced
between global comfort and local discomfort by draft, air temper-
ature gradient between head and feet, horizontal and vertical
radiant imbalances and feet temperature. A too simple modelling of
the human body as a heat exchanging homoeothermic system with
constant core temperature and of the control system that varies the
below-skin blood flow and sweat production depending on the
heat or cold noted, figured as main reasons for that. Several authors
proposed upgrades of the human model with the intention to come
to one overall comfort picture [12–14].
Also transient environmental conditions returned as a subject of
research. Measurements indicated that the mean skin temperature,
the head core temperature and the rate of change of skin temper-
ature are the governing factors in human thermal sensation in such
a case [15]. Zhang et al., for example, developed an advanced

40 25 40

20
30 30
Number

Number

Number

15
20 20
10

10 10
5

0 0 0
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
Vote Vote Vote

Fig. 2. Case 1: votes on the Fanger scale, left for those calling the thermal environment invariable, in the middle and right for those calling the thermal environment different from
day to day. The vertical axis gives the number of votes and the horizontal axis the vote.
 H.S.L.C. Hens / Building and Environment 44 (2009) 1399–1408 1401

40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1
2
30 3
4
Number

5
20
6
7
10 8
9
10
0
0 1 2 3
Vote

Fig. 4. Case 1: enquired draft votes (number of votes on the vertical axis, the vote on the horizontal axis) and spatial distribution of the votes over the trading room (the lightest
colour stands for no draft felt, the somewhat darker colour for some draft felt, the dark colour for drafty and the really dark colour for very drafty).

thermal model of the clothed human body that allows comfort
evaluation in quite realistic transient environments [16].
All theories and standards, however, could not avoid that
thermal comfort complaints emerged in larger numbers than
before in the air conditioned office buildings. That is illustrated in
the next paragraphs by two case studies. The first concerns the
trading room at the head-quarter of a large bank company. The
second involves a typical landscape office in a new high standard
office building.
2. Methods used
among employees working for a retail company with several
megastores in Belgium and the Netherlands [17].
The four comfort parameters were monitored using a Brüel &
Kjaer comfort meter with data storage on hard disk. That high
precision apparatus measures the dry bulb temperature, two semi-
spherical radiant temperatures, air speed and the dew point of the
air at the position it is standing. In addition, in the first case, also
Hobo sensors were used to monitor the dry bulb temperature and
relative humidity indoors. These may store three months of data,
measured every 10 min with a precision of 0.3  C for the dry bulb
temperature and 3% for relative humidity.

Basic philosophy applied is that comfort complaint analysis 3. The two cases
should be based on a double approach: at one side by enquiring all
employees involved, at the other side by a careful, at least working 3.1. First case
day long monitoring of the four comfort parameters at neck height
of a sitting person (dry bulb temperature, radiant temperature, 3.1.1. Building, office and HVAC-system
relative humidity, and air velocity) in different locations of the The office building involved is a large multi-story construction,
office floor. Neck height, in the case being 1.2 m above floor level, erected in Brussels, Belgium, during the 1990s. The trading room,
counts for the draft sensitivity of a seated person. with 166 employees working on a floor area of 840 m2, figures as
An example of the enquiry form is added. The package also a typical landscape office. The high occupation density with 5 m2
contained a page explaining the scales. The form evaluates overall per employee, three computer screens per employee (Fig. 1) and
thermal comfort using the Fanger scale from 3 to þ3, commented high illumination levels needed resulted in a very high internal
above. Draft is enquired using a scale from 0 to 3, with 0 for no draft load, up to 138 W/m2 on an average. The intense cooling load was
felt, 1 for some draft felt, 2 for drafty and 3 for very drafty. Also local removed with the aid of active chilled beams. Adaptive opportu-
thermal discomfort at feet, under-legs, hands and head is ques- nities in the office were very limited, as employees had a dressing
tioned using a Fanger-like scale from 3 to þ3. As many employees code to follow, while working with three computer screens made it
may have mixed complaints about global thermal comfort, a choice difficult to change posture. Temperatures could neither be
is offered between ‘constant’ and ‘variable’ global comfort with controlled individually, nor was it possible to open windows. Once
‘constant’ pointing to ‘each day the same’ and ‘variable’ to ‘different the trading room was in use, thermal comfort complaints of such
from day to day’. The form finally focuses on health complaints that intensity and persistency surfaced that the HVAC-design and con-
may be linked to Sick Building Syndrome. The form as it is now was tracting parties got a demand to change the system. Before any
developed by the Laboratory of Building Physics of the K.U. Leuven decision was taken, all parties required strong and decisive argu-
in the mid-1990s and first used for a large-scale comfort enquiry ments that justified the change. So, an in-depth comfort study was
initiated in June 2002, consisting of an enquiry, followed by an
extended measuring campaign [18].
Table 1
Case 1: health complaints. 3.1.2. Enquiry
Complaints Total number Answering % of the About 102 of the 166 employees filled in the form, resulting in
of answering ‘Yes’ total number a response ratio of 61.4% which made the enquiry quite represen-
individuals of answering tative. Thirty-three subjects judged the overall thermal environ-
individuals
ment as being invariable, while 69 estimated it as being different
Burning eyes 102 26 26
from day to day. Fig. 2 summarizes the votes given.
Headache 102 45 44
Blocked nose 102 38 37 Looking at the mean vote, the 33 employees calling the envi-
Coughing 102 24 20 ronment invariable ended up with a zero. The number of dissatis-
Dry throat 102 39 38 fied, however, totaled 11 individuals or 33% and not the predicted
Disappears during 77 64 83 5% or 2 individuals on 33 of the standard PMV/PPD curve. Those
the weekend?
judging the environment different from day to day ended up with
1402 H.S.L.C. Hens / Building and Environment 44 (2009) 1399–1408

Hobo's B&K comfort meters.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1 6 7 1 MA di
2 9 2 ma
3 11 13 3 do DO
4 4
5 10 12 5 DI VR
6 14 6 W
7 15 7 wo
8 08
9 8 9 vr
10 10

Fig. 5. Case 1: location of the Hobo T/RH sensors (left, numbers 6–16) and the two B&K comfort meters (right, letters ma, di, woe, do, vrij) in the trading room.

a mean vote 1.5 for cold and 2.1 for warm. As Fig. 3 shows, the
number of dissatisfied in both cases approached the predicted
value of the standard curve very closely, which was a remarkable
but fortuitous fact.
Fig. 4 summarizes the results for draft, which, as said, was
judged using a 4-vote scale: 0 for no draft, 1 for some draft, 2 for
drafty and 3 for very drafty. The mean vote noted was 1.2, i.e. some
draft, anyhow with 33 of the 102 employees complaining about
drafty and very drafty. The spatial distribution of the votes over the
trading room showed a diffuse pattern, with no draft and very
drafty side by side. Clearly, the thermal environment in the office
was far from uniform. Could that be the result of a too strong
downward jet of cool air produced by the active chilled beams?
Feet, under-legs, hands and head gave quite neutral mean votes,
ranging from 0.34 for the head over 0.18 for the under-legs to 0.14
for the hands. The standard deviation, however, was quite large for
3.1.3. Measurements and results
Between June 10 and June 14, 2002, 10 Hobo temperature and
relative humidity sensors (T/RH) logged the dry bulb temperature
and relative humidity, 1.2 m above floor level, in the positions 8–15,
distributed over the trading room, see Fig. 5. At the same time two
B&K comfort meters moved between 10 other positions, see Fig. 5.
In every position, dry bulb temperature, the two semi-spherical
radiant temperatures, air velocity and the dew point were regis-
tered during one working day, 1.2 m above floor level (Fig. 5).
The weather outdoors during the week the comfort measure-
ments were run is summarized in Table 2. The data reflect a typical
North Western European summer week: cool and unstable with
sun, rain and some wind.
Weekly mean dry bulb temperature (°C)

the head, resulting in 51 of the employees complaining of warm or 25

too warm or cold or too cold at the head. That, of course, seems to
24
be linked to the draft response.
Table 1 gives an overview of the answers noted on the questions
23
about health. Seventy-two employees complained about one or
more of the symptoms mentioned. Quite remarkable is the answer 22
on the final question: do the symptoms disappear during the
weekend. Sixty-four of the 72 complaining are affirmative. Is that 21
an indication for SBS? Perhaps. It may also reflect the stress linked
to the work as a trader. 20

19
Table 2
Case 1: the weather outdoors during the measuring week (daily means). 18
0 1 2 3
Daily Daily mean Daily mean Total Daily mean
mean air relative solar precipitation wind speed Voted in the enquiry (>0)
temperature humidity radiation (mm/day) (m/s)
Weekly mean dry bulb temperature (°C)

( C) (%) (W/m2) 25
Mon., 10/6/2002 13.1 85.0 153 4.8 1.8
Tue., 11/6/2002 14.1 82.0 170 4.8 1.4 24
Wed., 12/6/2002 16.2 86.5 135 6.4 1.9
Thurs., 13/6/2002 15.4 81.8 67 0.2 1.5 23
Fri., 14/6/2002 20.0 78.1 176 6.8 1.4
22

Table 3 21
Case 1, position 6: measured indoor temperature and relative humidity during
working hours: mean, maximum, minimum, standard deviation, enquiry vote by the
employee sitting closest-by.
20

Position 6, enquiry vote: Dry bulb temperature ( C) Relative humidity (%) 19

much too warm, much
Mean Max. Min. Std Mean Max. Min. Std
too cold, drafty
18
Mon., 10/6/2002 23.3 24.0 22.4 0.3 44.2 45.8 43.2 0.5 -3 -2 -1 0
Tue., 11/6/2002 22.9 23.3 22.7 0.2 45.8 46.7 44.0 0.6
Wed., 12/6/2002 23.2 23.5 22.6 0.2 46.1 46.7 45.3 0.3
Voted in the enquiry (<0)
Thurs., 13/6/2002 22.8 23.0 22.2 0.2 46.1 46.7 45.3 0.3
Fig. 6. Case 1: the enquired vote by the employee sitting closest-by and the measured
Fri., 14/6/2002 23.3 23.6 22.7 0.2 46.0 47.6 44.9 0.5
weekly mean inside air temperature during working hours.
 H.S.L.C. Hens / Building and Environment 44 (2009) 1399–1408 1403

3 are very stable, with hardly any up’s and down’s. That proves that
looking to daily and even weekly means has sense.
2 When the closest-by votes for all 10 positions are linked to the
Daily mean PMV (Fanger)

measured weekly mean dry bulb temperatures during working

hours then, as Fig. 6 shows, no pattern in a sense a vote 2 or 3
1
relating to higher and 2 and 3 to lower values could be detected.
One of course may argue that not the dry bulb temperature but the
0
resulting or operative temperature is the comfort parameter to be
considered. Anyhow, looking at the office lay-out, the operative
-1 temperature at the T/RH logger locations could not differ substan-
tially from the air temperature.
-2 In an analogous way, Fig. 7 links the daily mean of PMVs for
a summer Clo-value between 0.7 and 1 and a metabolism between
-3 1.2 and 1.4 Met, as calculated from the dry bulb temperature, mean
0 1 2 3 radiant temperature, air velocities and dew points logged by the
Voted in the enquiry (>0) B&K comfort meters during working hours, to the enquired votes by
the employees sitting closest-by. Metabolic rate used was taken
3
from ANSI/ASHRAE standard 55-2004 [6], while the Clo-value was
calculated based on the clothing habits observed in the trading
2
Daily mean PMV (Fanger)

room. The value of 1.4 Met accounted for the stressed working
conditions. Calculations in turn were based on the methodology
1 given in ANSI/ASHRAE standard 55-2004. Also here, no meaningful
correlation was found.
0 Fig. 8 focuses on the mean draft rate during working hours. Even
here, no correlation emerged between the enquired votes by the
-1 employees sitting closest-by and the values calculated from the dry
bulb temperature, air speed and turbulence intensity data as logged
by the B&K comfort meters, using the methodology given in ANSI/
-2
ASHRAE standard 55-2004. Of course, draft complaints may relate
to the extremes sensed, also, no reason for any complaint could be
-3
-3 -2 -1 0 detected as in none of the 10 positions the measured instantaneous
draft rate passed 15%!
Voted in the enquiry (<0)
Clearly, the measured data did not back the complaints.
Fig. 7. Case 1: overall comfort, relation between the enquired vote by the employee
sitting closest-by and the measured predicted mean vote during working hours. 3.2. Second case

3.2.1. Building, office and HVAC-system

Table 3 gives a glance of the conditions indoors by listing the
daily mean dry bulb temperature and relative humidity as logged
by the Hobo during working hours in position 6, together with the
maximum and minimum values and the standard deviation. The
enquired vote by the employee sitting closest-by is listed in the
upper corner left of that table. Apparently, the indoor conditions
15
12
DR (Fanger)
The office building involved, situated in Leuven, Belgium, was
designed and constructed shortly after the turn of the century, with
the landscape office under scrutiny located on the first floor. The
number of employees totaled 45. HVAC included a variable refrig-
erant volume heat pump system, complemented with electrical
base board heating and separate ducting for the preconditioned
ventilation air. Windows could not be opened and adoption in
dressing code for males was restricted to taking of the jacket or not.
Female employees could put-on a pull-over if needed. Persisting
complaints about thermal comfort and air quality motivated the
principal to order an overall thermal comfort evaluation. As in the
first case, data gathering consisted of an enquiry during the month
of September 2004, followed by an intense measuring campaign
during 10 consecutive winter days, from December 1 to December
12, 2004, and 10 consecutive summer days, from June 20 to July 1,

9
2005 [19].

3.2.2. Enquiry
6
The enquiry was conceived the same way as it was in the first
case. Thirty-six of the 45 employees filled in the form, a response
ratio of 80%, what made the results quite representative for the
3
whole of the occupants. Twenty subjects judged the overall thermal
environment as invariable from day to day, while 16 called it
different, ranging from cold on one day to warm the other day. Fig. 9
0 summarizes the votes. Looking at the mean vote, it can be seen that
0 1 2 3
those calling the environment invariable gave a 0.05. The number
Voted in the enquiry (>0)
of dissatisfied, however, totaled 4 on 20, i.e. 25% and not 1 on 20,
Fig. 8. Case 1: relation between the enquired draft vote by the employee sitting the 5% the standard PMV/PPD curve forwards (Fig. 10). Those
closest-by and the measured DR-value during working hours. judging the environment different from day to day represented
1404 H.S.L.C. Hens / Building and Environment 44 (2009) 1399–1408

10 10 10

8 8 8

Number
6

Number
6
Number

4 4 4

2 2 2

0 0 0
-3 -2 -1 0 1 2 3 1 2 3 4 5 6 7 -3 -2 -1 0 1 2 3
Vote Vote Vote

Fig. 9. Case 2: votes on the Fanger scale, left for those calling the thermal environment invariable, in the middle and right for those calling the thermal environment different from
day to day. The vertical axis gives the number of votes and the horizontal axis the vote.

Draft was judged using the 4 votes’ scale from 0 to 3, with 0 for
100
no draft felt, 1 for some draft felt, 2 for drafty and 3 for very drafty.
Fig. 11 summarizes the results. The average vote noted was 1.16, i.e.
90
some draft felt. As in case 1, the spatial distribution was rather
diffuse, with no draft felt and very drafty side by side. Yet, the votes
Predicted percentage of dissatisfied (PPD)

80
given correlated quite well with overall coldness felt.
70
Vdraft ¼ 0:12  0:63Vcoldness r 2 ¼ 0:58; F ¼ 39:4 (1)
60 with Vdraft being the draft vote and Vcoldness the somewhat cold, cold
or too cold vote. Such correlation backs the idea that subjects may
50 mix up draft with overall coldness and vice versa.
The health complaints noted are summarized in Table 4. The
40 average number of subjects complaining is higher than in case 1,
while 21 of the 24 subjects answered the question on disappear-
30 ance of all symptoms during the weekends with a ‘Yes’. Is that an
indication for SBS? Also here, it is difficult to confirm. Stress
20 because of the job to be done may be another cause.

10 3.2.3. Measurements and results

-3 -2
Predicted mean vote (PMV)
Fig. 10. Case 2: comparing the percentages of dissatisfied for the mean votes enquired
with the standard PMV/PPD curve.
a mean of 1.35 for the cold days and 1.56 for the warm days. At the
same time, as Fig. 10 shows, the number of dissatisfied was very
close to the predicted percentage of dissatisfied of the standard
PMV/PPD curve.
From December 1 to December 12, 2004 and from June 20 to
0 July 1, 2005, a B&K comfort meter moved over the landscape office,
-1 0 1 2 3 taking 10 positions. In each position, the apparatus registered dry
bulb temperature, the two semi-spherical radiant temperatures, air
speed and relative humidity during one working day 1.2 m above
floor level, the height where sitting employees have their neck
exposed (Fig. 12).
Table 5 lists the outside conditions during the 10 consecutive
winter and 10 consecutive summer working days. The winter
period was rather normal for a mild climate, while in the summer
period warmer days alternated with some colder days. Only the
first day, however, was really ‘warm’. The others were normal for

100
G2 J21 L21 Q21
80
G16
Percentage

S14 V15 X14 Y14 AB13 AC13 AF13 AG13

60
C11 S11 X12 Y12 AB11 AC11 AF12 AG12
40 M8 N8 Q8 R8 U8 V8 Y8 Z8 AC8 AD8 AG8 AH8
C8 M6 N6 Q6 R6 U6 V6 Y6 Z6 AC6 AD6 AG6 AH6
20 C2 G3a G3b K2 L2 O2 P2 S2 T2 W2 AA2 AB2 AE2 AF2 AI2

0
0 1 2 3
Vote
Fig. 11. Case 2: enquired draft votes (number of votes on the vertical axis, the vote on the horizontal axis) and spatial distribution of the votes and spatial distribution over the
landscape office (the lightest colour stands for no draft felt, the somewhat darker colour for some draft felt, the dark colour for drafty and the really dark colour for very drafty).
 H.S.L.C. Hens / Building and Environment 44 (2009) 1399–1408 1405

Table 4 Table 5
Case 2: health complaints. Case 2: the weather outdoors during the 10 winter and 10 summer days (daily
values).
Complaints Total number Answering % of the total
of answering ‘Yes’ number of answering Daily Daily Daily Daily Daily
individuals individuals mean air mean mean mean mean
Burning eyes 36 10 29 temperature relative solar wind wind
Headache 36 15 43 ( C) humidity radiation speed direction
Blocked nose 36 9 26 (%) (W/m2) (m/s) ( )
Coughing 36 13 37 Winter
Dry throat 36 10 29 Mon., 1/12/2004 3.3 97.8 70 0.8 222a
Disappears during 24 21 88 Tue., 2/12/2004 2.3 96.3 240 0.5 172
the weekend? Wed., 3/12/2004 0.7 94.1 300 0.38 152
Thurs., 4/12/2004 6.1 97.1 240 1.25 176
Fri., 5/12/2004 3.7 100 110 0.83 181
Mon., 8/12/2004 6.3 90.9 130 1.07 223
a mild climate. Fig. 13 shows the dry bulb temperature and the two Tue., 9/12/2004 0.6 100 100 1.17 220
semi-spherical radiant temperatures as measured on June 20, 2005. Wed., 10/12/2004 0.2 100 110 0.92 195
Although the three temperatures are very stable over the day, the Thurs., 11/12/2004 1.1 91.4 50 0.97 192
Fri., 12/12/2004 0.7 99.9 110 0.66 225
effect of solar gains around noon is slightly visible.
Fig. 14 links the enquired votes by the employees sitting closest- Summer
by to the measured mean PMV during the winter 8 h working day, Mon., 20/6/2005 25.5 55.4 310 0.83 264
as calculated for a metabolic rate 1.2 Met and a Clo-value between Tue., 21/6/2005 22.3 67.1 245 0.65 316
0.8 and 1.2. No correlation could be found. Fig. 15 repeats the graph Wed., 22/6/2005 21.8 60.0 290 0.94 23
Thurs., 23/6/2005 24.6 60.3 301 0.65 300
for the summer days, now for a metabolic rate 1.1–1.2 Met and
Fri., 24/6/2005 26.3 58.3 284 0.77 268
a Clo-value between 0.7 and 0.9. The measured summer mean of Mon., 27/6/2005 22.8 71.8 217 1.04 324
PMVs during working hours confirms that there is a comfort Tue., 28/6/2005 20.2 64.2 231 1.5 40
problem, though the enquired votes by the subjects sitting closest- Wed., 29/6/2005 22.1 55.1 290 1.16 20
Thurs., 30/6/2005 22.9 50.0 260 1.99 58
by do not match the measured values. As in case 1, metabolic rate
Fri., 1/7/2005 19.7 88.1 164 1.2 68
was taken from ANSI/ASHRAE standard 55-2004, while the clo-
a
value was calculated based on the clothing habits noted in the North is zero, East is 90 , South is 180 , and West is 270 .

landscape office.
as invariable in both cases deviates substantially from the 5% given
Fig. 16 shows the maximum draft rate measured during working
by the standard PMV/PPD curve. The question is if these results for
hours in each of the 10 positions in relation to the enquired vote by
a small number of respondents could end effectively in a deviation
the employee sitting closest-by. The average DR-values looked
from the standard curve if large number of individuals were
quite reassuring but the fluctuations over the day were consider-
questioned under the same circumstances. Application of a c2 test
able, reason why the maximum measured instead of the mean is
for 2 degrees of freedom (satisfied versus dissatisfied) on the
plotted in the figure. That maximum underlines draft complaints
distribution between satisfied and dissatisfied for the 33 respon-
are real though again not correlated to the enquiry results.
dents in case 1 and the 20 respondents in case 2 shows that the
Clearly, the measured data confirm that there is a thermal
probability that the measured data should approach the standard
comfort problem in summer. Correlation with the enquired votes
PMV/PPD curve for a much larger number of respondents is close to
by the employees sitting closest-by, however, is again inexistent.
zero.

4. Statistical relevance 5. Discussion

Though the mean vote is zero in both cases, the number of In case 1, the search for the reasons why no correlation could be
dissatisfied among those judging the overall thermal environment found between the measured comfort votes and the enquired ones,

Fig. 12. Case 2: 10 positions of the B&K comfort meter in the landscape office.
1406 H.S.L.C. Hens / Building and Environment 44 (2009) 1399–1408

30 where qintern is the internal heat gain and a an experimental

constant which links the air speed the occupants experience to the
heat gains the active chilled beams have to compensate, a speed
Dry bulb temperature (°C)

25
increase of 25–36% could be expected at full occupancy. That should
have raised the measured DR-value by some 5%. But even then, no
20
backing of the overall and local thermal comfort complaints was
achieved.
15
In case 2, the enquiry results confirm the observation that draft
as perceived by the subjects enquired and their response in terms
10 of coldness are linked. At the same time the comfort problems
reflected by the answers in the enquiry are not correlated to the
5 measured predicted mean votes and draft rates, even when these
confirm that there are thermal comfort problems, at least during
0 summer.
0 500 1000 1500 2000 2500 Thermal comfort complaints by the employees are of course the
Between June 20,10.35 am and 5.00 pm result of a long lasting experience with the indoor climate on their
spot in the landscape office, while the measurements are a snap-
Semi-spherical radiant temperatures (°C)

30 shot only, recorded during the 8 h of one working day. The dots in
the Figs. 6, 7, 8, 14, 15 and 16 also relate to one or two individuals
25 only, those seated nearest-by, while the PMV/PPD approach is
based on a statistical evaluation of the comfort votes of thousands
of individuals. The possibility that the enquired vote given by an
20
individual should match the predicted mean vote is quasi
inexistent.
15 Even more, difference in thermal comfort sensitivity between
individuals is perhaps much larger than reflected in the standard
10 PMV/PPD approach. The PMV/PPD curve, proposed by Mayer of the
Fraunhofer Institut für Bauphysik, seems to confirm this [20]. The
5 graph is skewed compared to the standard PMV/PPD curve. It
shows a higher number of dissatisfied at the cold side and reaches
0 the lowest predicted percentage of dissatisfied at a PMV of 0.5, with
0 500 1000 1500 2000 2500 a value of 17%. At PMV zero, that percentage equals 25%, 5 times the
Between June 20, 10.35 am and 5.00 pm standard value of 5% and much closer to the values enquired in both
cases.
Fig. 13. Case 2: dry bulb temperature and the two semi-spherical radiant temperatures There may be additional reasons. Each individual interprets
on June 20, 2005.
thermal sensation corresponding to the votes 3, 2, 1, 0, 1, 2 and
3 differently. Such differences are levelled out in a very large group
of respondents, but may distort the enquired results in small
groups, with a direct effect on the enquired number of dissatisfied.
started by noting some facts. The occupancy in the trading room
Objectively spoken that subjectivity induces uncertainty in the
during the measuring days never passed 48% of the maximum. The
results commented, what could be judged by testing statistical
internal heat gain also did not touch 138 W/m2, as presumed, but
significance. However, objectivity in comfort voting does not exist.
remained close to 57.5–70.5 W/m2. As a consequence, the chilled
Each vote expresses a subjective appreciation and should be
beams worked at part load, diminishing the temperature difference
considered that way. As such, the results in terms of dissatisfied,
between the room and the supply air when compared to full
etc., are deterministic, which means the distribution in enquired
occupancy. Using following formula for the mean air speed:
votes found should not be extrapolated to other cases.
pffiffiffiffiffiffiffiffiffiffiffiffiffi Expectation for improvement, thanks to the study, could also
v ¼ a 3 qintern (2) influence people’s responses when filling in the enquiry. Let us give

3 3

2 2
PMV (Fanger)

PMV (Fanger)

1 1

0 0

-1 -1

-2 -2

-3 -3
0 1 2 3 -3 -2 -1 0
Voted in the enquiry (>0) Voted in the enquiry (>0)

Fig. 14. Case 2: winter overall comfort. Relation between the enquired vote by the employee sitting closest-by and the measured predicted mean vote during working hours.
 H.S.L.C. Hens / Building and Environment 44 (2009) 1399–1408 1407

3 3

2 2

PMV (Fanger)
PMV (Fanger)
1 1

0 0

-1 -1

-2 -2

-3 -3
0 1 2 3 -3 -2 -1 0
Voted in the enquiry (>0) Voted in the enquiry (>0)

Fig. 15. Case 2: summer overall comfort. Relation between the enquired vote by the employee sitting closest-by and the measured predicted mean vote during working hours.

60 60

50 50

40
DR (Fanger)

40
DR (Fanger)
30 30

20 20

10 10

0 0
0 1 2 3 0 1 2 3
Voted in the enquiry (>0) Voted in the enquiry (>0)

Fig. 16. Case 2: draft. Relation between the enquired draft vote by the employee sitting closest-by and the measured DR-value during working hours for the winter (left) and
summer (right) periods.

a negative vote in terms of too warm or too cold. That will motivate the standard PMV/PPD curve seems to underestimate the number
the company to agree with an upgrade of the indoor environmental of dissatisfied.
conditions. The overall conclusion is one should be very careful when
interpreting the results of thermal comfort campaigns. The actual
6. Conclusions standards help but should not be considered as absolute references.
The individual state of mind that expresses satisfaction with the
Two cases where the employees had persisting comfort thermal environment is too diverse for that when small groups are
complaints have been evaluated by means of a comfort enquiry and considered. The ASHRAE standard 55-2004 is right when stating
detailed measurements. In both cases, the enquiry backed the that subjective evaluation is required. On the other hand, of course,
complaints. Anyhow, the number of dissatisfied noted at a mean a reference method as given in the standards mentioned [6,7] is
vote zero was higher than the 5% from the standard curve. needed.
In none of the two cases, however, the measured mean PMVs
during working hours revealed any relationship with the votes the
Acknowledgement
subjects working close-by the logging points gave in the enquiry.
The measurements even declared case 1 problem-free, while in
We thank the companies involved for their willingness in
case 2 problems were only noted in summer, but again without
distributing and collecting the enquiries and their help during the
clear relationship with the enquiry results.
measuring campaigns.
One may argue that the standardised PMV/PPD curve stems
from thousands of individual responses, while in the two cases
commented, only 102 and 36 employees participated in the Appendix. Enquiry form
enquiry. Direct comparison between measured and enquiry results
was even restricted to the 10 subjects closest-by. With small 1 Global thermal comfort: Constant Variable
numbers, individuality colours the responses much more than with (strikethrough what does not fit)
larger numbers. Even the interpretation of the seven voting If constant
possibilities, from 3 over 0 to þ3, varies between individuals. Too 3 2 1 0 1 2 3
cold, cold, somewhat cold, neutral, somewhat warm, warm and too
warm apparently cover a different content for different people. Also
expectations may colour enquiry results. And, as Mayer showed, Put a cross below the number that fits with your experience.
1408 H.S.L.C. Hens / Building and Environment 44 (2009) 1399–1408

If variable 4 Personal remarks

The one extreme ...........................
3 2 1 0 1 2 3 ...........................
....... (add another page if necessary).
Put a cross below the number that fits with your experience.
The other extreme
3 2 1 0 1 2 3 References

[1] de Grave A. Bouwfysica. Brussels: Uitgeverij SIC; 1957. 220 pp. [in Dutch].
Put a cross below the number that fits with your experience. [2] Missenard A. Cours superieur de chauffage, ventilation et conditionnement de
l’ air. In: Livre IV. Paris: Editions Eyrolles; 1947. 168 pp. [in French].
2 Local thermal comfort [3] Cammerer JS. Wärme- und Kälteschutz in der Industrie. Berlin/Göttingen/
Heidelberg: Springer-Verlag; 1962. 497 pp. [in German].
Draft [4] Benzinger TH. The physiological basis for thermal comfort. In: Fanger PO,
0 1 2 3 Valbjorn O, editors. Indoor climate. Copenhagen: Danish Building Research
Institute; 1978. p. 441–76.
[5] Fanger PO. Thermal comfort. New York: McGraw-Hill Book Company; 1972.
Put a cross below the number that fits with your experience. [6] ASHRAE, standard 55-2004. Thermal environmental conditions for human
Temperature feeling at the lower legs and feeds occupancy; 2004.
3 2 1 0 1 2 3 [7] ISO, standard 7730. Moderate thermal environments – determination of the
PMV and PPD indices and the conditions for thermal comfort; 2005.
[8] Humphreys MA. Clothing and the outdoor microclimate in summer. Building
Put a cross below the number that fits with your experience. and Environment 1977;12:137–42.
Temperature feeling at the head [9] Baker, Standeven. Thermal comfort for free-running buildings. Energy and
Buildings 1996;23:175–82.
3 2 1 0 1 2 3 [10] ASHRAE. Field studies of thermal comfort and adaptation. In: ASHRAE Tech-
nical Data Bulletin, vol. 14, number1. Atlanta: W.S.Comstock; 1998. 156 pp.
[11] Zhu S, Kato S, Yang L. Examination of thermal adaptive effect of postural and
Put a cross below the number that fits with your experience.
positional adjustment of a seated human body exposed to spot airflow.
Temperature feeling at the hands ASHRAE Transactions 2007;113(Part 2) (CD).
3 2 1 0 1 2 3 [12] Airaksinen M, Tuomaala P, Holopainen R. Modeling human thermal comfort.
Proceedings of Clima 2007;2007:10–4. June (published as CD).
[13] Tuomalaa P, Airaksinen M, Holopainen R. A concept for utilizing a detailed
Put a cross below the number that fits with your experience. human thermal model for evaluation of thermal comfort. Proceedings of Clima
2007;2007:10–4. June (published as CD).
[14] Stolwijk JAJ. A mathematical model of physiological temperature regulation in
3 Did you suffer from the following irritations/problems during the
man. NASA contractor report, NASA CR-1855, Washington DC; 1971.
working hours in the trading room (strikethrough what does not [15] Dusan F, Lomas K, Stohrer M. First principles modelling of thermal sensation
fit): responses in steady-state and transient conditions. ASHRAE Transactions
2003;109(Part 1) (CD-ROM).
Burning eyes Yes No
[16] Zhang H, Arens E, Huizinga C. Advanced human thermal comfort model,
Headache Yes No
<http://www.cbe.berkeley.edu>.
Stuffed nose Yes No [17] Verbeeck G, Hens H. Case study on the thermal comfort in a store. In: Proceedings
Coughing Yes No of the CLIMA 2000 conference, Brussels; 30 August–2 September 1997.
Dry throat Yes No [18] LBF. Comfort study in a trading room, report 2001/15 (5); 2001 [in Dutch].
[19] LBF. Comfort problems in a landscape office, report 2004/9 (2); 2004 [in
Did the complaints disappear during the weekends? Dutch].
Yes No [20] Mayer E. Ist die bisherige Zuordnung von PMV und PPD noch richtig? Ki Lüft-
und Kältetechnik 1998;54:575–7 [in German].