Proceedings of 8th Windsor Conference: Counting the Cost of Comfort in a changing world Cumberland Lodge, Windsor, UK, 10-13 April 2014. London: Network for Comfort and Energy Use in Buildings, http://nceub.org.uk

Presenting LOBSTER, an innovative climate chamber, and the analysis of the effect of a ceiling fan on the thermal sensation and performance under summer conditions in an office-like setting

Marcel Schweiker1*, Sabine Brasche2, Maren Hawighorst1, Wolfgang Bischof2 and Andreas Wagner1

1 Karlsruhe Institute of Technology, Building Science Group, Englerstr. 7, 76131 Karlsruhe, Germany, marcel.schweiker@kit.edu 2 University Hospital Jena, Institute of Occupational, Social, Environmental Medicine and Hygiene, Department of Indoor Climatology, Bachstr. 18, 07743 Jena, Germany

Abstract

Thermal comfort studies have been performed so far either in closed climate chambers with controlled conditions or non-controlled conditions during field studies. Detailed analyses of mechanisms behind the adaptive comfort models are therefore hardly possible. This paper presents a newly constructed climate chamber in Karlsruhe (Germany) along with the complete chain from subjective experiments, via data analyses, model development and implementation into dynamic building energy simulation until the formation of a decision base for or against a renovation measure for a confined case. The objective of this experimental series conducted in summer 2013 was the analysis of the effect of a ceiling fan on the perceived thermal comfort and performance. The results suggest that it is not the control itself, which leads to a higher acceptance of increased indoor temperatures in summer, but the effectiveness of the control. For the analysis of the performance tests, only minor differences in the performance under the distinctive conditions were observed.

Keywords: Adaptive comfort, fan usage, occupant behaviour, neutral temperature, simulation

1 Introduction

The foundations of thermal comfort studies have been laid in closed climate chambers (Fanger, 1972). Today these studies are criticised for the lack of adaptive opportunities subjects had during the experiments. At the same time, the studies conducted within the development of the adaptive comfort model are field studies missing controlled conditions (deDear et al., 1997, Humphreys and Nicol, 1998). Detailed analyses of individual portions of the adaptive mechanisms are therefore hardly possible on the existing dataset.

This paper presents a newly constructed climate chamber in Karlsruhe (Germany) along with the results of a first experimental series conducted during the summer 2013. These results are implemented into a dynamic building simulation in order to form a decision base for renovation measures.

The objective of the first experimental series was the analysis of the effect of a ceiling fan on the perceived thermal comfort and mental performance. The ceiling fan was

choappl

Theand con

2 LO

TheEnvand

In oofficroomproftilteglazwinthro

In aconbeforoom

Thewithdaylclos

Witon t

sen becauselied easily d

e hypothesespreference

ditions. The

OBSTER –

e Laboratovironmental

occupant b

order to facce rooms hm has integfiles two wed; the top zing; the ondows are tiough the bui

addition to cept involvore enteringm.

e whole of th daylight light througsed.

th respect tothe experim

e it is - comduring remo

s tested were votes as weir distinctio

– an adapti

ry for Oc Research (

behaviour.

cilitate studhave one fagrated in th

windows andlight windo

opaque baluilted mechailding contr

Figure 1. F

the possibves two deceg the office

the glass suguidance frgh the uppe

o studies onmental setup

mpared to otodelling pro

re related towell as the on was base

ive test faci

ccupant Be(LOBSTER

dies on adaaçade to thehe post andd two top liows can onustrade is anically eithrol system. O

Floor plan

ility to opeentralized f

e space and

urface can brom ROMA

er part even

n occupant p - multiple

thers - a ratojects.

o differenceresults of

ed on the de

ility

ehaviour, R) was desi

aptive comfe outdoor ed beam struight window

nly be tiltedequipped w

her by pressOpening the

en one or bfloor convec

d a fan drive

be shaded bA (see Fig.n when the l

behaviour, e adaptive

ther cheap i

es in the thethe perform

egree of per

Satisfactiongned for st

fort, the twenvironmentucture madws. The wid (see Fig.with vacuusing a buttoe windows h

Figu

both of thectors able ten exhaust

by electrica. 3). The dlower part o

subjects caopportuniti

investment

rmal sensatmance testsmitted ceili

n, Thermaludies on ad

o identical t (see Fig. de of insulandows can 2). The glam insulatio

on on the whas to be do

ure 2. One of t

e windows, o heat or cosystem in t

ally driven Vdaylight guiof the blind

an be grantees such as

which coul

tion, accepts between ting fan usag

l comfort daptive com

fully equip1). Each o

ated aluminbe opened

azing is a ton panels.

window catcone by hand

the office roo

the ventilaool the inlethe back of

Venetian blidance provds is comple

ted – depencontrolling

ld be

ance three ge.

and mfort

pped ffice nium

d and riple The

ch or d.

ms

ation et air f the

linds vides etely

nding g the

winhanthe rese

Fivecapisurftemsuch

3 Ex

Thecurrwer

Dursun contthe on rredu

In aconttem9:30belotheypostcom

ndow openind, all those researcher

earcher thro

e of six suillary tube face individ

mperatures wh panel exis

xperimenta

e experimenrent objectivre chosen: i-

ring the i- cprotection

trol the ceilceiling fan reverse moduced to a mi

addition, thetrolled non

mperature ca0am until 4ow the adapy were 1K atulate of th

mparable fro

ng, the Veneinteraction

or an algorugh LabVie

urfaces (allsystem, wh

dually. In thwas not in tsts.

Figu

al design

ntal design pves. With th-, i+, and iP

conditions, state and

ling fan stawas also allde, so that iinimum.

e set point n-steady-sta

alculated acc4:30pm theptive comfoabove the che adaptive om day to da

etian blindsns – except rithm througew.

except thehich allows e frameworthe hands o

ure 3. Venetian

presented byhe focus on.

the subjectlighting le

ate in additilowed durinits effect on

temperaturate. They cording to Deir set poinort temperatomfort bancomfort m

ay despite c

s or a ceilinga full opengh the build

e post andchanging th

rk of these eof the subje

n blinds with

y Schweiken the effect

ts were alloevel. Duringon to the cong the iP con changing t

res of the wstayed at

DIN EN 15nt temperatuture until 3K

nd accordingmodel, one changes in t

g fan (see Fning of the wding control

d beam façahe set pointexperiments

ects – in fac

daylight guida

er et al. (20of a ceiling

owed to cong the i+ seontrols of t

onditions. Ththe air velo

wall surface1K below

5251 betweeure was inK above it,g to categorcan expect

the outdoor

Figs. 2 and 3windows – l system ac

ade) are act temperatus the controct, they we

ance

12) was adag fan, only t

ntrol the wiettings, theyhe i- settinghus, the ceicity at the w

es and the aw the adaen 9am and

ncreased lin i.e. at the

ry I (see Figthat the th

weather con

3). On the ocan be don

ccessible for

ctivated wiure of each ol of the surere not told

apted to suithree condit

indow openy were ablg. The usagiling fan wawork space

air supply waptive comd 9:30am. Fnearly fromend of the

g. 4). Underhermal strainditions.

other ne by r the

ith a wall

rface that

t the tions

ning, le to ge of as set

was

were mfort From

m 1K day, r the in is

In tocon

The(frothreconsess

EveThecontand

Durquewas(4-smovand

Fi

Table 1.

Female Male All

otal 21 subjditions. Tab

e subjects wom 9 am tillee distinctivditions wassions, the ne

ery morninge mode of ttrolled by tiP-sessions

ring the dastionnaire w

s used to assscale), thervement as wdominance

igure 4. Temp

. Average and

N 10 11 21

jects (11 mable 1 summa

were asked l around 5 pve sessions s randomizext third wit

g, the subjecthe fan (revthe subjectss.

ay, subjectwith 30 itemsess the thermal accepwell as three (Lang et al

Figure 5. T

erature profile

d standard dev

Age [a25 (sd=4.625 (sd=4.025 (sd=4.2

ale, 10 femaarizes their

to work onpm with 30 introduced

zed, i.e. aroth i+ session

cts were tolverse or nos via a web

ts had to ms roughly ermal sensattance (4-sc

ee items rell., 1993).

Timeline of sur

e of wall surfa

viation of subje

a] H6) 1700) 1872) 179

ale) were chbasic chara

n their ownmin lunch above had

ound one tns and the l

ld, whether ot) was not b-based inte

fill out aevery 90 mtion vote (Acale), perclated to the

urveys and phy

aces and the a

ects’ gender, a

Height [cm]0 (sd=6.0) 7 (sd=5.8) 9 (sd=10.6)

hosen as subacteristics.

n during 3 cbreak) in th

d to be condthird of thelast group w

they couldcommunic

erface, whic

a computerminutes (see ASHRAE 7ceived air

perceived

ysiological me

ir supply syste

age, height an

Weigh63 (sd=783 (sd=73 (sd=

bjects based

consecutivehe test facilducted oncee subjects

with iP sessi

use the ceicated. The cch was unlo

-based 5-mFig. 5). Th

-scale), thermovement,levels of pl

easurements

tem

nd weight

ht [kg] 7.0) 10.8) 13.7)

d on their he

e days for 8lity. Each oe. The ordestarted wit

ions.

iling fan or ceiling fan ocked durin

minute comhe questionnrmal prefer, preferredleasure, aro

ealth

8 hrs f the er of th i-

not. was

ng i+

mfort naire ence

d air ousal

Twiperfmin

For distrwithwithpossatten

In atemcont

3 H

Theshowtogeconsessaccolow

4 A

4.1

Besobse

ice a day (bformance tenutes.

the Tsai-Pributed numhin 40 secohin which tsible (Wargntion and co

addition, phmperature, s

tinuously in

Hypotheses

e hypothesiwn for the tether with dition i+ asions. For ording to th

wer due to a

Fi

nalysis met

Data prepa

ides data merved value

The optemperamiddle

The neuaccordineutral

TnG = T

before lunchests in a row

artington, smbers from nds for eac

the subjectsgocki et aloncentration

hysical parasurface temn a one minu

s regardingthermal senan increas

are suspectethose obta

he literature.non-efficien

igure 6. Hypo

thods

aration

maintenance,es:

perative temature and aof the room

utral temperng to the mtemperature

Tg + (0 – TS

h and beforew: the Tsai-P

subjects wer1 to 99. T

ch sheet (Ams had to solv., 1999). Tn (Brickenk

ameters of tmperatures, ute interval

g the thermnsation votesed air veled to be hiained durin. They couldnt control an

theses related

, the follow

mperature wair velocity m. rature by Gr

modificationse to be

V) / R,

e the end ofPartington,

re given twThey had to mmons, 195lve as many

The d2 is a kamp, 2002

the indoor humidity,

l.

mal comforte in Figure 6locity, the igher, i.e. c

ng the iP sd be highernd correspo

d to session typ

wing items w

was derivemeasured

riffith’ meths described

f the experimaddition an

wo sheets oflink the nu

55). The ady additions

test for th).

and outdoo and air

t votes obt6. Due to ththermal co

closer to nesessions twr due to the onding dissa

pe and therma

were derived

ed based oadjacent to

hod, TnG, wby Rijal et

ment), they nd the d2 ta

f papers witumbers in addition task of 5 two-di

he assessme

or environmvelocity w

ained in eahe higher deomfort voteutral, com

wo outcomehigher degratisfaction.

al comfort vot

d from one o

on air temp the workp

was calculateal. (2008), w

had to do taking in tota

th 25 randoascending otook 5 min

digit numberent of selec

ment such awere meas

ach sessionegree of contes obtained

mpared to thes are possree of contro

tes.

or more dire

perature, gplaces close

ed for each which give

three al 15

omly order nutes rs as ctive

s air ured

n are ntrol d in he i- sible ol or

ectly

globe e the

vote s the

(1)

with Tg being the globe temperature and R a factor representing the expected change in TSV for each degree rise in the globe temperature. Tg is used as an approximation to operative temperature in their study, so that To could be used for this study without problems. R was set to be 0.33, which is related to the findings of Fanger (1972) that, all else being equal, the comfort vote increases by one unit for each three degree rise in globe temperature.

For the addition task, the speed was calculated according to the number of correct answers per second.

For the Tsai-Partington-test, the total number of correct links was calculated for both sheets, representing the total number of correct links within 80 seconds.

For the d2 test, the concentration performance was calculated according to Brickenkamp (2002).

4.2 Statistical analysis

In order to reveal differences between the sessions in thermal comfort votes and performance the Wilcoxon signed rank test for related samples (Wilcoxon, 1945) was evaluated using the statistical software SPSS. As input to the test, a) the means of all measurements for each session type (experimental day) were calculated for each subject and b) the means of the last two measurements of each session type (day) were taken.

The usage of the ceiling fan was recorded in percentage of the maximum speed. For the model development, these values were recoded to show “0” when the fan was switched of and “1” in the other case. Such dichotomous data can be used to derive a logistic regression model of the form

e 1

1 p

Op (-fan Ceiling , (2)

with pCeiling fan being the probability that one can observe the ceiling fan being switched on, α and β the coefficients derived and θOp the operative temperature.

4.3 Implementation into dynamic building simulation

The logistic regression model of ceiling fan usage for the i+ session was then implemented into a dynamic building simulation. First, an idf (EnergyPlus) model of a simple 4m width and 6m deep office cell with the South facing wall adjacent to the outdoors was created using DesignBuilder. The South façade has a wall-to-window ratio of 50%. The walls have a U-value of 1.5 W/(m2 K), the window a U-value of 1.72 W/(m2 K) and a g-value of .69 representing an existing not refurbished building standard from before 1978 in Germany. The building was situated in Freiburg/ Germany. There was neither cooling nor a sun shading device implemented. Schedules and activity levels were chosen to be typical for an office.

The idf model was then coupled via MLE+ (Nghiem, 2014) with a MatLab-model containing the code for the probabilistic ceiling fan usage model. Though, EnergyPlus does not offer a ceiling fan, for each timestep, the prevailing operative temperature was passed from EnergyPlus to MatLab. Within MatLab the probability of the ceiling fan was calculated according to the logistic regression model and compared to a random number. In case the random number was higher than the probability, the ceiling fan was considered to be switched on. At the end of each of the 100 simulation

runsmul

In acomduri

5 R

FiguopertimedesischeThemeareac

In Fshowprotwer

Thecolusessrestrmee

Theoperthe diffform

Fadap

s, the numbltiplied by 1

addition, themfort range ing i- and i+

Results

ure 7 preserative tempes of the coign is showeduled, i+ ae latter is duasurements.ct to increas

Figure 8, sewn alongsidtection statere meant to b

e total numumn of Tabsion were exrooms shoretings, whic

e mean valurative tempfollowing c

fer. The lattmer was due

Figure 7. Meaptive comfort t

ber of times100W, whic

e number owas define

+ sessions.

nts the meaperature andomfort vote

wn. While and iP sessi

ue to the sim Due to tim

sed thermal

et point, mede with thees. The tembe as explai

mber of voteble 2. Withxpected. Thrt before a ch could not

ues and stanperature, Top

columns of ter was expe to differen

an and standartemperature fo

the s

steps the cech was consi

of hours beied as 2K a

an differencd adaptive ces. In additi- sessions ions are in

mple controlme constrainloads either

easured sure behaviouramperatures s

ined above.

es in total ah all 21 subhe missing v

scheduled t have been

ndard deviatp, relative huTable 2. T

pected due tnt interactio

rd deviation ofor each sessioet point accor

eiling fan widered to be

ing outsideabove the m

ce together comfort temtion the set

tend to bethe middle

l algorithm ints in the pr solar or du

rface, air anal variables

show that re.

and for eacbjects givinvotes happen

vote, or havoided.

ations of runumidity, RH

The mean Tr

to the diffeons of subjec

of the differencon type and intrding to the ex

was switchee a medium

e the comfomedian neu

with the stmperature fot point accoe cooler at of the dayimplementepreparation ue to any oc

nd derived os for the wieal conditio

ch sessionng all votesned due to s

having been

nning meanH, and air vrm’s are ver

erence of cects with the

ce between opterval of comfxperimental de

ed on was spower.

ort range wautral temper

tandard devor each sessording to th

the end oy up to 2K hed during th

phase the ccupancy.

operative teindow, ceilions were wa

type is sho, a total ofsubjects havn absent du

n outdoor tevelocity, v, ary similar, weiling fan uir controls.

perative tempefort vote. The esign.

summed up

as derived. rature obse

viation betwsion type athe experimeof the day higher than

his first seriecontrol did

emperaturesing fan andarmer than

own in the f 126 votesving been toue to impo

emperature, are presentewhile Top a

usage, while

erature and th dashed line s

and

The erved

ween t the ental than

n set. es of d not

s are d sun

they

first s per o the rtant

Trm, ed in nd v e the

e hows

Fexp

Tru

Fig. 8. Exampperimental daythe set point t

tempera

Table 2. Numbunning mean o

Data (subset)

all

i-

i+

iP

le of temperatys (in this casetemperature foature in green

ber of votes, moutdoor tempe

veloc

N

356 4

119 4

120 4

117 4

ture values ane the first day

for the surfacen and the other

mean values aerature, Trm, opcity, v, for the

TSV T

4.4 (0.9)

4.4 (1.0)

4.2 (0.7)

4.6 (1.0)

nd behaviouraly was the i- seses; the operativr colours repre

and standard dperative temp

e full data set

Trm [°C]

19.3 (1.5)

19.1 (1.5)

19.5 (1.4)

19.3 (1.5)

l data for one ssion). The redve temperaturesent the five

deviations of thperature, Top, rand each cond

Top [°C]

26.7 (2.3)

25.8 (2.4)

26.9 (2.1)

27.2 (2.1)

set of experimd line in the toe is drawn in psurface tempe

hermal sensatrelative humiddition

RH [%]

45 (

48 (

45 (

42 (

ments, i.e. threop element shpurple, the aireratures.

tion vote, TSVdity, RH, and

] v [m/

(7) 0.17 (0

(7) 0.11 (0

(7) 0.27 (0

(6) 0.14 (0

ee hows r

V, air

/s]

0.15)

0.05)

0.21)

0.11)

5.1

Figuto e(firsstatiiP sbein

ThevoteTabstatisenstem

Fig

T

s

Compariso

ure 9 showseach sessionst three voting to sensesessions. Thng most com

e results of es and the ble 3. For istically sigsation votes

mperatures b

gure 9. Mean

Table 3. Medicom

Thermal sensation vo

Thermal preferenceThermal

acceptancNeutral

temperatur

PMV

on of therm

s the mean n type at eaes): the vote the most che afternoonmfortable, fo

the Wilcoxmean of thall items

gnificant cs differed stetween i- an

and standard

ian, Z- and p-vmfort. Signific

i

ote 4.

e

-.

e

-.

re

25

.4

mal comfort

and standarach survey ptes obtainedcomfortablen votes shoollowed by

xon Signed he calculate

except PMhange betw

tatistically snd i+ sessio

deviation of thand binne

value of Wilccant difference

Mediai- i+

33 4.17

33 -.17

50 -.33

5.1 26.6

42 .35

t votes and

rd deviationperiod. Thed during i+e, followed ow that the the iP and t

Rank testsd PMV for

MV, the Wween i- ansignificant bons.

hermal sensatd for each sur

coxon signed res (p<0.05) ar

an iP

7 4.67

7 -.50

-.67

6 25.9

.39

d performan

n of thermale tendencie

+ sessions aby those fri+ sessions

the i- sessio

s for the dar the six m

Wilcoxon sind iP-sessiobetween i+ a

tion votes grourvey period

rank tests for fre marked with

Result Wi- to i+

-.93 (.353)-.992 (.321)-1.33 (.183)-2.73 (.006)-.40

(.689)

nce betwee

l sensation vs vary durinre the lowerom the i- ss are sensed

ons.

aily mean theasurementgned-rank ons. In adand iP sessi

uped accordin

four items relah bold charact

Wilcoxon sigi+ to iP-2.77 (.006)-.766 (.444)-.987 (.324)-1.30 (.192)-1.65 (.099)

en sessions

votes accorng the morest, i.e. subjsessions andd in averag

hermal comts are showtest showe

ddition, theions and ne

ng to session ty

ated to thermaters.

gned rank teP i- to i

) -2.0(.039

) -2.2(.028

) -2.7(.006

) -2.4(.016

) -1.5(.114

rding rning jects d the ge as

mfort wn in ed a rmal utral

ype

al

est iP 6 9) 0 8) 4 6) 2 6) 8 4)

The results of the Wilcoxon Signed Rank tests for the mean of the last two thermal comfort votes and calculated PMV of each day are shown in Table 4. For all items, the Wilcoxon signed-rank test showed a statistically significant change between i+ and iP-sessions. In addition, neutral temperatures show statistically significant differences between all sessions. There is no statistically significant difference between i- and i+ or i- and iP sessions for the calculated PMV.

The results of the Wilcoxon Signed Rank tests for the performance tests are shown in Table 5. The Wilcoxon signed-rank tests showed that none of the sessions did elicit a statistically significant change (p<0.05) in any performance. Indeed, median ratings were close to similar. The analysis of the test batteries performed in the afternoon alone did not show any significant differences either.

Table 4. Median, Z- and p-value of Wilcoxon signed rank tests for four items related to thermal comfort for the last two measurements of the day. Significant differences (p<0.05) are marked with

bold characters.

Median Result Wilcoxon signed rank test i- i+ iP i- to i+ i+ to iP i- to iP

Thermal sensation vote

5.00 4.45 5.05 -3.07 (.002)

-3.00 (.003)

-.14 (.886)

Thermal preference

-.67 -.36 -.64 -1.91 (.057)

-2.17 (.030)

-.073 (.942)

Thermal acceptance

-.76 -.59 -1.0 -1.09 (.276)

-2.50 (.012)

-1.14 (.253)

Neutral temperature

23.8 26.9 25.2 -3.63 (.000)

-2.66 (.008)

-2.21 (.027)

PMV .67 .64 .69 -.295 (.768)

-1.96 (.050)

-1.20 (.230)

Table 5. Median, Z- and p-value of Wilcoxon signed rank tests for four items related to performance. “Speed” is the number of correct addition tasks per second, “number of correct links” is the number of

correct links drawn on two sheets within 40 seconds each.

Median Result Wilcoxon signed rank test i- i+ iP i- to i+ i+ to iP i- to iP

Speed (Addition)

.080 .080 .077 -.469 (.639)

-.035 (.972)

-.785 (.433)

Number correct links (Tsai-Partington)

23.5 23.5 24.0 -.187 (.852)

-.226 (.821)

-.040 (.968)

Concentration performance (d2)

181 187 188 -.261 (.794)

-.052 (.958)

-.052 (.958)

5.2

Thesimisubjcoef

Linethe spee

F

Tab

5.3

Figutemshowhigh

Duetest with

The941and neut

Logistic m

e logistic regilar (Fig. 1jects are usfficients of

ear models maximum ed was chos

igure 10. Obs

ble 6. Values o

Intercept Operative

Dynamic b

ure 11 showmperature, op

ws, that theher during s

e to the probrun to test

h a mean of

e number ohrs during for the i+

tral tempera

model of ceil

gression mo10). In botsing their cthe ceiling

of the chosspeed) howsen during i

served and fitttemperature

of coefficientsi+ ses

e temperatur

building sim

ws the outpuperative tem

e model persummer sea

babilistic chrun. The m

f 893hrs and

of hours outthe i+ sess sessions itatures stated

ling fan usa

odels for thh cases, thceiling fan fan usage m

sen strengthwever show iP-sessions

ted probabilitie as predictor f

s, standard errssions (Nagelk

V-2

re .8

mulation

ut of 1 of thmperature arforms as onson, than in

haracter of tminimum tod a standard

tside the coions. The ct is 26.6°Cd above.

age

he ceiling fahe operative

is around model durin

h (subjects cthat at the

compared t

ies to observe for all data an

ror and probabkerke’s R2-ind

Value 23.6 841

he 100 simuand ceiling ne would sun winter.

the model, otal usage tid deviation o

omfort rangcomfort ran

C ± 2K. Th

an usage due temperatu28°C. Tablg i+ session

could choossame oper

to i+-session

the ceiling fannd individual s

bilities of the dex for model

Standard er±.655

±.0237

ulation runsfan state. T

uspect. The

the usage pime is 756hof 57hrs.

ge is 1811hge for the iey were ob

ring i+ and ure at whicle 6 shows ns.

se between 0ative tempens.

n running usinsession types

ceiling fan us: .488)

rror Pro<<

with regardThe data of

usage of th

pattern diffehrs, the max

hrs during i- sessions ibtained usin

d iP-sessionsch 50% of the values

0% and 60%erature a hi

ng the operati

sage model du

obability <0.001 <0.001

rd to outdoof the ceilinghe ceiling fa

ers slightly fximum 102

i- sessions is 25.1°C ±ng the mea

s are f the s for

% of igher

ve

uring

or air g fan an is

from 28hrs

and ± 2K an of

6 D

As aimpsuit

For inteal. occuStillfor tit w200cont

Theconta nomodthe levesolu

Thepartobtatemcan

Figure 11. O

iscussion a

already shoplemented inable in orde

this projecteractions to

(2012), onupants werel, a statisticthe sessions

was not. Thi4), which strol.

e iP sessiontrol over theot neglectabde. With a ldifference

el out such ution for fur

e analysis otly show staained at th

mperature, Tbe seen, th

utput of 1 of 1

and conclus

own in Schwn a climate er to reveal

t, the effectbe research

nly the intee allowed t

cally signifis, when theis compliesshowed tha

ns were intee ceiling fanble effect olinear regrein the ceilieffect, allo

rther studies

of differenceatistically she last tw

TnG, differshat the therm

100 simulation

sions

weiker et alchamber hafurther insig

t and usage hed. In contreraction opto open the cant differe

e usage of ths with findiat occupant

ended to ten, while it d

on the air veession analying fan’s peowing the ss.

es in thermignificant do measure significantmal sensati

n runs using t

. (2012), caaving at leaghts into ad

of a ceilingrast to the epportunities

window anence was fohe ceiling fings of prevts thermal s

est a so-caldoes not havelocity due

ysis, it was erformance subjects the

mal comfort differences, ements of tly betweenion vote is

the probabilist

areful develast one winddaptive proc

g fan was chexperiments of the cend/or use th

ound in the fan was allovious researsensations d

lled placebove any effecto the ceil

further showby increas

e choice of

votes betwpartly not.each day

n the three higher for i

tic ceiling fan

oped experidow facing cesses.

hosen vicars presented eiling fan he ceiling fmean neutr

owed comparch (see e.gdiffers with

o-effect of ct. In realitying fan runwn that subsing the spea single sp

ween the se The analyshowed thsession typi+ sessions

usage model.

rimental desthe exterior

riously for oin Schweikdiffered, w

fan at all timral temperatared to the g. Brager eth the degre

giving subjy, there wasnning in revbjects reacteeed. In ordepeed could

ession typesysis of the vhat the nepes. In Figu

compared

.

signs r are

other ker et while mes. tures ones t al.,

ee of

jects s still verse ed to er to be a

s did votes utral

ure 9 to i-

during the first periods. In the afternoon, the thermal sensation vote is lower, i.e. subjects perceived the conditions as more comfortable. This would support the hypothesis that at the beginning, having control positively influences on the occupants thermal sensation but at the end of the day, such control needs to be effective in order to remain such influence. An analysis compared to the theory of alliesthesia (Parkinson et al., 2012), is difficult to perform based on the existing data due to the number of interaction types and the lack of comfort votes close to the interactions.

The occupant behaviour model for the ceiling fan usage considered in this study – a multivariate logistic regression model – is compared to the models presented in the literature (see e.g. Haldi and Robinson, 2009) one of a simple type. Discrete-time Markov chain models and others would be usefully implemented in order to increase the validity of the results. Such analysis is beyond the scope of this paper and will be presented elsewhere.

The air movement and temperature distribution used for this analysis is based on a one-node-room model. This is valid for the measured data (taken in the middle of the room) as well as the simulated data. Therefore, the validity between statistical model and simulation is given. In order to increase the accuracy, detailed investigations of the air flow profile of the existing ceiling fan as presented in Voss et al. (2013), would be meaningful.

The dynamic building simulation showed, that the elevated comfort range due to the ceiling fan nearly halved the hours with temperatures above the comfort range. In the chosen case study, the resulting hours above the comfort level are with more than 800 hrs still above an acceptable number of around 260 hrs (= 3% of a year as denoted in DIN EN 15251). The reduction of hours outside the comfort range comes through a relatively easy measure.

The installation costs of a ceiling fan could be estimated depending on the situation between 150€ and 500€ per office. The running electricity cost for each ceiling fan can be calculated to be in average 893 hrs/a x 100 W x 0.25ct/kWh = 22.3 €/a.

In conclusion, the results suggest that it is not the control itself, which leads to a higher acceptance of increased indoor temperatures in summer, but the effectiveness of the control. For the analysis of the performance tests, only minor differences in the performance under the distinctive conditions were observed.

Comfort votes and simulation suggest, that the installation of a ceiling fan does not only improve thermal comfort of occupants, but could be a cost effective measure resulting in a lower cooling demand. Further, building concepts using naturally ventilation will work in a wider range of climate conditions.

Acknowledgement

This project is funded by the German Federal Ministry of Economics and Technology (BMWi) with the project ID: 0327241C. The new test facility LOBSTER was funded by the German Federal Ministry of Economics and Technology (BMWi) with the project ID: 03ET1035B and supported by industrial partners. The authors are indebted to the Baden-Württemberg Stiftung for the financial support of this research project by the Eliteprogramme for Postdocs.

References

Ammons, CH (1955), Task for the study of perceptual learning and performance variables. Perceptual and Motor Skills, 5, 11-14.

Brager GS, Paliaga G, de Dear R (2004), Operable windows, personal control, and occupant comfort, ASHRAE Transactions 2004;110(2), 17-35.

Brickenkamp R (2002), Test d2-Aufmerksamkeits-Belastungs-Test, Manual. Göttingen: Hogrefe-Verlag GmbH.

de Dear R, Brager G, Cooper D (1997), Developing an adaptive model of thermal comfort and preference, in: Final Report on ASHRAE Research Project 884, Macquarie University Sydney.

DIN EN 15251 (2007), Indoor environmental input parameters for design and assessment of energy performance of buildings addressing indoor air quality, thermal environment, lighting and acoustics, German version EN 15251:2007.

Fanger PO (1972), Thermal comfort, analysis and applications in environmental engineering. McGraw-Hill, New York.

Haldi, F, Robinson, D (2012), Interactions with window openings by office occupants, Building and Environment 44, pp. 2378-2395.

Humphreys MA, Nicol JF (1998), Understanding the adaptive approach to thermal comfort, ASHRAE Transaction, 104(1), 991-1004.

Lang PJ, Greenwald MK, Bradley MM, Hamm AO (1993), Looking at pictures: affective, facial, visceral, and behavioral reactions, Psychophysiology, 30(3), 261-273.

Nghiem, TX (2014), MLE+: a Matlab-EnergyPlus Co-simulation Interface, URL: http://www.seas.upenn.edu/~nghiem/mleplus.html, accessed 14.01.2014.

Parkinson, T, de Dear, R, Candido, C (2012), Perception of Transient Thermal Environments: Pleasure and Alliesthesia, Proceedings of 7th Windsor Conference, Windsor, UK.

Rijal HB, Tuohy P, Humphreys MA, Nicol JF, Samuel A, Raja IA, Clarke J (2008), Development of adaptive algorithms for the operation of windows, fans, and doors to predict thermal comfort and energy use in Pakistani buildings, ASHRAE Transaction 114(2), 555-573.

Schweiker M, Brasche S, Bischof W, Hawighorst M, Voss K, Wagner A (2012), Development and validation of a methodology to challenge the adaptive comfort model, Building and Environment, 49, 336-347.

Voss, K, Voß, T, Otto, J, Schweiker, M, Rodriguez-Ubinas, E (2013), Investigation of ceiling fans for improving summer thermal comfort, Proceedings of the 2nd Central European Symposium on Building Physics, Vienna, Austria, 9-11 September 2013.

Wargocki, P., Wyon, D.P., Baik, Y.K., Clausen, G. and Fanger, P.O. (1999b), Perceived air quality, Sick Building Syndrome (SBS) symptoms and productivity in an office with two different pollution loads, Indoor Air, 9, 165–179.

Wilcoxon, F. (1945), Individual Comparison by Ranking Methods, Biometrics Bulletin, (1)6, 80-83.