at SciVerse ScienceDirect

Building and Environment 49 (2012) 25e32

Contents lists available

Building and Environment

journal homepage: www.elsevier .com/locate/bui ldenv

Adaptive comfort: Analysis and application of the main indices

Simone Ferrari*, Valentina ZanottoDepartment of Building Environment Science & Technology (BEST), Politecnico di Milano, Via Bonardi 3, 20133 Milano, Italy

a r t i c l e i n f o

Article history:Received 7 June 2011Received in revised form9 August 2011Accepted 28 August 2011

Keywords:Thermal comfortAdaptive approachPassive coolingItalian climatic context

* Corresponding author. Tel.: þ39 (0) 223999483; fE-mail address: simone.ferrari@polimi.it (S. Ferrar

0360-1323/$ e see front matter � 2011 Published bydoi:10.1016/j.buildenv.2011.08.022

a b s t r a c t

Active climatization is currently one of the main causes of energy use in buildings. Since it aims atproviding indoor environmental conditions that are comfortable for most of the building occupants, theway these conditions are determined is very important in the framework of energy optimization.

Indoor comfort conditions are conventionally expressed in terms of steady temperature levels (e.g. 20e26 �C). Differently, the adaptive approach determines temperature levels that are unsteady and followthe variability of the outdoor climate.

Even if this alternative approach has proven to be very effective in providing mitigated indoortemperatures, agreement about its formulation and its practical application is still lacking.

In this paper some of the available formulations of the adaptive approach are described and adopted todetermine comfort temperatures for three different Italian climatic contexts. Moreover, the discomfortlevels for a case-study room are estimated, by the means of a dynamic building energy simulation model,according to both the conventional and the adaptive approaches.

� 2011 Published by Elsevier Ltd.

1. Introduction

Thermal comfort is defined as that condition of mind whichexpresses satisfaction with the thermal environment [1]. Actually,building users simply relate thermal conditions to the airtemperature levels, regulated by managing the set-points of theclimatization systems with subsequent effects on energyconsumption.

In order to assess the quality of thermal environments in detail,the main international standards regarding the determination ofthe indoor comfort conditions [1e3] refer to the work of Fanger[4]. The heat balance of occupants staying in an indoor environ-ment is defined in [4] in terms of personal and environmentalparameters affecting the heat exchanges between the human bodyand the room. In order to clarify the related physics equation, theresulting heat exchange values were later connected to the 7-points thermal sensation scale [1] by collecting the responses ofmore than 1000 people to different environmental conditionsprovided inside a climatic chamber. As a result the Predicted MeanVote (PMV) and the Predicted Percentage of Dissatisfied (PPD)indices were defined: in these indices the ideal state for theoccupants, called “neutrality”, corresponds to the thermal

ax: þ39 (0) 223999484.i).

Elsevier Ltd.

equilibrium with the indoor environment (equivalence of inwardand outward heat flows).

In standards [1e3] the Fanger equation is however used todetermine (according to conventional clothing level and standardair humidity and velocity) allowable reference ranges of operativetemperature for different building uses, which correspond todifferent activity levels.

Several studies pointed out some flaws in this approach. First ofall, in climatic chamber tests human beings are considered aspassive “sensors” detecting the environmental conditions, that donot interact with the room in anyway: this situation is very peculiarand deeply different from what happens in most of real buildings.Moreover, another important observation deals with the choice ofequating comfort to the neutrality of heat exchanges between thebody and the environment: the definition of ideal thermal envi-ronment as the neutral one is based on a very deterministicapproach, which does not take into account the psychological andcultural aspects of comfort and can therefore be questioned [5].

For these reasons, alternative studies were carried out analysingthe occupants’ sensation and preference inside actual buildings,and brought to the development of another method to assess theindoor thermal conditions: the adaptive approach.

During the last decades, several of such field studies wereconducted, with different specific purposes and results, bringing todifferent formulations of the adaptive approach: this paperdescribes the main characteristics of the available adaptive indicesand compares their application to the Italian context.

Nomenclature

Tco indoor comfort operative temperature [�C]Text,ref outdoor reference temperature [�C]Trm running mean temperature [�C]Tdm daily mean temperature [�C]va air velocity [m/s]

Subscriptsn present day

S. Ferrari, V. Zanotto / Building and Environment 49 (2012) 25e3226

2. The adaptive approach

From the 1970s, field studies on environmental conditions andcomfort determination inside real buildings started taking place,and results pointed out important differences between the PMV/PPD predictions and the actual thermal sensation/preferenceexpressed by the building occupants [6].

Considering the results of these studies, an alternative approachto the definition of “comfortable environmental conditions” wasdeveloped. The main assumptions of this theory, called “adaptive”[7e9], regard:

- the ability of human beings to adapt themselves to the envi-ronmental conditions (through conscious or unconsciouschanges in their metabolic rate or clothing level) and to interactwith the environment in order to adapt it to their needs(through available environmental controls);

- the influence of thermal experience on the occupants’ expec-tations regarding the indoor conditions, which can be short-term, due to the recent weather, or long-term, related to thegeneral climate they are used to.

The huge amount of data collected in these studies [10,11]allowed a statistical analysis which, among all the consideredenvironmental parameters, revealed a direct correlation betweenthe indoor comfort temperature and the outdoor one that iscommonly expressed by the adaptive approach founding equation

Tco ¼ a$Text;ref þ b (1)

where a is the slope of the function and b is its y-intercept, bothstatistically derived by the analysis of the collected data.

As already introduced, several adaptive indices have beendeveloped during the last decades: the following sections analysethe elements characterizing the various formulations.

2.1. a and b values

The main difference between the applications of the adaptiveapproach is in the equation itself, which, as previously mentioned,is derived by the statistical analysis of extensive field studies.According to the assumptions of the related study and to the ob-tained results, both the slope (a) and the y-intercept (b) of theequation change: the slope value, in particular, represents thecorrelation between Tco and the Text,ref, in other words the “adap-tiveness” of the equation.

2.2. Outdoor reference temperature

The outdoor reference temperature (Text,ref) is the only indepen-dent variable of the adaptive equation, and theway it is determinedis very important in defining the kind of thermal experience taken

into account in the index. Among the adaptive approach applica-tions, two main kinds of temperature are considered: the monthlyaverage one (Tmm) and the running mean one (Trm).

The monthly average temperature (Tmm) was the first one to beused. Since it is based on the historical series of air temperatures ina specific location, it represents a typical climate and is thereforeconnected to the occupants’ long-term experience.

During the 1990s, Nicol and Humphreys [9,12] suggested the useof a progressive value for the outdoor temperature, following theassumption that exponentially weighted average data would allowa higher reliability in the relationship between indoor and outdoortemperature. The running mean temperature (Trm) was thereforeintroduced in the adaptive equation. In its general expression, Trm isan average of the mean daily temperatures of a certain number ofdays immediately before the analysed one, weighted according totheir time distance:

Trm;n ¼ ð1� aÞ$�Tdm;n�1 þ a$Tdm;n�2 þ a2$Tdm;n�3

þa3$Tdm;n�4 þ.�

(2)

where a is a reference constant value, ranging between 0 and 1(recommended 0.8).

Being based on the recent daily temperature data, the runningmean temperature is connected to the occupants’ short-termexperience.

2.3. Acceptability range amplitude

There is a limited interval of temperatures around the idealcomfort one (Tco, calculated by the means of Eq. (1)) that can beconsidered as comfortable according to a specific quality level.These intervals are called “acceptability ranges” and are definedthrough successive temperature thresholds.

In the available indices, the ranges amplitude is usually definedby the means of constant values deriving from the statisticalanalysis of the field studies results. In one case [12], however,a different approach was adopted, with a variable interval widthcalculated according to the comfort temperature value: theresulting correlation reveals a range of temperatures comfortablefor the occupants that becomes narrower as the outdoor conditionbecomes warmer.

2.4. Applicability

Due to its assumptions, the adaptive approach has some appli-cability limitations, that are still a matter of discussion.

First of all some indices adopt a climatic restriction, according tothe fact that the adaptation dynamics usually refer to warmconditions. In fact the adaptive mitigation strategies, that are basedon interactions with the external climate variability, are not equallyeffective considering the harshness of cold conditions. Most of thefield studies were therefore conducted considering only warmclimates and/or the warm season and some of the adaptive equa-tions can be applied only in these contexts.

Moreover, it was verified that the adaptation dynamics havedifferent importance whether considering “naturally ventilatedbuildings”, with manually operable windows and without activesystem, or “HVAC buildings” equipped with sealed facades andmechanical ventilation: different equations have been consistentlyderived.

Actually most of real buildings, particularly in Europe, do notclearly fall into one of the two categories but lie somewhere in-between (e.g. conditioned buildings with operable windows,hybrid ventilated buildings, etc.). Some indices (e.g. [1] and [3])

ASHRAE index comfort ranges

15

20

25

30

35

40

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40Outdoor reference temperature [°C]

Indoor tem

perature [°C

]

80% acc. upper limit 90% acc. upper limit80% acc. lower limit 90% acc. lower limit

Fig. 1. Acceptability ranges calculated according to the ASHRAE index.

S. Ferrari, V. Zanotto / Building and Environment 49 (2012) 25e32 27

consider all these cases as “HVAC buildings”, even if this impliesunderestimating the occupants adaptation potential.

Recent studies, however, found out that the adaptationdynamics are as effective in the so-called mixed-mode buildings(with hybrid ventilation, performed both by windows operationand by mechanical air distribution) as in the “naturally ventilatedbuildings” [13,14]. This is due to the importance of the “perceivedadaptive opportunity” in influencing the thermal sensation/pref-erence of human beings. In fact, in [15], the “perceived adaptiveopportunity” is defined as a psychological factor, which directlydepends on the availability of the environmental controls whetherthey are actually used or not.

2.5. Weak spots

The statistical nature of the approach is a strong weakness,because it implies that the resulting equations are fully reliable onlywhen the considered boundary conditions are similar to the onestaken into account into the original studies.

Another weakness of the adaptive approach is the fact that it isfocused only on temperature, underestimating the effect of otherindoor parameters such as air velocity and humidity. As acknowl-edged by most standards, the first one affects the comforttemperature, raising its value. For this reason, Nicol [16] suggests toincrease the comfort temperature level, in case of air velocityhigher than 0.1 m/s, by

7� 504þ 10$v0:5a

(3)

The effect of air humidity is universally known in combinationwith high temperatures [2]: according to Nicol [16] the increase ofhumidity affects the width of the acceptability range around theideal comfort temperature, but there is still no simple equationavailable to integrate it with temperature.

3. Adaptive indices considered in this study

As introduced before, in the last few years several field studieswere performed and several adaptive equations were developedfrom their results. Among the available adaptive indices, thefollowing ones were selected and analysed, as the most adoptedand the ones deriving from the largest field studies.

3.1. ASHRAE

In the 1990s, ASHRAE commissioned a specific research projectfrom deDear and Brager [17], which collected information froma lot of different field studies performed worldwide, for a totalamount of 21,000 data points whichwere later elaborated to get thefollowing equations:

Tco ¼ 0:31$Text;ref þ 17:8 (4)

in case of naturally ventilated buildings and for outdoor tempera-tures ranging from 5 �C to 32 �C, and

Tco ¼ 0:11$Text;ref þ 21:45 (5)

in case of “HVAC buildings”. In this case Text,ref is the monthlyaverage outdoor temperature.

It is clear that the “adaptiveness” (represented by the slopevalue) is much higher in case of unconditioned buildings.

The research found two acceptability ranges, that correspond topercentages of satisfied occupants and are defined by constantvalues: an 80% acceptability range connected to a temperature

interval of �3.5 �C for typical application and a 90% acceptabilityrange connected to a temperature interval of �2.5 �C in casea higher level of thermal comfort may be desired.

The first equation resulting from this study was later imple-mented in the American Standard regarding the assessment of thethermal conditions in indoor environments [1], with the followinglimitations:

- for the summer season, in particular for Text,ref ranging between10 �C and 33.5 �C;

- in case of buildings where occupants can directly operatewindows;

- in case of buildings where the occupants perform low meta-bolic rate activity (<1.3 MET).

In Fig. 1 the resulting acceptable ranges of indoor operativetemperature according to the outdoor reference temperaturevalues are shown.

3.2. ACA

The Smart Control and Thermal Comfort project (SCAT),promoted by the European Commission, aimed to reduce energyuse due to air conditioning systems by varying the indoortemperature through the use of an “adaptive algorithm” [12]. In thisstudy, ended in 2000, 5 buildings were surveyed in 5 Europeancountries, characterized by different climatic conditions, for a totalamount of 25 cases, covering several specific functions and HVACsystems (naturally ventilated buildings, climatized buildings andmixed-mode buildings). From the different national resultsa general European equation, called Adaptive Comfort Algorithm(ACA), was developed:

Tco ¼ 0:302$Text;ref þ 19:39 (6)

where Text,ref is the running mean temperature calculated fora suggested time interval of 3.5 days.

This equation is considered reliable only for outdoor tempera-tures higher than 10 �C, while below this limit a constant value ofTco ¼ 22.88 �C has to be considered.

The acceptability range amplitude is determined by

�0:189$Tco þ 6:35; (7)

which expresses an inverse correlation to the comfort temperature.The choice of this kind of acceptability range can be easily

ACA index comfort range

30

35

40

em

perature [°C

]

Upper limit

Lower limit

S. Ferrari, V. Zanotto / Building and Environment 49 (2012) 25e3228

explained by the fact that the index was developed, in this case,taking into account also climatized buildings. The activation ofcooling systems is most likely when the indoor and outdoortemperatures are at their highest, in fact, and therefore the adap-tation dynamics become less effective and the acceptability rangebecomes narrower.

The resulting acceptable ranges of indoor operative temperatureaccording to the outdoor reference temperature values are shownin Fig. 2.

15

20

25

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40Outdoor reference temperature [°C]

Indoor t

Fig. 2. Acceptability range calculated according to the ACA index.

3.3. ATG

In the Netherlands, the findings of deDear and Brager [17] and inparticular both Eqs. (4) and (5) were used to develop the localstandard regarding the comfort condition determination, calledAdaptive Temperature Limit (ATG) [18].

In this case, the outdoor reference temperature is the runningmean one, calculated by

Trm ¼ Tdm;n þ 0:8$Tdm;n�1 þ 0:4$Tdm;n�2 þ 0:2$Tdm;n�3

2:4: (8)

Eq. (8) is based on a time interval of 4 days starting from thecurrent one: the index use is limited to analysis during the designphase, through building simulation, and/or during the servicephase, through data registered in the field.

The main peculiarity of the Dutch norm, however, is that itapplies the adaptive approach to every kindof building,whether it isconditioned or not: buildings are, in fact, divided into two categories(Alpha and Beta) which differ by the accorded “adaptive opportu-nity” (mostly meaning the accessibility of environmental controls,operating bothwindows andHVAC system). A flowchart is providedtogether with the standard to sort out the buildings between thetwo categories: this sorting tool is being assessed and is likely to beadjusted according to the lessons learnt by its practical application.Even if it is notdefinitive, this approach is very interesting, because itattempts to avoid the ambiguities connected with the traditionaldistinction between naturally ventilated and HVAC buildingswithout underestimating the occupants’ adaptation potential.

Regarding the acceptability ranges, the original 90% and 80%limits suggested by Ref. [17] are adopted together with an addi-tional 65% limit according to the following interpretation:

Table 1Upper comfort limits suggested by the Dutch ATG index.

Acceptability class Only for Alpha buildingsand Text,ref > 12 �C

A e 90% Tco < 0.11Text,ref þ 22.70 Tco < 0.31Text,ref þ 20.30B e 80% Tco < 0.11Text,ref þ 23.45 Tco < 0.31Text,ref þ 21.30C e 65% Tco < 0.11Text,ref þ 23.95 Tco < 0.31Text,ref þ 22.00

Trm ¼ Tdm;n�1 þ 0:8$Tdm;n�2 þ 0:6$Tdm;n�3 þ 0:5$Tdm;n�4 þ 0:4$Tdm;n�5 þ 0:3$Tdm;n�6 þ 0:2$Tdm;n�7

3:8(10)

- class A (90%) represents “very good” indoor conditions, and isappropriate for buildings with relatively sensitive users or withhigh comfort requirements;

- class B (80%) represents good indoor conditions, and isappropriate for common offices;

- class C (65%) represents acceptable indoor conditions, and isappropriate in case of existing buildings.

The resulting equations defining the upper comfort limits arecollected in Table 1. From these equations it can be noted that, foroutdoor reference temperatures lower than 12 �C (i.e. in the winterseason) the reference equation for both building categories is thesame, the one with lower adaptive potential: this reflects the factthat adaptation is more effective in summer.

Regarding the lower limits, it was decided to maintain the sameequations regardless for season and building type, with the loweradaptive potential slope value (0.11) and y-intercept values of 20.20(class A), 19.45 (class B) and 18.95 (class C).

Fig. 3 shows the acceptable ranges with the upper limits trendchanging over 12 �C of outdoor reference temperature, according toAlpha type buildings equations.

3.4. CEN

According to the results of a specific task group and partly on thefindings of the SCAT project [19], the European Committee forStandardization (CEN) introduced the adaptive approach in stan-dard EN 15251 in 2007 [3].

The basic equation is

Tco ¼ 0:33$Text;ref þ 18:8 (9)

and is considered reliable only for outdoor reference temperaturesbetween 10 �C and 30 �C. In this case, Text,ref is the running meantemperature calculated for a time interval of 7 days, according to

The acceptability ranges define three requirement levels accordingto different building types:

- category I, which corresponds to high level of expectations(PPD < 6%), recommended in case of very sensitive and fragileoccupants, is determined by a temperature interval of �2 �C;

Selected cities climate comparison

Monthly average outdoor air temperature

02468

10121416182022242628

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Outdoor A

ir T

em

perature [°C

]

Milano Roma Palermo

Fig. 5. Comparison of the average monthly outdoor temperature values for theselected cities.

ATG index comfort ranges - Alpha building type

15

20

25

30

35

40

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40Outdoor reference temperature [°C]

In

do

or te

mp

era

tu

re

[°C

]

65% acc. upper limit 80% acc. upper limit 90% acc. upper limit

65% acc. lower limit 80% acc. lower limit 90% acc. lower limit

Fig. 3. Acceptability ranges calculated according to the ATG index for Alpha buildings.

S. Ferrari, V. Zanotto / Building and Environment 49 (2012) 25e32 29

- category II, which corresponds to a normal level of expectation(PPD < 10%), recommended in case of new buildings andrenovations, is determined by a temperature interval of �3 �C;

- category III, which corresponds to a moderate level of expec-tation (PPD < 15%) and can be used in case of existing build-ings, is determined by a temperature interval of �4 �C.

Even if [3] introduces the adaptive approach within the Euro-pean norms, its applicability is very limited:

- to evaluations in the summer season;- to buildings used for low metabolic rate activities (<1.3 MET);- to buildings where occupants can freely operate windows andchange their clothing level;

- to buildings without any HVAC system.

In Fig. 4 the resulting acceptable ranges of indoor operativetemperature according to the outdoor reference temperaturevalues are shown.

4. Comparative analysis

This paper reports the results of a comparative study among thepreviously described adaptive indices in determining the summercomfort conditions for the cities of Milano, Roma and Palermo,

CEN index comfort ranges

15

20

25

30

35

40

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40Outdoor reference temperature [°C]

In

do

or te

mp

era

tu

re

[°C

]

III -upper limit II - upper limit I - upper limit

III - lower limit II - lower limit I - lower limit

Fig. 4. Acceptability ranges calculated according to the CEN index.

which are characterized by very different climates (Fig. 5), coveringa wide range of the Italian conditions.

The study is divided in two parts:

- a first comparison of the upper comfort limits for the selectedlocations, determined according to the test reference year(TRY) data;

- an analysis of the adaptive comfort level estimation, accordingto the selected indices, by assessing the summer indoorthermal conditions of a case-study office room.

The indices were interpreted considering the following accept-ability ranges, which are the ones appropriate for common officebuildings:

- 80% acceptability for ASHRAE and ATG indices;- limit calculated according to Eq. (7) for ACA index;- Category II for CEN index.

4.1. Upper limit comparison

Figs. 6e8 show the daily values of the upper comfort limitsaccording to the indices throughout a year, together with thetraditional baseline of 26 �C (which is the summer comfort limit formoderate activity according to Refs. [2,3]).

Since they are based on the outdoor temperature trend, all theadaptive indices reveal a pronounced yearly variability.

The CEN limit is the one allowing the highest indoor tempera-tures: this is due to its formulation, which shows both the highestslope value (0.33), determining a comfort temperature that followsthe outdoor reference one more precisely, and the highest y-intercept. Differently, the apparent lower “adaptiveness” of the ACAindex is due to the acceptability range determination: since therange becomes narrower as the comfort temperature increases, thelocal maxima are less pronounced. It is also interesting to point outthe differences between the ASHRAE and the ATG indices, whichare based on the same basic equation and differ just in the outdoorreference temperature: the ASHRAE index, in fact, gives monthly

Palermo

22

23

24

25

26

27

28

29

30

31

32

Te

mp

era

tu

re

[°C

]

ASHRAEACAATGCEN

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Fig. 8. Comparison of the upper comfort limits calculated according to the selectedadaptive indices for Palermo.

Milano

22

23

24

25

26

27

28

29

30

31

32

Te

mp

era

tu

re

[°C

]

ASHRAEACAATGCEN

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Fig. 6. Comparison of the upper comfort limits calculated according to the selectedadaptive indices for Milan.

S. Ferrari, V. Zanotto / Building and Environment 49 (2012) 25e3230

values instead of the daily ones given by ATG and therefore does notprovide smooth changes as time passes by.

The graphs also reflect the fact that the ASHRAE and CEN indiceshave stronger limitations regarding the application of the adaptiveapproach: they do not provide, in fact, a continuous comfort limitthroughout the year, in particular in the colder locations, because ofthe lack of specification in case of outdoor reference temperaturelower than 10 �C. However, the first index, being based on theoutdoor monthly mean temperature, is characterized by a limitedbut continuous application interval, while the CEN index, which isbased on the more variable running mean temperature, is providedwith a very discontinuous range.

4.2. Case-study office room

The summer indoor environmental conditions of the case-studyoffice room were analysed by the means of dynamic simulationsperformed with Energy Plus [20], which requires a detaileddescription of the space in terms of its geometry, thermal proper-ties of the construction materials and usage patterns with relatedinternal heat production (i.e. people, lighting and equipments).

The “base case” office is characterized by a 21 m2floor area,

a 63 m3 inside volume and a 10.5 m2 external wall with anunshaded 5 m2 window, surrounded by similar spaces (withina larger building, as shown in Fig. 9). The building elements arecharacterized by constructions that are very common in existingbuildings in Italy: two layers of hollow bricks and in-betweeninsulation for the external walls (U-value 0.5 W/m2 K), a concrete

Roma

22

23

24

25

26

27

28

29

30

31

32

Te

mp

era

tu

re

[°C

]

ASHRAEACAATGCEN

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Fig. 7. Comparison of the upper comfort limits calculated according to the selectedadaptive indices for Rome.

and masonry solution for the horizontal structures (insulated incase of the roof, with U-value of 1.0 W/m2 K) and double glazingwith wooden frame for the windows (U-value 2.7 W/m2 K). Theoffice is a 2 occupants room equipped with two personalcomputers, a printer and a fax, with four fluorescent tubes and twosmaller table lamps (overall internal heat load of 40 W/m2): thepattern is based on hourly variable usage rates within typicalworking week (from Monday to Friday) and day (from 8:00 to20:00).

Settings regarding ventilation consider a constant air changerate due to infiltration (0.5 h�1) and additional natural ventilationfor an amount of 2 h�1, operated only when the indoor temperatureexceeds both 26 �C and the outdoor one. In order to comply withthe requirements of all the considered approaches, the roombehaviour was analysed without active climatization devices (free-running conditions).

The room was simulated for the four main orientations (North,South, East, West).

The effect of some basic passive cooling strategies (i.e. shadingand enhanced natural ventilation) was also investigated, defininga “best case” office:

- shading was planned as optimized considering orientation,with a properly dimensioned fixed overhang for the southernfaçade and mobile external shading for the eastern andwestern ones (operated only in case of direct solar radiation onthe surface);

- enhanced ventilation was applied to the additional ventilationrate (operated according to indoor and outdoor air temperatureconditions) to reach an overall amount of 10 h�1.

3.5

6.0 6.0

3.03.51.4

Fig. 9. Plan (left) and section (right) of the case-study office room (highlighted in grey)within the larger building. Dimensions are expressed in m.

Table 2Thousands of discomfort degree-hours calculated according to the various indices, for the base and best case office and considering the different orientations.

Milan Rome Palermo

ASHRAE ACA ATG CEN ASHRAE ACA ATG CEN ASHRAE ACA ATG CEN

Base case officeN 8.2 8.2 6.6 4.6 10.3 10.7 8.6 6.0 12.3 13.9 11.2 8.1S 13.6 13.8 11.9 9.2 16.5 17.1 14.7 11.4 20.0 21.9 18.7 14.5E 10.3 10.4 8.7 6.4 13.0 13.5 11.3 8.5 15.6 17.3 14.3 10.8W 13.4 13.5 11.7 9.1 16.7 17.2 14.9 11.8 20.4 22.3 19.2 15.3Best case officeN 3.7 3.6 2.4 1.5 5.1 5.3 3.8 2.3 5.5 6.6 4.4 2.6S 3.1 2.8 1.8 1.0 4.4 4.7 3.2 1.9 4.8 5.8 3.7 2.1E 3.2 3.0 1.9 1.1 4.1 4.4 2.9 1.6 4.0 5.0 3.0 1.6W 3.3 3.1 2.0 1.1 3.9 4.3 2.8 1.5 4.0 5.0 3.0 1.5

Table 3Percentage reduction of discomfort degree-hours determined according to thevarious adaptive indices compared to the ones calculated according to theconventional set-point (26 �C) for the base case and the best case office. The valuesare average among the different orientations.

Base case office Best case office

ASHRAE ACA ATG CEN ASHRAE ACA ATG CEN

Milano 25.3% 24.6% 36.6% 52.5% 45.1% 48.8% 67.1% 80.7%Roma 28.2% 25.6% 37.3% 52.5% 50.3% 46.9% 64.2% 79.5%Palermo 34.3% 27.5% 39.3% 53.5% 60.7% 51.4% 69.5% 83.4%

S. Ferrari, V. Zanotto / Building and Environment 49 (2012) 25e32 31

According to what stated in [3], the results of the simulation areanalysed in terms of discomfort degree-hours (DH), which take intoaccount both the amount of time operative temperature fallsoutside the comfort range and the difference between the actualtemperature and the allowed one: the amount of DH becomesa general index describing the building performance, since it can beconsidered proportional to the building seasonal energy demand.

Table 2 summarizes the resulting discomfort degree-hours,sorted by climate, room orientation and equipment (base or bestcase), and adaptive index. The results mostly reflect the observa-tions made in the previous section, with the CEN index being themost “forgiving” in case of high internal temperatures and the ACAindex being the strictest one. Interestingly, again, the ATGdiscomfort degree-hours are less than the ASHRAE ones, even ifthey are based on the same equation: this can indicate that theshort-term experience (represented by the running meantemperature) is stronger in determining “forgiving expectation”

Thousands of discomfort degree-hours

Base case office - orientation South

0

5

10

15

20

25

30

Milano Roma Palermo

ASHRAE ACA ATG CEN 26

Fig. 10. Thousands of discomfort degree-hours calculated according to the selected ada

than the long-term one (represented by the monthly averagetemperature).

In order to facilitate the reading of the results, the discomfortdegree-hours assessed by the means of the conventional practice(with a 26 �C comfort limit, as suggested by Refs. [2,3]) have alsobeen calculated and compared to the ones presented in Table 2.

The adaptive indices always evaluate less DH than the 26 �Climit, which could be translated in equivalent energy savings ifa cooling system was regulated according to an adaptive set-point.The value of DH percentage reduction results, summarized inTable 3, range between a minimum 20% (ACA in Milano) anda maximum 60% (CEN in Palermo) for the base case office.

As an example, the values for the case-study offices facing Southare presented in Fig. 10. All the adaptive indices follow the differentclimatic conditions with tolerance, while the conventionaltemperature limit, which is stricter, determines a higher growth ofthe discomfort DH as the climate becomes warmer. Among theadaptive indices, however, the ASHRAE one shows a slightlydifferent trend compared to all the others because of the monthlybased outdoor reference temperature.

Moreover, when the introduction of the passive cooling strate-gies is taken into account, it is also evident that the adaptiveapproach appreciates the related mitigation effects more than theconventional one. The adaptive indices estimate an average (amongthe considered building locations and orientations) decrease indiscomfort DH going from the base case to the best case that isbetween 65% (ACA) and 80% (CEN) while the one estimated by the26 �C operative temperature limit is around 55%.

Thousands of discomfort degree-hours

Best case office - orientation South

0

5

10

15

20

25

30

Milano Roma Palermo

ASHRAE ACA ATG CEN 26

ptive indices and to the conventional set-point (26 �C) for the different locations.

S. Ferrari, V. Zanotto / Building and Environment 49 (2012) 25e3232

5. Conclusions

The adaptive approach is now considered in the main interna-tional standards about thermal comfort [1,2] and is also the topic ofinternational research projects (e.g. ThermCo [21] and Common-Cense [22]) and groups (e.g. NCEUB [23]), with a huge amount ofavailable reference data and literature.

However, there are still some open issues: first of all regardingits formulation, since each of the indices considered in this paperare connected to a specific interpretation of the overall approach,and secondly regarding the limitations on the practical applicationof the adaptive comfort theory.

The implementation of the adaptive approach in the comfortstandards is usually restricted to the assessment of the summerperformances of naturally ventilated and unconditioned buildings,in particular concerning the design phase and the service phase.Acknowledging the fact that adaptation takes place also in condi-tioned buildings provided with “perceived adaptive opportunity”[13,24], however, it looks reasonable to extend the adaptiveapproach to all kinds of buildings, whether through a universal less“forgiving” index (as in the ACA case) or through different valuesaccording to the building type (as in the ATG case). In this picture, itis important to highlight that less restrictive standards regardingthe indoor operative temperature levels could reflect in lessrestrictive HVAC system set-points being allowed and in thepassive cooling strategies being promoted, with subsequent lowerenergy consumption scenarios.

At the present time, it is unlikely to consider office buildings inthe Italian context that are completely free-running in summer, andso the best potential of the adaptive approach consists in the actualmanagement of air conditioning and cooling systems throughvariable set-points. Among the analysed indices, the only oneseligible for such an application are the ACA and the ATG ones. Thefirst index, which was developed assuming its adoption to deter-mine the indoor temperature level, is already in a useful form.However, because of the way it assigns the acceptability rangeamplitude, the ACA index is the less effective one in reducing thediscomfort DH. The ATG index, on the other hand, has a wideracceptability range because it sorts buildings according to the“perceived adaptive opportunity”: it would therefore bring tohigher energy savings while promoting buildings that give theoccupants the freedom to behave “adaptively”. The problem for itsapplication is that this index relies on a Trm form which is notsuitable for a system management, because it also takes intoaccount the current day mean temperature, which remainsunknown when the indoor temperature is set. The validity of theATG equations when using a different form of the Trm should betherefore verified before suggesting their adoption to actually setan air conditioning system.

It would be interesting to test these eligible adaptive indices inmanaging some real case-study buildings, characterized bydifferent HVAC system types and overall management philoso-phies, in order to analyze the real effects on the climatization

system operation and the actual occupants’ response in terms ofperceived thermal comfort.

References

[1] ASHRAE Standard 55 e thermal environmental conditions for human occu-pancy. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers; 2004.

[2] ISO 7730 e ergonomics of the thermal environment e analytical determina-tion and interpretation of thermal comfort using calculation of the PMV andPPD indices and local thermal comfort criteria. Geneva: International StandardOrganisation; 2005.

[3] EN 15251 e indoor environmental input parameters for design and assess-ment of energy performance of buildings addressing indoor air quality,thermal environment, lighting and acoustics. Brussels: European Committeeof Standardization; 2007.

[4] Fanger PO. Thermal comfort: analysis and applications on environmentaltechnology. Copenhagen: Danish Technical Press; 1970.

[5] Humphreys MA, Hancock M. Do people like to feel "neutral"? Exploring thevariation of the desired thermal sensation on the ASHRAE scale. Energy Build2007;39:867e74.

[6] Humphreys MA. Field studies of thermal comfort compared and applied. BuildServ Eng 1976;44:5e27.

[7] Nicol JF, Humphreys MA, Sykes O, Roaf S. Standards for thermal comfort:indoor air temperature standards for the 21st century. London: Chapman &Hall; 1995.

[8] deDear RJ, Brager GS. The adaptive model of thermal comfort and energyconservation in the built environment. Int J Biometeorol 2001;45:100e8.

[9] Nicol JF, Humphreys MA. Adaptive thermal comfort and sustainable thermalstandards for buildings. Energy Build 2002;34:563e72.

[10] Humphreys MA. Thermal comfort temperatures worldwide e the currentposition. Renew Energy 1996;8:139e44.

[11] Brager GS, deDear RJ. Thermal adaptation in the built environment: a litera-ture review. Energy Build 1998;27:83e96.

[12] McCartney KJ, Nicol JF. Developing an adaptive algorithm for Europe. EnergyBuild 2002;34:623e35.

[13] Brager GS, Baker L, Occupant satisfaction in mixed-mode buildings. In:Proceedings of the air conditioning and the low carbon cooling challengeconference; 2008.

[14] Zhang H, Arens E, Pasut W. Thresholds for thermal comfort and perceived airquality. In: proceedings of the adapting to change: new thinking on comfortconference; 2010.

[15] Baker N, Standeven M. A behavioural approach to thermal comfort assess-ment. Int J Sol Energy 1996;19:21e35.

[16] Nicol JF. Adaptive thermal comfort standards in the hot-humid tropics. EnergyBuild 2004;36:628e37.

[17] deDear RJ, Brager GS. ASHRAE RP-884 Final Report: developing an adaptivemodel of thermal comfort and preference. Atlanta: American Society ofHeating, Refrigerating and Air-Conditioning Engineers; 1997.

[18] Van der Linden AC, Boerstra AC, Raue AK, Kurvers SR, deDear RJ. Adaptivetemperature limits: a new guideline in the Netherlands. A new approach forthe assessment of building performance with respect to thermal indoorclimate. Energy Build 2006;38:8e17.

[19] Nicol JF, Humphreys MA. Derivation of the adaptive equations for thermalcomfort in free-running buildings in European standard EN15251. BuildEnviron 2010;45:11e7.

[20] Crawley DB, Lawrie LK, Winkelmann FC, Buhl WF, Huang YJ, Pedersen CO,et al. Energy plus: creating a new generation building energy simulationprogram. Energy Build 2001;27:83e96.

[21] ThermCo, homepage of ThermCo [web page]; 2008. http://www.thermco.org/cms/front_content.php%4fidcat¼57 [accessed 05.11].

[22] CommonCense, homepage of CommonCense [web page]; 2008. http://www.commoncense.info/ [accessed 05.11].

[23] NCEUB, homepage of NCEUB [web page]; 2011. http://nceub.commoncense.info/ [accessed 05.11].

[24] ThermCo. Interrelation between different comfort parameters and theirimportance in occupant satisfaction. Report for the Thermal Comfort inBuildings with Low-Energy Cooling European Project; 2009.