User Tools

Site Tools


Adaptive versus Heat Balance Comfort Models

Author: Jürgen Schnieders, Passive House Institute.

Why PHI is not using adaptive comfort models

A common definition of thermal comfort is given in [ASHRAE 55] as ”that condition of mind that expresses satisfaction with the thermal environment“. There is general agreement that the main external physical parameters which determine thermal comfort are air temperature, radiative temperature, humidity and air velocity. The right combination of these parameters depends mainly on a subject’s activity and clothing.

There are currently two fundamentally different approaches in the judgement of thermal comfort: The first are the heat balance models, which basically assume that thermal comfort is achieved if the body temperature can be held in a narrow range, skin moisture is low and physiological effort of regulation is minimized ([ASHRAE 2005]). These models have been developed in huge, carefully controlled laboratory experiments. They resulted in the international standard ISO 7730 which allows for the calculation of the predicted mean vote, PMV.

In [ISO 7730:2006] three comfort categories are distinguished, with different temperatures in summer and winter due to different clothing levels. For sedentary activity, category A allows for a minimum of 21 °C, category B for 20 °C and category C for 19 °C, each at a relative humidity of 40%. Accordingly, during summer, category A allows for a maximum of 25.5 °C, category B for 26 °C and category C for 27 °C, each at a relative humidity of 60%.

The heat balance models have also been confirmed by many field studies, mostly in air-conditioned office buildings in which the predicted comfort temperatures have actually been achieved.

Since the beginning of the 1970s, so-called adaptive comfort models have slowly gained importance. These models claim that a much broader temperature range is considered comfortable by users of free-running buildings because they can adapt to varying boundary conditions. Depending on the study, comfort temperatures up to 35 °C are predicted.

The adaptive models do not rely on a physiological model of the human body, but relate e.g. comfort temperatures and ambient temperatures empirically.

In the view of the Passive House Institute, there is a number of strong arguments against the application of adaptive comfort models.

A comprehensible indication of how ambient conditions should influence the comfort temperature of people inside a building could not be found in any study. People do, of course, adjust their clothing to the interior temperatures (which in turn depend on the exterior temperatures for non-air-conditioned buildings), a fact that is already accounted for by the heat balance models. Often, vague references to physiological and psychological factors are used to justify the correlation.

It appears safe to say that physiological acclimatisation does not influence the comfort conditions, although it will influence the ability of people to withstand uncomfortably hot or cold conditions, thus changing the vote for a certain set of uncomfortable conditions as well as the width of the comfort range. If, for example, the indoor temperature in a subject’s general environment and in a particular building occasionally rises to 15 K above the comfort temperature, a 10 K temperature difference may result in a vote of +2 only, whereas the same condition would be rated +3 by a subject who rarely experiences such temperatures.

Then, there are some inconsistencies in the proposed use of adaptive models.

  • In practice, adaptive models are only applied to the upper limit of the comfort range, whereas winter comfort is usually considered to be independent of the ambient temperature.
  • The models are only applied to ‘free running’ buildings, whereas for air-conditioned buildings the heat balance models are suggested (see figure below). It appears quite strange that the thermal vote of a person should depend on whether an air conditioning system or e.g. a combination of shading and night ventilation is providing the same ambient conditions he or she is subjected to.
Fig. 1: Adaptive comfort ranges according to [de Dear 1997] for air-conditioned and free-running buildings (90% acceptability range)

  • In all adaptive models known to the author, a linear relationship between ambient temperature and indoor comfort temperature is assumed, without any reference to possible limits of validity. This appears inappropriate.

An indication for the limits of validity can be found in the comprehensive ASHRAE report RP-884 [de Dear 1997]: the preferred temperature (i.e. the temperature at which the same fraction of subjects preferred a warmer temperature and a cooler temperature) is documented for 116 buildings with and without air-conditioning. 2 of these 116 preferred temperatures are below 21 °C, 3 are above 27 °C. It may be concluded that temperatures above 27 °C are hardly considered ideal in any climate.

Contrary to this, in most studies adaptive comfort temperatures as a function of ambient conditions have been determined indirectly not only from actual votes “comfortable”, but from all votes in the experiment. In particular, for indoor temperatures far away from comfort (usually correlated with high ambient temperatures), where none of the subjects votes neutral, a linear relationship between the distance from the comfort temperature and the vote on the comfort scale is generally assumed. This procedure introduces a systematic error into the estimate of the neutral temperature which may be decisive. For example, [de Dear 1997] notes that users of air-conditioned buildings are twice as sensitive to changes in indoor operative temperature than users of naturally ventilated buildings.

The assumption of a linear relationship between indoor temperature and comfort vote in an otherwise uncontrolled field experiment is obviously wrong: Assume that people are able to adjust their clothing within a certain range, which will result in a certain range of temperatures in which they feel comfortable. Then the comfort vote will deviate from 0 only above and below this range, and the relationship of comfort vote and temperature gets non-linear, particularly in the range of interest.

Another important aspect of the RP-884 results is that air velocities in naturally ventilated buildings were often about 0.4 to 0.5 m/s, much higher than in air-conditioned buildings. This will reduce warm discomfort and thus change the votes for higher temperatures.

Part of the explanation may also be the so-called range effect. [McIntyre 1980] points out that, when exposed to a range of stimuli and equipped with a category scale, subjects will distribute the stimuli more or less evenly across this scale.

It must be concluded that, if the most comfortable temperature is below the temperature range covered in a study, this study will not allow for conclusions about this most comfortable temperature.

To summarize, the adaptive and heat balance models result in similar predictions over a relatively wide range of temperatures up to approximately 25 °C. This is also supported by [Fanger 2002], who reanalysed the data from the RP-884 database and found good agreement of the PMV predictions and the field measurements. The methodological basis that results in higher predicted comfort temperatures under certain conditions, however, appears very weak and cannot be accepted for design purposes.


[ASHRAE 2005] American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE): ASHRAE 2005 Fundamentals.

[ASHRAE 55] American Society of Heating, Refrigerating, and Air-Conditioning Engineers: ANSI/ASHRAE 55-2004, Thermal Environmental Conditions for Human Occupancy, ASHRAE, Atlanta, April 2004

[de Dear 1997] de Dear, Richard J., Gail Brager, and Donna Cooper: Developing an Adaptive Model of Thermal Comfort and Preference, Final Report, ASHRAE RP-884. Sidney/Berkeley, March 1997

[Fanger 2002] Fanger, Ole and Jørn Toftum: Extension of the PMV model to non-air-conditioned buildings in warm climates. Energy and Buildings 34, 6 (2002) 533-536

[ISO 7730:2006] DIN EN ISO 7730:2006-05, Ergonomie der thermischen Umgebung – Analytische Bestimmung und Interpretation der thermischen Behaglichkeit durch Berechnung des PMV- und des PPD-Indexes und Kriterien der lokalen thermischen Behaglichkeit (ISO 7730:2005); Deutsche Fassung EN ISO 7730:2005. Beuth-Verlag, Berlin, May 2006

[McIntyre 1980] McIntyre, D.A.: Indoor Climate. Applied Science Publishers, London 1980

See also

Articles in the member area

basics/building_physics_-_basics/thermal_comfort/heat_balance_vs_comfort_models.txt · Last modified: 2015/06/29 09:16 by mschueren