Adaptive Thermal Comfort

PMV assumes a fixed clo, no personal control, and a steady climate - reasonable in a sealed, air-conditioned office. In a naturally ventilated building every one of those assumptions fails. Occupants control their environment: they open and close windows, run ceiling fans, adjust blinds. They change clothing with the season and even the hour. And they hold different expectations - someone who walked in from a 34 C street does not expect, or want, a 22 C interior; 28 C with a breeze feels like relief.

When researchers ran large field surveys - de Dear and Brager's global database, and many since - the pattern was consistent: in naturally ventilated buildings, the temperature people actually voted 'comfortable' was higher in hot weather and lower in cold weather, tracking the outdoor climate. PMV, blind to all of this, systematically over-predicted discomfort in these buildings. The fix was not to patch PMV but to build a different kind of model - one that predicts comfort from the outdoor climate itself.

The stakes of getting this right are practical, not academic. If you assess a naturally ventilated building with the static model, you will conclude it is uncomfortable for large stretches of the year and 'therefore' needs air conditioning - a conclusion that is both wrong about the occupants and enormously expensive in energy and carbon. Whole regions of vernacular and low-energy architecture, from courtyard houses to verandah-wrapped bungalows, only make sense when you judge them by the right comfort model. Choosing static or adaptive is thus one of the first and most consequential decisions in any comfort study.

The adaptive model's central idea is simple and elegant: the indoor comfort temperature is a straight-line function of the recent outdoor temperature. Warmer outside, warmer the comfortable indoor temperature; cooler outside, cooler.

The outdoor input is not today's peak but a smoothed running mean (ASHRAE 55 calls it the prevailing mean outdoor temperature; EN 16798 the running mean) - a weighted average of the last several days, which captures how people acclimatise to a spell of weather rather than a single hot afternoon. The ASHRAE 55 adaptive comfort temperature is approximately:

T(comfort) = 0.31 x T(outdoor running mean) + 17.8 C.

So at a running mean of 20 C, the comfort temperature is about 24 C; at 30 C outdoors, it rises to about 27 C. Rather than one fixed set-point, you get a sloping comfort line that follows the season. A simulation that outputs indoor operative temperature can be plotted straight onto this chart to see, hour by hour, whether the building stays within the comfortable band as the climate swings - the natural way to assess a free-running design.

Why does the comfortable range widen so much? Because people adapt in three distinct ways, and the model bundles all of them.

Behavioural adaptation is the biggest and the most designable: physical actions that change the heat balance - opening a window, switching on a fan, closing a blind, removing a jacket, moving to a shaded seat, drinking something cool, slowing down. Give occupants these controls and they will steer their own comfort across a wide band.

Physiological adaptation (acclimatisation) is the slower shift in the body itself over days and weeks - people genuinely tolerate heat better after a week of it, which is why the running mean, not the instantaneous temperature, is the right outdoor input.

Psychological adaptation is expectation and perception: when people understand and control their environment, and expect it to vary with the season, they judge the same conditions as more acceptable. A naturally ventilated building where you opened the window yourself feels comfortable at a temperature that would draw complaints in a sealed office you cannot touch. Design implication: the model only holds if occupants actually have the controls - operable windows, reachable fans, adjustable shading. Take the controls away and adaptive comfort collapses back toward the narrower static range.

The adaptive comfort line is drawn with acceptability bands around it, because - as with PPD - you accept a percentage of dissatisfied. Two standards codify it.

ASHRAE Standard 55 gives an adaptive method for naturally ventilated spaces, with an 80% acceptability band (about +/-3.5 C around the comfort line, for normal use) and a tighter 90% acceptability band (about +/-2.5 C, for higher-expectation spaces). It applies when there is no mechanical cooling running, occupants can open windows, and the running mean sits roughly between 10 C and 33.5 C.

EN 16798-1 (the European standard, successor to EN 15251) uses a similar sloping line with categories I to IV, from tightest to loosest, mapping onto the same idea. Both are read the same way in practice: run the building free-running, get the hourly indoor operative temperature from the simulation, and count what fraction of occupied hours falls inside the chosen band. A design that stays within the 80% band for most occupied hours is comfortable without air conditioning - which is exactly the target for low-energy design in a mild or warm climate.

Two cautions keep the adaptive model honest. First, it has boundaries of validity: it applies to occupant-controlled, free-running spaces within a defined running-mean range, and outside those bounds (a sealed building, deep cold, extreme heat) you fall back to other criteria. Second, the band is about acceptability, not indifference - 80% acceptable still means one in five is dissatisfied, so the tighter 90% band is right for spaces with higher expectations or vulnerable occupants. Used with those caveats, the adaptive chart becomes the single clearest way to communicate a passive design's comfort to a client: one plot, the band, and the cloud of simulated hours sitting mostly inside it.

The adaptive model is not just an assessment tool - it changes what you design for. It says a building can be comfortable across a wide, sloping temperature band if it is free-running and occupants can adapt, which shifts the design goal from 'hold a fixed set-point with machinery' to 'stay within the adaptive band with passive means and give people controls'.

Concretely that means: operable windows placed for cross-ventilation; ceiling fans so occupants can add air movement (which, as the next lesson shows, extends the band further); shading and thermal mass to keep indoor operative temperature tracking - not overshooting - the running mean; and a mixed-mode strategy where air conditioning only cuts in when the free-running temperature would leave the band. In simulation terms, you run the model free-running against the local EPW file, plot operative temperature on the adaptive chart, and iterate the passive measures until occupied hours inside the band are high enough. This is the workflow behind genuinely low-energy comfort - and it is the natural bridge to how India's own standard, IMAC, adapts these ideas to Indian climates and buildings.

There is a cultural point hiding in the physics, too. The adaptive model quietly rehabilitates a great deal of vernacular wisdom - deep verandahs, high ceilings, shaded courtyards, operable shutters, the afternoon slowing-down - as rational comfort strategy rather than nostalgia. Each of those gives occupants a way to adapt, which is exactly what the model rewards. For a designer, that reframes 'low-tech' features as high-performance ones, and gives you the evidence to argue for them against the reflex to seal and cool. The next lesson shows how India measured its own version of this and turned it into a working standard.