Despite 60 years of research, the relationship between thermal sensation, comfort and acceptability is poorly understood. Laboratory studies only tell half the story. So can research in real buildings fill the gap?
Thermal comfort research is in crisis. So says comfort guru Mike Humphreys, who believes there is a serious conflict between the results of laboratory-based studies and data obtained from real buildings.

It is very difficult to associate conditions in fully operational and occupied buildings with laboratory-based models. Yet it is the laboratory-based models that have driven national and international thermal comfort standards.

As clients begin to recognise the relationship between comfort and productivity, it is increasingly important that predictive models are based on field studies. And for that to happen we need to develop cost-effective and efficient means of obtaining and processing information from the real world.

A recent BSRIA project, partly funded by the DETR, has developed a means of obtaining

The results can be related to physical parameters – such as temperature, humidity and air velocity profiles – obtained from building energy management systems.

BSRIA has developed and validated such software on field trials conducted in two buildings located in the south-east of England. The first building (Building A) was sparsely occupied and the second (Building B) was densely occupied.

Comfort field trials

Fifteen participants in Building A took part in the thermal comfort field study, which was conducted during an autumn/winter season. Over 500 computer-based responses were collected over a period of four weeks. (Analysis of this data was found to correlate with the results of other field studies conducted in similar environments).

The software was designed to be installed on a pc in about ten minutes, and the questionnaire itself, which popped up regularly on occupants' screens, took about a minute to answer.

Interestingly, the computer-based questionnaire caused less distraction than a paper-based questionnaire. Nearly 70% of the participants stated that the questionnaire did not distract them (see table 1 overleaf).

Interestingly, the computer-based questionnaire caused less distraction than a paper-based questionnaire

The results obtained demonstrated the success of the software. Formal feedback sessions with the participants were conducted a week after the completion of the study. The comments received from this pilot session suggested that some changes needed to be made to the software prior to the larger scale study of Building B.

A spring/summer study was conducted in Building B. Forty participants took part in this study. Over 2000 responses were received over a five-week study period. As with the autumn/winter study, the results of the spring/summer study were comparable with other field studies conducted in similar environments but using a different medium of data collection.

In-depth analysis of the larger (spring/summer) study showed the derived neutral temperature was 20.5°C (the indoor dry bulb temperature most likely to produce the response zero or neutral on the predicted mean vote (pmv) comfort voting scale).

PMV is based on laboratory studies and forms the basis of ISO 7730. This states that the optimum temperature in a typical office environment (under a specified set of conditions) is 24.5°C, a value which differs by 4°C from the calculated neutral temperature.

This study, like many others, highlights the discrepancies between field and laboratory-based studies. Moreover, the differences between the various measures of thermal comfort, such as the neutral and preferred temperatures, show that relationships between thermal sensation, comfort and adaptability are still not clear.

Results obtained from the two field studies indicated less sensitivity to temperatures away from the neutral temperature than theory predicts. Only 11% of the variance in the comfort vote was explained by the indoor dry bulb temperature; this rose to 12% when other physical and personal parameters were taken into account.

Differences in the comfort votes by gender were significant. It appears that female participants took action to avoid thermal discomfort.

So what can we conclude from this work?

The primary conclusion that can be drawn from the two field studies is that the designed and implemented Thermal Comfort Field Data Acquisition software is an effective mechanism for administering thermal comfort studies. The results were consistent with the results of prior field studies of thermal comfort conducted with more traditional survey methods.