XIV INTERNATIONAL CONFERENCE ON ENVIRONMENTAL ERGONOMICS Stylianos Kounalakis ⋅ Maria Koskolou (Eds), Greece 2011

338

ASSESSMENT OF URBAN OUTDOOR THERMAL COMFORT BY THE UNIVERSAL THERMAL CLIMATE INDEX UTCI

Peter Bröde, Eduardo L. Krüger & Francine A. Rossi

INTRODUCTION

Within the recently completed European COST Action 730, the Universal Thermal Climate Index (UTCI) was made available as operational procedure for assessing the outdoor thermal environment in the core fields of human biometeorology (Bröde et al. 2010). The aim of UTCI was to characterize the thermal stress defined by the combined influence of air temperature, radiation, humidity and wind on an equivalent temperature scale (Bröde et al. 2009). The simulated dynamic response of an integrated thermo-physiological and behavioral clothing model was used to derive this scale and to establish UTCI threshold values defining different categories of thermal stress (Fig. 1).

This study applied UTCI to analyze urban thermal comfort data from a survey on pedestrians in Curitiba, South Brazil (Krüger et al. 2010). The focus was on choosing the relevant reference category for urban thermal comfort research: (i) the ‘no thermal stress’ category with 9 °C ≤ UTCI ≤ 26 °C or (ii) the sub interval 18 °C ≤ UTCI ≤ 26 °C. The latter had been shown (Bröde et al. 2010) to comply with the definition of the ‘Thermal Comfort Zone’ (TCZ) by the Glossary of Terms for Thermal Physiology (2003): ‘The range of ambient temperatures, associated with specified mean radiant temperature, humidity, and air movement, within which a human in specified clothing expresses indifference to the thermal environment for an indefinite period’.

Fig 1. Concept of UTCI as categorized equivalent temperature scale derived from the dynamic response of a thermophysiological model coupled with a behavioral clothing model.

XIV INTERNATIONAL CONFERENCE ON ENVIRONMENTAL ERGONOMICS Stylianos Kounalakis ⋅ Maria Koskolou (Eds), Greece 2011

339

METHODS

In structured interviews (Krüger et al. 2010), passers-by on a pedestrian street in Curitiba provided information on their weight, height, age and gender, on their clothing as well as on their thermal sensation, affective evaluation, thermal preference and thermal tolerance using standardized scales (ISO 10551 1995). Measurements of air temperature, mean radiant temperature, relative humidity and wind speed were performed in parallel to the interviews and were used to calculate actual UTCI values (Bröde et al. 2009).

The statistical significance of the association between the votes on the 7-point scale of thermal sensation (from -3:’cold’ over 0:’neutral’ to +3:’hot’) and on the 4-point scale of affective evaluation (0:’comfortable’, 1:’slightly uncomfortable’, 2:’uncomfortable’, 3:’very uncomfortable’) was assessed by Fisher’s Exact Test (Mehta & Patel 1983). Further, the respondents were classified into three ordered categories of thermal comfort based on their votes on thermal sensation and affective evaluation (cf. Table 1): (1) ‘cold discomfort’ (sensation<0 & evaluation>0), (2) ‘comfortable / neutral’ (sensation=0 or evaluation=0), (3) ‘warm discomfort’ (sensation>0 & evaluation>0). Predicted probabilities for those ordered three categories of thermal comfort with UTCI as explanatory variable were determined by fitting an ordinal logistic regression model (Harrell 2001).

RESULTS

The contingency table (Table 1) showed a highly significant association between thermal sensation and affective evaluation (Fisher’s Exact Test p<0.001). From 1685 interviewees, 1210 persons voted ‘neutral’ on the sensation scale or ‘comfortable’ on the affective scale with more than one third feeling both neutral and comfortable (Table 1). 188 persons were classified into the ‘cold discomfort’ group, and 287 showed ‘warm discomfort’. Table 1. Contingency table with frequencies and percentages (grey) observed for categories of thermal sensation (sl.=’slightly’) and affective evaluation. Colored cells indicate neutral or comfortable values (green), cold discomfort (blue), and warm discomfort (red), respectively.

Affective Evaluation Thermal Sensation Frequency Percentage

-3 cold

-2 cool

-1 sl. cool

0 neutral

1 sl. warm

2 warm

3 hot Total

comfortable: 0 1 0.06

34 2.02

173 10.27

597 35.43

229 13.59

74 4.39

9 0.53

1117 66.29

sl. uncomfortable: 1 8 0.47

42 2.49

103 6.11

87 5.16

109 6.47

86 5.10

30 1.78

465 27.60

uncomfortable: 2 9 0.53

11 0.65

10 0.59

6 0.36

7 0.42

27 1.60

20 1.19

90 5.34

very uncomfortable: 3

0 0.00

3 0.18

2 0.12

0 0.00

1 0.06

1 0.06

6 0.36

13 0.77

Total 18 1.07

90 5.34

288 17.09

690 40.95

346 20.53

188 11.16

65 3.86

1685 100.00

XIV INTERNATIONAL CONFERENCE ON ENVIRONMENTAL ERGONOMICS Stylianos Kounalakis ⋅ Maria Koskolou (Eds), Greece 2011

340

Fig 2. Probabilities of the individual categories of thermal comfort predicted by ordinal logistic regression analysis with UTCI as explanatory variable. Also shown are the boundaries of the thermal stress categories from the UTCI assessment scale and the sub range of UTCI values compliant to the definition of the ‘Thermal Comfort Zone’ (TCZ) by the Glossary of Terms for Thermal Physiology (2003).

Diagnostic tests on the goodness-of-fit of the ordinal logistic regression model confirmed that the proportional odds assumption underlying the model was tenable (1-df chi-square=0.07, p=0.80), that the influence of the explanatory variable UTCI was highly significant (p<0.001), and that the predictive ability of the model was reasonable as indicated by the value of 0.780 of the concordance index c of rank correlation.

As expected, the regression model predicted higher probabilities of cold discomfort for decreasing UTCI values as well as higher probabilities of warm discomfort with increasing UTCI (Fig. 2). Considering the probability curves for the individual categories of thermal comfort in relation to the UTCI assessment scale (cf. Fig. 1) revealed that the ‘Thermal Comfort Zone’ (TCZ) was a better descriptor of the region with comfortable or neutral sensations than the ‘no thermal stress’ category, since the probability for comfort was about 80% in the TCZ range, and the probabilities for both cold and warm discomfort stayed below 20% (Fig. 2).

CONCLUSION

The results suggest that UTCI may provide for useful predictions of outdoor thermal comfort summarizing the influence of air temperature, wind, humidity and radiation in sub-tropical urban areas. Though there was only little evidence advocating for a re-calibration of the UTCI assessment scale by adapting the threshold values between the different thermal stress categories, the sub range describing the ‘Thermal Comfort Zone’ may be considered as a preferable reference category in urban thermal comfort analyses compared to the wider ‘no thermal stress’ range. The semantic labels applied to the heat and cold stress categories require further validation using data from surveys and from epidemiological studies.

XIV INTERNATIONAL CONFERENCE ON ENVIRONMENTAL ERGONOMICS Stylianos Kounalakis ⋅ Maria Koskolou (Eds), Greece 2011

341

REFERENCES

Bröde, P., Fiala, D., Blazejczyk, K., Epstein, Y., Holmér, I., Jendritzky, G., Kampmann, B., Richards, M., Rintamäki, H., Shitzer, A. & Havenith, G (2009). Calculating UTCI Equivalent Temperature. In: Castellani, J.W., Endrusick, T.L. (eds.): Environmental Ergonomics XIII, Proceedings of the 13th International Conference on Environmental Ergonomics, Boston, MA, 2-7 August 2009, University of Wollongong, Wollongong, pp 49-53

Bröde, P., Jendritzky, G., Fiala, D. & Havenith, G. (2010). The Universal Thermal Climate Index UTCI in operational use. In Proceedings of Conference: Adapting to Change: New Thinking on Comfort, Cumberland Lodge, Windsor, UK, 9-11 April 2010, Network for Comfort and Energy Use in Buildings, London, 6pp

Glossary of terms for thermal physiology (2003). Third Edition revised by The Commission for Thermal Physiology of the International Union of Physiological Sciences (IUPS Thermal Commission), Journal of Thermal Biology 28:75-106

Harrell, F.E. (2001). Regression Modeling Strategies with Applications to Linear Models, Logistic Regression, and Survival Analysis. Springer, New York

ISO 10551 (1995). Ergonomics of the thermal environment - Assessment of the influence of the thermal environment using subjective judgement scales. International Organisation for Standardisation, Geneva

Krüger, E.L., Givoni, B. & Rossi, F.A. (2010). Outdoor comfort study in Curitiba, Brazil: Effects of gender, body weight and age on the thermal preference. In Proceedings of Conference: Adapting to Change: New Thinking on Comfort, Cumberland Lodge, Windsor, UK, 9-11 April 2010, Network for Comfort and Energy Use in Buildings, London, 12pp

Mehta, C.R. & Patel, N.R. (1983). A Network Algorithm for Performing Fisher’s Exact Test in r×c Contingency Tables. Journal of the American Statistical Association, 78: 427–434

Peter Bröde:

Leibniz Research Centre for Working Environment and Human Factors (IfADo), Ardeystr. 67, 44139 Dortmund, Germany, broede@ifado.de

Eduardo L. Krüger:

Universidade Tecnológica Federal do Paraná, Curitiba PR, Brazil

School of the Built and Natural Environment (BNE), Glasgow Caledonian University, UK

Francine A. Rossi:

Universidade Tecnológica Federal do Paraná, Curitiba PR, Brazil

BB O O K O F O O K O F AA B S T R A C T SB S T R A C T S

(Short and Extended form)

Editors

Stylianos Kounalakis • Maria Koskolou

XIV INTERNATIONAL CONFERENCE ON ENVIRONMENTAL ERGONOMICS Stylianos Kounalakis ⋅ Maria Koskolou (Eds), Greece 2011

2

ENVIRONMENTAL ERGONOMICS XIV / Editors: KOUNALAKIS

STYLIANOS, KOSKOLOU MARIA ISSN

National and Kapodestrian University of Athens