ASHRAE NA-04-2-3-2004 A Stochastic Approach to Thermal Comfort-Occupant Behavior and Energy Use in Buildings《乘员热舒适性行为与建筑物的能源使用的一种随机的方式》.pdf

《ASHRAE NA-04-2-3-2004 A Stochastic Approach to Thermal Comfort-Occupant Behavior and Energy Use in Buildings《乘员热舒适性行为与建筑物的能源使用的一种随机的方式》.pdf》由会员分享，可在线阅读，更多相关《ASHRAE NA-04-2-3-2004 A Stochastic Approach to Thermal Comfort-Occupant Behavior and Energy Use in Buildings《乘员热舒适性行为与建筑物的能源使用的一种随机的方式》.pdf（15页珍藏版）》请在麦多课文档分享上搜索。

1、NA-04-2-3 A Stochastic Approach to Thermal Comfort-Occupant Behavior and Energy J. Fergus Nicol Use in Buildings ABSTRACT This paper presents the results of surveys of the use of simple controls-opening of windows, the closing of window blinds, and the use of lighting, heaters, und fans-by building

2、occupants. Information is also presented on the use of air conditioning in mixed-mode buildings. The surveys were conducted in the UK, Pakistan, und throughout Europe. The data are analyzed to show how the use of each control varies with outdoor temperature. The paper discusses the application of su

3、ch results to the simulation of occupied buildings. INTRODUCTION: ENERGY, COMFORT, AND BUILDINGS In the wake of the Kyoto Agreement, there is an interna- tional imperative to reduce energy consumption and its asso- ciated anthropogenic emissions that contribute to global climate change and pollution

4、. As much as 50% of all energy is used in buildings-half of it for the provision of indoor climate control for occupant comfort (Santamouris and Wout- ers 1994). Thus, the provision of comfort has a major bearing on energy consumption and carbon dioxide emissions. The European Union Directive on the

5、 Energy Performance of Buildings (EU 2003) came into force on January 5,2003, with aims to “promote the improvement of the energy performance of buildings . and promote the convergence of building stan- dards towards those (with) ambitious levels.” In the UK, the Royal Commission on Environmental Po

6、llution (RCEP) published a report (HMSO 2000) calling for a 60% reduction in carbon emissions by 2050 in order to stabilize the atmo- spheric carbon dioxide at 550 ppm. The Energy Review (HMSO 2002) of the governments Performance and Innova- tion Unit (PIU) sets targets for energy savings through Mi

7、chael A. Humphreys increased energy efficiency (EE) in buildings at 20% by 20 10 with a further 20% by 2020. A government paper (DTI 2003) endorsed the RCEP recommendations, stating that “the UK should put itself on a path towards a reduction in carbon diox- ide emissions of some 60% from current le

8、vels by about 2050.” Naturally ventilated (NV) buildings typically use less than half as much energy as those with air conditioning (AC) (Kolokotroni et al. 1996). Encouraging the use of NV build- ings would therefore seem a good way to begin the promotion of energy efficiency. But there is no accep

9、ted way to predict the energy use of occupied W buildings or to ensure that their occupants will find their indoor climate comfortable. Of the different possible approaches to EE for existing buildings, close understanding and control of heating, lighting, and ventilation systems, together with a de

10、eper understanding of human themial comfort needs, has the greatest potential to deliver savings. Current international standards (IS0 1994) for thermal comfort are based on steady-state models developed from laboratory experiments. While acceptable for AC buildings, these standards present building

11、 designers with problems because they are poor at predicting occupant responses in NV buildings (deDear and Brager 1998), tending to overpredict levels of discomfort in variable conditions. This means that using current standards tends to encourage tightly controlled AC solutions. Part of the proble

12、m for the simulation of NV buildings is to account for the behavior of building occupants and the vari- ability of the weather. Because no clear model exists of occu- pant behavior, building simulators tend to assume best practice or some simplified model of behavior. Alternatively, they are Fergus

13、Nicol is a professor in the Oxford Centre for Sustainable Development, School of the Built Environment, Oxford Brookes University, Oxford, UK, and the Low Energy Architecture Research Unit (LEARN), London Metropolitan University. Rev. Michael Humphreys is a professor in the Oxford Centre for Sustain

14、able Development, School of the Built Environment, Oxford Brookes University, Oxford, UK, and honorary fellow at the Centre for the Study of Christianity and Culture, Regents Park College, University of Oxford. 554 02004 ASHRAE. encouraged to favor AC buildings where use of controls is dictated by t

15、he building management system. Likewise, the weather is ofen assumed to have precise characteristics. In effect, what is being simulated is past experience rather than predictions for the future. New up-to-date approaches to modeling people, buildings, and weather are needed. This paper presents a m

16、ethod to develop algorithms to predict likely occupant behavior using records of occupant behavior in field surveys of thermal comfort. The algorithm produced is expressed as the likelihood a particular control is used in terms of the physical conditions-either inside or outside the building. Occupa

17、nt behavior is assumed to be a response to those conditions. Behavior is related to the oppor- tunity given by the building for occupants to modi the indoor climate to their liking. Knowledge of occupant behavior will allow building simulations to predict the likely range of indoor conditions more r

18、ealistically. MODELING PEOPLE AND BUILDINGS According to the adaptive principle, “if a change occurs such as to produce discomfort, people react in ways which tend to restore their comfort ” (Humphreys and Nicol 1998). Building occupants use clothing, activity, or building controls, such as windows,

19、 blinds, heaters, or fans, to avoid discomfort, doing so in ways that vary between cultures. Comfort is not a given to be defined but a goal to be sought (Shove 2003F conventions, occupants, and buildings interact. Occupant control of indoor conditions has been considered to be a weak- ness of NV bu

20、ildings. Because building managers do not control the actions ofbuilding occupants, it is felt that they will increase energy consumption (e.g., Bruant et al. 1996). If behavior is directed toward the comfort “goal” of the OCCU- pants, the use of controls may not always be detrimental to energy use.

21、 Models of NV buildings need to account for occu- pant behavior (Nicol 2001) and the variability of weather. Adaptive behavior takes two forms: actions that help the subject to become comfortable in the prevailing conditions and actions that suit the environment to the subject. This paper concentrat

22、es on the latter type of behavior, but the former is also important. Figure 1 uses the data from a survey in Pakistan (Nicol et al. 1999). It shows three types of adaptive actions that affect the temperature that subjects find comfortable: the clothing insulation, the metabolic rate, and the air spe

23、ed. Also shown is the perceived skin moisture of the subjects. The adaptive process is demonstrated: as the temper- ature increases, the clothing insulation falls and the air move- ment increases (mainly due to the use of fans-see “Use of Fans” section). The metabolic rate hardly changes, but this m

24、ay be because the metabolic rate is described by the activity of the subject. Fanger and Toftum (2002), among others, have suggested that the metabolic rate for a given activity may fall Adaptive actions to change comfort temperature (Pakistan) 1.75 - 1 1 25 1 0 75 ._ . - ._ . . ,+Clothing +Air move

25、ment +Metabolic rate +Skin moisture 10 15 20 25 30 35 40 Mean globe temperature OC Figure 1 Pakistani subjects use of personal controls” to help them suit themselves to the temperature they find: change ofclothing (clo) change ofmetabolic rate (met, 9 change of skin moisture (scale: O = none, I = sl

26、ight, 2 = moderate) change of air movement (m/s) from fans and windows) ASHRAE Transactions: Symposia 555 1 o. 9 a, 0.8 o 0.7 E 8 0.6 B 0.5 o 0.4 5 0.3 o e 0.2 o. 1 O o 0 = v) .a- C O .- 12 14 16 18 20 22 24 26 28 30 32 34 36 38 Mean indoor temperature “C Figure 2 Proportion of Pakistani office work

27、ers who are comfortable at diferent mean indoor temperatures. Comfort is dejned as those who vote 3 (comfortably cool), 4 oust comfortable), or 5 (comfortably warm) on the seven-point Bedford scale as a proportion of all votes. with rising temperature as people become more economical with their acti

28、ons. The results of these adaptive actions are shown in Figure 2 (from Nicol et al. 1999), which shows that few people report discomfort in indoor temperatures between 20C and 30C. The other type of adaptive strategy is for the subject to modifj the environment in search of comfort by using the cont

29、rols that the building provides. Controls play an important part in determining occupant satisfaction with any building, particularly if it is naturally ventilated. Baker and Standeven (1 995) have characterized the richness of control provided by a building as its “adaptive opportunity,” and Leaman

30、 and Bordass (1 997) have found that occupants are more “forgiv- ing” of a building that provides good opportunities for control. Some recent studies of thermal comfort in buildings have recorded the use of building controls in the viciniy of subjects as part of the information collected. Patterns o

31、f behavior have emerged from these studies relating to the way use is made of controls by building occupants that are reported in this paper. The use of controls is clearly influenced by physical condi- tions, but their use tends to be governed by a stochastic rather than a precise relationship. Thu

32、s, there is not a precise temper- ature at which everyone opens a window, but as the tempera- ture rises, there is an increased probability that they will do so (Raja et al. 2001; Nicol et al. 1999). It has been suggested by Nicol and Raja (1997) that temperatures in buildings should be calculated “

33、both as an expected temperature and as a proba- bilistic variation about it” in order to judge whether a building will provide an acceptable environment. This suggests a stochastic model of occupant behavior, presented as an algo- rithm for the likelihood ofa control being used rather than a simple

34、odoff condition. The indoor environment and the energy use will then be modeled as a likelihood distribution rather than as a discrete value. FIELD STUDIES OF THE USE OF CONTROLS IN BUILDINGS Three databases are used here to make an estimation of the use of controls. One results from field studies o

35、f subjective comfort in offices in five European countries-Sweden, UK, France, Portugal, and Greece (McCartney and Nicol 2001). The database comprises some 4,655 full records from subjects in 25 buildings, of which 1 1 were naturally ventilated (1,649 records). Similar databases of some 5,000 result

36、s (3,600 natu- rally ventilated) wholly from the United Kingdom (UK) (McCartney et al. 1998) and 7,000 from Pakistan (Nicol et al. 1999) are also available. The databases include the comfort responses of the subjects, records of the indoor thermal envi- ronment (air and globe temperatures, humidity,

37、 and air speed), and the outdoor temperature and humidity. In addition, obser- vations were made of the subjects clothing, activity, and use of various controls (windows, window blinds, doors, lighting, fans, heaters and, where applicable, air conditioning). This paper reports on some probability al

38、gorithms relat- ing occupant behavior to indoor and outdoor temperature. The paper suggests ways in which this information could be used 556 ASHRAE Transactions: Symposia Use of fans as a function of outdoor temperature in Pakistan C .- O 5 0.4 - A/ a L OI - o 4-b 5 10 15 20 25 30 35 40 Mean outdoor

39、 temperature Figure 3 Use ofprobit analysis to illustrate how theproportion ofpeople who use fans change with outdoor temperature. Rth increasing temperature, the proportion ofpeople who use fans increases (joints). The probit line represents the best fit to thepoints using the relationship in Equat

40、ion 2. to inform the simulation of occupied buildings. The intema- tional scope of the data means that estimates can be made of the extent to which climate and culture are a factor in the rela- tionship. Analysis of the Results The data collected about the use of controls in these stud- ies is essen

41、tially binary-window open or closed, blind up or down, etc. The likelihood that any particular control is being used can be said to depend on the thermal environment and, in particular, the temperature. Thus, as the indoor or outdoor temperature increases, the likelihood that a window is opened-or a

42、 fan switched on-increases. One powerful method of analyzing such processes is probit analysis (Finney 1964). The probit procedure assumes that the likelihood of an event happening increases as the “intensity” of the stimulus (for instance, temperature) increases. The traditional model assumes an un

43、derlying Gaussian relationship between the stimulus and the probabilistic response, but other causal distri- butions can be used (Finney 1964). The model used in this study is the Logit model. Logb I (1 -p) = U + bx (1) where Log I (1 -p) = the logit function and is assumed to be linearly dependent

44、on the value of x P = the probability of an event having taken place aandb = constants X = a variable (in this case a temperature or thermal index). The values of a and b are determined by performing a probit regression for the Logit function using the cases where a particular event has occurred (wi

45、ndow opened, blind down, etc.) at different values ofx. A weighted regression analysis of the Logit against the values of x then gives estimates of the values of the constants a and b. Once the values of the constants are known, a curve can be constructed linking the index x and the probability of t

46、he particular event using Equation 2, which is derived from Equa- tion l above. 1 (2) p = e(a+w / (1 + e(a+w The method has often been used in thermal comfort stud- ies to investigate the changing incidence of discomfort with increases in temperature or some other environmental index (see, e.g., Web

47、b 1964, Fanger 1970, and many subsequent studies). Figure 3 is an example of how probits can be used to analyze some data from the Pakistan survey to show how the probability that people will use fans changes as the outdoor temperature rises (see also Figure 9b). The curve is the probit regression l

48、ine, which comes from the data represented by the points, which represent the mean proportion of fans running binned for every two degrees on outdoor temperature. . Commonly used statistical packages such as SPSS can be used to calculate the values of a and b. ASHRAE Transactions: Symposia 557 PRESE

49、NTATION AND DISCUSSION OF THE RESULTS The controls whose use has been analyzed here are the opening of windows, the drawing of window blinds or curtains, the use of lighting, the use of heating, and the use of fans. The use of air conditioning (AC) in mixed-mode build- ings has been separately analyzed. The results of the analysis are displayed in Figures 4 through 1 1. These analyze the use of these controls against both indoor and outdoor tempera- tures. In most cases, the correlation with indoor temperature is similar to that with outdoor temperature (