Diurnal and weekly variation of anthropogenic heat emissions in singapore

14 Figure 2.2 Mean hourly heat release from vehicles in a Vancouver suburb…… 15 Figure 2.3 Weekly and daily normalised electricity load profiles for Toulouse... LIST OF SYMBOLS E k hour

DIURAL AD WEEKLY VARIATIO OF ATHROPOGEIC

HEAT EMISSIOS I SIGAPORE

QUAH KHAI LI AE

B.A (Hons.), University of Canterbury

of work

I would also like to thank the following people and organisations who have assisted in the provision of data, namely, Ms L.T Soh, Ms S.P Lim, Ms C Ang, Mr D Phoon and Mr M Chew from the Land Transport Authority, Professor L Norford from the Massachusetts Institute of Technology, the Ministry of National Development, Mr E Toh, Mr J Chin, Ms L Ganesh from the Energy Market Authority, Ms C Ng from the West Coast Town Council, Mr H.J, Cheng from the Bukit Panjang Town Council, Mr Goh from Clementi Community Centre, Ms M Chung from Faith Montessori Centre,

Ms I Leong from the National University of Singapore (Office of Estate Development) and Mr Y.Y Choon from the National Environment Agency

I am also thankful to the many friends who have supported and encouraged me through this challenging period

Most importantly, none of this would have been possible without the love, patience and encouragement of my parents and brother

TABLE OF COTETS

ACKOWLEDGEMETS……… i

TABLE OF COTETS……… ii

SUMMARY……… v

LIST OF TABLES……… vii

LIST OF FIGURES……… ix

LIST OF SYMBOLS……… ……… xii

CHAPTER 1: ITRODUCTIO……… 1

1.1 Context and Rationale……… 1

1.2 Objectives……… 5

1.3 Structure of Thesis……… 6

CHAPTER 2: REVIEW OF RESULTS FROM AD METHODS USED I PAST ATHROPOGEIC HEAT FLUX DESITY STUDIES……… 7

2.1 Magnitude of Anthropogenic Heat Flux Densities in Different Cities……… 7

2.2 Methods for Estimating Anthropogenic Heat Flux Density…… 12

2.2.1 Overview of inventory-based approach ……… 12

2.2.2 Inventory-based bottom-up modelling approach ……… 13

2.2.3 Inventory-based top-down modelling approach……… 20

2.2.4 Energy budget closure approach.……… 25

2.3 Discussion of Inventory-Based and Energy Budget Closure Approaches ……… 26

2.3.1 Comparisons between the inventory-based and energy budget closure approaches.……… 26

2.3.2 Comparisons between the bottom-up and top-down modelling approaches……… 27

CHAPTER 3: BACKGROUD IFORMATIO O STUDY AREAS…… 29

3.4 City-wide Energy Consumption Patterns and Trends.……… 34

3.5 Location and Description of the Selected Study Areas………… 37

CHAPTER 4: COCEPTUAL FRAMEWORK AD ITS APPLICATIO

4.3 Anthropogenic Heat Emissions from Buildings.……… 54 4.3.1 Estimation of electricity consumption: EUI approach… 59 4.3.2 Estimation of electricity consumption: Household-cum-

common area electricity usage approach……… 61 4.3.3 Estimation of electricity consumption: Mixed approaches 65 4.3.4 Mapping electricity load profile data to electricity

consumption estimates……… 66

4.4 Anthropogenic Heat Emissions from Human Metabolism……… 69 4.4.1 Calculation of metabolic heat production values……… 70 4.4.2 Estimation of hourly population……… 71

5.2 Anthropogenic Heat Emissions from Buildings……… 78

5.3.1 Diurnal variation of QM……… 81

5.3.2 Weekly variation of QM……… 83

5.4 Total Anthropogenic Heat Flux……… 84

5.4.1 Diurnal variation of QF……… 84

5.4.2 Weekly variation of QF……… 87

CHAPTER 6: DISCUSSIO……… 89

6.1 Overview……… 89

6.2 Factors Influencing Diurnal and Weekly Estimates of QF…… … 89

6.2.1 Influence of traffic volume……… 89

6.2.2 Influence of building electricity consumption patterns.… 93 6.2.3 Influence of population numbers……… 98

6.2.4 Influence of QV, QB and QM on diurnal and weekly QF estimates……… 100

6.3 Comparison of QF against Net All-Wave Radiation Flux Density and UHI Intensities……….……… 106

6.3.1 Comparison with net-all wave radiation flux density…… 107

6.3.2 Implications for UHI intensities……… … 108

CHAPTER 7: COCLUSIO……… 111

7.1 Summary of Results……… 112

7.2 Recommendations for Future Research and Final Thoughts…… 114

REFERECES……… 116

APPEDIX A: RAW DATA FOR CALCULATIG QF ……… 126

APPEDIX B: EQUATIO FOR CALCULATIG EERGY USE ITESITY……… 133

APPEDIX C: GROSS FLOOR AREA CALCULATIOS……… 135

APPEDIX D: ESTIMATIO OF BUILDIG ELECTRICITY COSUMPTIO USIG MIXED APPROACHES … 138

APPEDIX E: ELECTRICITY LOAD PROFILES……… 144

APPEDIX F: BODY SURFACE AREA CALCULATIO ……… 154

APPEDIX G: APPROACH FOR ESTIMATIG WORKER, SHOPPER AD PEDESTRIA POPULATIOS AT COM… ……… 156

SUMMARY

The anthropogenic heat flux density, QF, is unique to urban environments and can

potentially be an important component of the energy balance of the building-air volume, particularly in densely populated cities with high energy demands Through

its role in the energy balance, QF will influence a city’s thermal environment, ambient

air quality and other attributes of the urban climate system, albeit to different extents

as anthropogenic heat emissions are not uniform across the city

The present study estimates the temporal variability of QF for three common land use types found in Singapore, namely, (i) commercial, (ii) high-density public housing and

(iii) low-density private housing, between October 2008 and March 2009 QF is estimated by considering separately the three major sources of waste heat in urban environments, which are heat release from vehicular traffic, buildings and human

metabolism, respectively These components of QF are calculated by applying a combination of the top-down and bottom-up modelling approaches of energy consumption within the local setting In order to place the emission of this

anthropogenic heat in a wider context, QF is compared against other components of the energy balance of the building-air volume and with urban heat island intensities observed in Singapore

Results show that over a 24 hour period, magnitudes of mean hourly QF reach maximum values of 113 W m−2 in the commercial, 18 W m−2 in the high-density public housing and 13 W m−2 in the low-density private housing areas, respectively Buildings are found to be the major source of anthropogenic heat in each study area,

contributing to between 49–83% of QF on weekdays and 46–82% on weekends The

spatial and temporal variations of QF are attributed to differences in traffic volume,

building energy consumption and population density Comparisons show that QF is

equivalent to 87% of the net-all wave radiation flux density at the commercial site whereas this percentage is considerably smaller for the two residential areas

The detailed QF data obtained from the present study can be included in urban climate models to allow researchers to quantify and gain further insights into the way in which

QF affects the local climate of cities and its potential contribution to urban heat islands

LIST OF TABLES

Table 2.1 Summary of mean anthropogenic heat flux density (QF) for different

cities ……… ……… ……… 8 Table 2.2 Net heat combustion and fuel density values of vehicles in a

Vancouver suburb according to fuel type ……… … 15 Table 4.1 Inventory-based modelling approaches used to calculate the

magnitude of the components of QF……… 43 Table 4.2 Fuel type according to vehicle class……… 51 Table 4.3 Net heat combustion and fuel density of unleaded petrol and diesel… 51 Table 4.4 Mean fuel economy of different vehicle classes in Singapore……… 52

Table 4.5 Representative values of the energy used per vehicle according to the

vehicle class……… 53

Table 4.6 Building categories and sub-categories and the corresponding

number of buildings……… ………… 58 Table 4.7 Average EUIs of non-residential buildings according to building

category/sub-category……… 60 Table 4.8 Mean monthly electricity consumption per household according to

household type…….……… 62 Table 4.9 Mean hourly normalised common area electricity consumption

according to block type……… 65 Table 4.10 Temporal scale at which electricity consumption estimates of the

various building categories/sub-categories were available……… … 67 Table 4.11 Building categories/sub-categories and household types without

representative load profile data and their matched building category/sub-category and household type with available load

Table 4.12 Metabolic heat production values of activity types most commonly

carried out in COM, HDB and RES……… 71

Table 4.13 Estimated population of COM, HDB and RES at any hour of the

‘sleep’ and ‘active’ periods, on weekdays and weekends………… 72 Table 4.14 Total number of non-working residents, workers, shoppers and

pedestrians for COM on weekdays and weekends……… …… 74

Table 6.1 QF magnitudes of various city-centres/commercial areas … ……… 101

Table 6.2 QF magnitudes of various suburban residential areas……… 104 Table 6.3 Weekday and weekend mean hourly QF values for COM, HDB and

RES……… 106

LIST OF FIGURES

Figure 1.1 Growth of urban population by region, 1950–2050……… 1

Figure 2.1 Temporal profiles of traffic counts for major and minor roads in a

Vancouver suburb……… 14 Figure 2.2 Mean hourly heat release from vehicles in a Vancouver suburb…… 15 Figure 2.3 Weekly and daily normalised electricity load profiles for Toulouse 17

Figure 2.4 Mean hourly heat release from human metabolism in a Vancouver

suburb……… 19 Figure 2.5 Diurnal metabolic rates representative of Greater Manchester…… 20 Figure 2.6 Daily normalised traffic flow profiles for various US cities and

states, Toulouse and the Gyeong-In region of South Korea ……… 22 Figure 2.7 Weekly normalised traffic flow profile for Toulouse……… 23

Figure 2.8 Representative summer and winter electricity load profiles for

various service regions across the US…… ……… 24 Figure 3.1 Location map of Singapore……… ………… 29 Figure 3.2 Land use map of Singapore in 1958 and 2005……… ……… 31 Figure 3.3 Climograph of Singapore based on measurement carried out

between 1982 and 2008……… 32 Figure 3.4 Distribution of Singapore’s end-use energy consumption in 2007

according to consumption sector……… 35 Figure 3.5 Distribution of energy consumed in Singapore for 2006 according

to types of energy……… 35

Figure 3.6 Distribution of electricity consumed by a typical household in

Singapore according to type of appliance……… 37 Figure 3.7 Location of study areas……… ……… 38 Figure 3.8 Land use patterns and spatial extent of the three study areas……… 39 Figure 3.9 Photographs of land-use characteristics at the three study areas…… 40

Figure 4.1 Mean diurnal traffic count profiles for COM, HDB and RES… … 47

Figure 4.2 Vehicle classes considered in the present study and their

distribution as a proportion of the total number of vehicles registered in Singapore……… 49 Figure 4.3 Selected road junctions within COM……… 54 Figure 4.4 Variation of monthly electricity consumption of different

household types……… 62 Figure 5.1 Diurnal variation of QV for COM, HDB and RES……… 76 Figure 5.2 Diurnal variation of weekday and weekend QV for COM, HDB

and RES……… ………… 77 Figure 5.3 Weekly variation of QV for COM, HDB and RES……… 78

Figure 5.4 Diurnal variation of QB for COM, HDB and RES……….………… 79 Figure 5.5 Diurnal variation of weekday and weekend QB for COM, HDB and

RES……… 80 Figure 5.6 Weekly variation of QB for COM, HDB and RES ……….…… 81

Figure 5.7 Diurnal variation of QM for COM, HDB and RES ………… …… 82 Figure 5.8 Diurnal variation of weekday and weekend QM for COM, HDB and

RES……… 83 Figure 5.9 Weekly variation of QM for COM, HDB and RES ……… … 84

Figure 5.10 Diurnal variation of QF, Qv, QB and QM for COM, HDB and

Toulouse, COM, HDB and RES……… 91

Figure 6.2 Weekday and weekend mean hourly traffic counts for COM, HDB,

RES and Toulouse……… 92

Figure 6.3 Winter-time total diurnal energy use profile of offices, commercial

buildings and hotels in Tokyo and the diurnal building energy use profile of COM……… 95

Figure 6.4 Diurnal building energy use profiles of HDB, RES and the

residential areas in the Greater Manchester region……… 96

Figure 6.5 Winter-time total diurnal QF profile of offices, commercial

buildings and hotels in Tokyo and the diurnal QF profile for COM 100

Figure 6.6 Mean hourly electricity consumption of an

office-cum-shopping-centre building and hotel in COM on weekdays and weekends…… 102 Figure 6.7 Mean hourly QF values for suburban Vancouver, HDB and RES… 103

Figure 6.8 Mean hourly QF, QB, QV and QM values for suburban Vancouver,

HDB and RES……… 105 Figure 6.9 Mean hourly net all-wave radiation flux density at RES and

anthropogenic heat flux profiles for all 3 study areas……… 108

LIST OF SYMBOLS

E k hourly electricity consumption of building k W

ORC estimated total office and retail space in COM m2

ORN estimated total number of office workers and retail staff in the

planning zones of COM

persons

ORS total office and retail space in the planning zones of COM m2

QM human metabolism anthropogenic heat flux density W m-2

QV vehicular traffic anthropogenic heat flux density W m-2

RN estimated total number of retail staff in the planning zones of

COM

persons

RS total retail space in the planning zones of COM m2

WD estimated weekday worker population at COM persons

WE estimated weekend worker population at COM persons

Chapter 1

ITRODUCTIO

1.1 Context and Rationale

The world is undergoing the largest wave of urban growth in history (United Nations [UN], 2008) During 2008, it saw for the first time more than half of its population living in urban areas, and this figure will surge to almost 70 per cent in 2050 mainly due to natural growth and migration (UN, 2008) A large part of this expected future urban growth will be absorbed by cities of the less developed regions (Figure 1.1), most of which are located in the sub(tropics) (defined as the region between 23.5–35 ºN and 23.5–35 ºS)

Figure 1.1 Growth of urban population by region, 1950–2050

(Source: UN, 2008)

The pace of urbanisation in less developed sub(tropical) cities is often so rapid that local governments do not have sufficient time to react appropriately and adequately to this sudden influx of people into the cities in terms of proper land use planning, the

Less developed regions

More developed regions

building of new urban infrastructure (e.g houses and roads) and the provision of basic urban services, such as power, water and sanitation (UN, 2008) This consequently brings about a myriad of climate-related environmental problems (e.g poor dispersion

of pollutants and high levels of heat stress) together with a list of social and economic problems Sadly, many of these cities are in a weak position to handle their environmental problems in a sustainable manner because they lack the adequate financial, technical and scientific resources to do so However, urban growth in most cities of the less developed sub(tropical) countries is still at its initial stages and this presents opportunities for the governments of these countries to develop and implement effective strategies, such as climate-sensitive urban designs to solve and mitigate the environmental problems brought about by urbanisation

In order to effectively tackle the climate-related environmental problems faced by less developed sub(tropical) cities, it is essential to first have a better understanding of the nature of their physical climatology Fundamental to this understanding is the building-air volume formulation of the urban (surface) energy balance (UEB) which is expressed by Oke (1987) as:

where the flux densities are: Q* net all-wave radiation, QF anthropogenic heat, QH

turbulent sensible heat, QE turbulent latent heat, ∆QS net heat storage, and ∆QA net heat advection Positive values on the left-hand side of equation (1.1) are inputs to the system, while positive values on the right-hand side are outputs or losses

The nature of the physical climatology of a sub(tropical) city was first studied by

Oke et al (1992) Thereafter about eight similar energy balance studies with results

published in English peer-reviewed literature have been carried out (Roth, 2007)

However, none of these studies has examined the term, QF,in any detail Essentially,

QF is the energy released from human activities, such as via vehicle fuel combustion, heating and cooling of buildings, industries and human metabolism This particular term has often been neglected in UEB studies regardless of whether they were carried out in sub(tropical) cities or in other regions, primarily because of its modest magnitude relative to the other fluxes of the UEB and also due to the difficulties in

accessing the magnitude of QF (e.g Garcia-Cueto et al., 2003; Spronken-Smith et al.,

2006)

Nevertheless, the interest in quantifying QF has been growing over the last couple of

years, especially for densely populated mid-latitude cities such as Tokyo, Seoul and London This trend is perhaps due to an increased awareness of the findings of past anthropogenic heat flux density studies which have shown that under certain

conditions, QF can potentially be an important or even dominant component of the

UEB and that it can potentially influence urban climates For example, Grimmond (1992) observed that more than 10% of the wintertime energy input (Q* + QF) in

Vancouver was accounted for by QF, while Inchinose et al (1999) estimated QF in central Tokyo to reach 400 W m−2 in the day during summer months and as much as

1 590 W m−2 during the early morning hours of winter Temperature simulations by Fan and Sailor (2005) suggested that the nighttime heat island effect of Philadelphia,

which was augmented by about 2 ˚C – 3 ˚C in winter, could be attributed to QF In

fact, the best documented example of the impact that QF has on urban climates is

perhaps the urban heat island (UHI) effect (a phenomenon where nocturnal temperatures in the urban areas are observed to be higher relative to their rural,

undeveloped surroundings) This is reasonable since QF constitutes an additional

source of heat that urban areas are subject to, which is absent in rural areas Despite anthropogenic heating being just one of the many factors contributing to the UHI effect, it can play a relatively major role in the development of the nighttime UHI, particularly during winter as exemplified in the above-mentioned case of Philadelphia

In view of an increasingly urbanised world, the effects of the UHI phenomenon on urban energy consumption, human comfort and health will take on an increasingly important role within the context of developing more sustainable cities Overall,

findings from past studies indicate that QF is most pronounced during wintertime and

in cities with a high population density and intensive levels of commercial activity

(e.g Inchinose et al., 1999; Sailor and Lu, 2004; Pigeon et al., 2007)

While an increasing number of QF studies are being carried out in mid-latitude cities,

the same cannot be said of cities located in other regions In order to obtain a more comprehensive understanding of anthropogenic heat emissions and their effects on the urban thermal environment and other attributes of the urban climate system, it is

important to bridge the gaps in our knowledge of QF Furthermore, most

observational programmes neglect QF and treat ∆QS as the energy balance residual

from Q*, QH and QE (e.g Spronken-Smith et al., 2006) One of the main problems associated with such an approach is that the estimates of ∆QS will be wrong if

magnitudes of QF are large (e.g Christen and Vogt, 2004; Pigeon et al., 2007) Therefore, in consideration of this problem, it is important to explicitly quantify QF so

that more accurate estimates of ∆QS can be used to model and/or explain the various

urban climate processes and phenomena of interest (e.g boundary layer development

and UHI effect) In response to the reasons provided above for studying QF, it is

appropriate to conduct a detailed investigation of the anthropogenic heat flux density for a sub(tropical) city as a start, given that most of the expected future urban growth will take place in this region

The tropical city-sate of Singapore (1˚22´N, 103˚48´E) is an ideal study area for the

purpose of investigating QF for a sub(tropical) city as it is not only highly

commercialised, but also densely populated (according to data from the UN [2009], Singapore is the world’s second most densely populated country with a population density of 6 500 person km−2); thus, its QF magnitude can potentially be quite significant The findings of this study will serve as a valuable addition to the small

knowledge ‘bank’ of (sub)tropical urban climate studies, particularly (sub)tropical QF

studies As mentioned above, building up this knowledge ‘bank’ by means of

conducting QF related studies in cities across different climatic zones will allow for a

more holistic understanding of the nature of anthropogenic heat emissions All this knowledge would eventually aid researchers and policymakers develop more effective strategies to prevent or mitigate climate-related environmental problems associated with the rapid urban growth that cities in the less developed sub(tropical) region are

facing

1.2 Objectives

The main objectives of this research are:

• To estimate the anthropogenic heat flux density in Singapore using the based energy consumption method

inventory-• To observe and document the diurnal and weekly variations of the anthropogenic heat flux density for the three land use types in Singapore, which are (i) commercial, (ii) high-density public housing, and (iii) low-density private housing

• To assess the potential impact of the estimated anthropogenic heat flux density on the urban climate in Singapore by obtaining indications of its contribution to the urban (surface) energy balance and UHI intensities observed

conceptual framework used to estimate QF The diurnal and weekly variations of QF

are presented in Chapter Five and the main findings are discussed in Chapter Six Lastly, a summary of the thesis and recommendations for future research on this topic are given in Chapter Seven

Chapter 2 REVIEW OF RESULTS FROM AD METHODS USED I PAST

ATHROPOGEIC HEAT FLUX DESITY STUDIES

2.1 Magnitude of Anthropogenic Heat Flux Densities in Different Cities

QF has been estimated for a range of different cities, most of which are located in latitude regions as mentioned in Chapter 1 Results from these studies indicate that

mid-the magnitude of QF varies greatly between and within cities, depending on per capita energy use, population density, meteorological conditions and background climate

According to Oke (1988), for example, the mean annual magnitude of QF for large

cities ranges from 20 W m−2 to 160 W m−2 QF values for US cities are in the range of

20 W m−2 to 40 W m−2 in summer and 70 W m−2 to 210 W m−2 in winter (Taha, 1997)

Table 2.1 summaries the QF values from past anthropogenic heat flux density studies

Local space and time variations are masked by the annual average QF values listed in Table 2.1 Therefore, where possible, the Table also includes seasonal values For example, in Klysik’s (1996) study, the entire urban area of Lodz was estimated to

have a mean annual QF value of 29 W m−2, but in the winter and summer months, its mean monthly values were 54 W m−2 and 12 W m−2, respectively In addition, for the

same study, the estimated values of QF for the densest area of Lodz differ substantially from the ones calculated for a suburban residential area, as also shown in Table 2.1

The effects of population density on QF can be observed, with particular reference to the cities of Fairbanks and Reykjavik Although both cities are located on the same

latitude (64˚N), their mean annual QF values differ substantially Steinecke (1999) postulated that the higher population density of Reykjavik was the main contributing

factor to its high QF value

Besides population density, the magnitude of QF also depends on energy consumption The latter depends on many factors including climate (which influences the demand for space heating and cooling), urban transportation system, building type and level and type of economic activity Table 2.1 illustrates some of these points For

example, the higher summer time QF values (41 W m−2) observed in Hong Kong relative to its winter-time values (32 W m−2) were attributed to energy consumption levels being higher during the summer period compared to that during the winter period, largely due to strong demands for air-conditioning in the summer months (Newcombe, 1976) Thus, it is important to consider the unique interactions of space, time, energy consumption and population density when attempting to understand the magnitudes and variations of the anthropogenic heat flux density in a city

Overall, the magnitude of QF has been evaluated for various purposes Grimmond (1992), Christen and Vogt (2004) and Offerle et al (2005) estimated QF at the local scale for the purpose of incorporating this term into the suburban energy balances of

Vancouver, Basel and Lodz respectively Fan and Sailor (2005) calculated QF in order

to include it in their simulation of Philadelphia’s temperature fields; the simulations

suggested that QF contributed 2 ˚C–3 ˚C to the night-time heat island in Philadelphia Regardless of the purpose for estimating QF, it is apparent from Table 2.1 that apart from Hong Kong, Mexico City and Singapore (present study) which are located in the tropics (i.e 0–23.5 ˚N and 0–23.5 ˚S), all of the other cities are located in mid-latitude regions

2.2 Methods for Estimating Anthropogenic Heat Flux Density

Unlike the other terms of the UEB, QF cannot be directly measured Therefore, two

approaches have been developed to estimate the anthropogenic heat flux density in urban environments, which are the inventory-based and energy budget closure approaches The former can be further divided into the bottom-up and the top-down modelling approaches Table 2.1 also lists the approaches that were used to estimate

the values of QF for various cities Each of these approaches will be described and

discussed in greater detail below

2.2.1 Overview of inventory-based approach

Estimating QF from inventories of energy consumption is a classical approach and has

been applied in many anthropogenic heat flux density studies (e.g Grimmond, 1992;

Klysik, 1996; Sailor and Lu, 2004; Pigeon et al., 2007; Smith et al., 2009) since its

earliest application, which is likely to be that by Torrance and Shum (1975) who

estimated the mean annual QF of an unspecified densely populated city to be

83.7 W m−2 This particular approach relies on utility scale consumption data or data

obtained from energy consumption surveys Most studies that estimate QF using this approach assume that all energy consumed is instantaneously released into the

atmosphere as sensible waste heat (e.g Grimmond, 1992; Inchinose et al., 1999; Sailor and Lu, 2004; Lee et al., 2009) Although QF can also be released into the atmosphere in the form of latent heat, only very few studies using the inventory-based

approach, partition QF into the sensible and latent heat flux density components (e.g

Pigeon et al., 2007) due to the difficulties in estimating their magnitudes The overall idea of the inventory-based approach is to estimate QF by adding up the amount of waste heat produced by the main anthropogenic heating sources in urban

environments, which are vehicles, buildings (includes commercial, residential, industrial, institutional and all other building types and structures) and human metabolism An elaboration on the manner in which the bottom-up and top-down modelling approaches have been used to calculate the magnitude of these three

components of QF is given in the following sections

2.2.2 Inventory-based bottom-up modelling approach

The bottom-up modelling approach relies on estimating energy consumption at scales (e.g individual buildings) in order to scale the information up to larger scales of interest (e.g city-scale) Studies which have employed this approach include those of

small-Grimmond (1992), Heiple and Sailor (2008) and Smith et al (2009).

The quantification of heat release from vehicles based on this approach has been

rather simplistic in some QF studies, especially in the earlier studies due to a lack of detailed input data, such as vehicle counts over the time period of interest and heat release for different vehicle types For example, Klysik (1996) had to use hourly records of fuel purchases to determine the amount of fuel consumed by vehicles in his study area located in Lodz as he did not have access to detailed traffic profile data

Similarly, due to inadequate data, Inchinose et al (1999) had to assume that traffic

density was spatially uniform throughout Tokyo However, there have been several studies which have examined heat release from vehicles in a more rigorous manner by using the refined bottom-up modelling approach developed by Grimmond (1992) This refined approach expresses heat release from vehicles as a function of the number of vehicles travelling within the study area, vehicle distance travelled and

energy used per vehicle It has been applied by Grimmond (1992) and Smith et al

(2009) to estimate heat release from vehicles in Vancouver and Greater Manchester respectively In these two studies, detailed traffic count data was obtained for the major and minor roads identified in the study areas An example is provided in Figure 2.1 which shows the mean daily traffic count profiles obtained for Vancouver in Grimmond’s (1992) study Using traffic count and road length data, the total distance travelled by all vehicles in the study areas were calculated and subsequently multiplied by the respective energy used per vehicle data to determine the total amount of anthropogenic heat released

Figure 2.1 Temporal profiles of traffic counts for major and minor roads in a

Vancouver suburb (Source: Grimmond, 1992)

According to Grimmond (1992), the energy used per vehicle is dependent on its fuel economy and the net heat combustion and density of its fuel type used Grimmond (1992) determined a representative fuel economy value of 11 253 m litre−1 The net heat combustion and fuel density values corresponding to different fuel types are given in Table 2.2

Table 2.2 Net heat combustion and fuel density values of vehicles in a Vancouver

suburb, according to fuel type

(x106 J kg−1)

Density range†(kg litre−1)

The mean hourly distribution of heat release from vehicles which was eventually calculated is shown in Figure 2.2 From this Figure, it was observed that in the case of Vancouver, the maximum mean hourly heat release from vehicles is approximately

7 W m−2, recorded at about 18:00 h In contrast, the mean daily heat release from vehicles for Greater Manchester was as much as 70 W m−2 at major traffic junctions

of the motorway network and it accounted for about one-third of the total daily

anthropogenic heat flux density (Smith et al., 2009)

Figure 2.2 Mean hourly heat release from vehicles in a Vancouver suburb

(Source: Grimmond, 1992)

In order to estimate heat release from buildings, the bottom-up modelling approach

has been used in most QF studies (e.g Torrance and Shum, 1975; Klysik, 1996;

Inchinose et al., 1999; Smith et al., 2009) This approach typically involves the use of

statistical data to estimate the energy consumption characteristics of a group of buildings that is considered to be representative of the building stock found within the city/area of interest The acquired data is then combined with detailed representations

of the spatial distribution of buildings types found within the city of interest and their corresponding typical energy consumption data In the absence of detailed data, the results produced from most of the earlier studies were fairly simple in terms of their spatial and temporal variations For example, Torrance and Shum (1975) were forced

to assume that industrial energy consumption rates were constant throughout their study period and that space heating rates were simply proportional to ambient air temperature

Later studies which used the bottom-up modelling approach to estimate heat release from buildings produced more refined results, both spatially and temporally as the

required data was more readily available For example, Inchinose et al (1999)

developed energy load profiles for nine building types representative of the residential and commercial sector buildings in Tokyo based on building load survey data These load profile data were subsequently linked to a geospatial building database to produce a spatially and temporally detailed representation of building-related heat

release Another example of this approach is the work of Pigeon et al (2007) who

developed diurnal and weekly electricity load profiles using real-time hourly and daily electricity consumption data (based on temporal scales, this particular example can be considered as a bottom-up modelling approach because hourly and daily data are scaled-up to diurnal and weekly temporal scales) Figure 2.3 illustrates these load profiles which were applied together with statistical information on the typical mean

electricity consumption of the building types and their spatial distribution, to determine the magnitude of building-related heat releases for a particular area of Toulouse Findings from the study indicate that within the area of interest, heat release from building electricity consumption is the dominant term of the total anthropogenic heat flux density, and that it exhibits weak annual variation with values

in the order of 30 W m−2 in winter and 20 W m−2 in summer It should be highlighted

that the building load profiles data obtained from past QF studies are specific to their

corresponding cities Thus, they cannot be readily applied to other cities or regions

Figure 2.3 (a) Weekly and (b) daily normalised electricity load profiles for the entire city of Toulouse The weekly and daily cycles are normalised by the average total

weekly and daily traffic counts, respectively (Source: Pigeon et al., 2007)

There have been several studies which focused solely on estimating heat release from

buildings (e.g Kikegawa et al., 2003; Kikegawa et al., 2006; Hsieh et al., 2007;

Heiple and Sailor, 2008) The special interest in this topic is perhaps due to the

complicated process of quantifying this particular component of QF Essentially, these studies estimate building-related heat release in a fairly similar manner to that described above However, the main difference is that this group of studies obtains

energy consumption characteristics of the set of buildings that are representative of the building stock through modelling techniques instead of statistical data as in the case of

Inchinose et al (1999) and Pigeon et al (2007) Eventually, these energy consumption

simulations of the prototypical buildings are integrated with a database containing information on building characteristics (e.g building type and size) to determine

energy consumption estimates for each building of interest Hsieh et al (2007)

conducted such a study in Taipei where 10 building prototypes were developed A similar study was also carried out by Heiple and Sailor (2008) in which 22 commercial and 8 residential building prototypes were used to represent the building stock of Houston

Heat release from human metabolismhas been omitted in most anthropogenic heat

flux density studies (e.g Klysik, 1996; Pigeon et al., 2007; Lee et al., 2009) because its effects on QF is typically modest compared to the other two components of QF In spite of this, Grimmond (1992) incorporated this particular component into her study for completeness sake In using the bottom-up modelling approach to estimate heat release from human metabolism, Grimmond (1992) first divided each day into two time periods, ‘active’ (07:00 h – 23:00 h) and ‘sleep’ (23:00 h – 07:00 h) Following this, metabolic rates representative of these two time periods were determined using human metabolic heat production data from Oke (1978) Figure 2.4 illustrates the diurnal variation obtained by Grimmond’s (1992) study, and as expected, it was found

to be the least important component of QF

Figure 2.4 Mean hourly heat release from human metabolism in a Vancouver suburb

(Source: Grimmond, 1992)

Smith et al (2009) also accounted for human metabolism in their study Census

population data was used to estimate the total number of people in the study area while the representative metabolic rates that were used to calculate the metabolic related heat releases are given in Figure 2.5 Based on the population and metabolic

rates data, Smith et al (2009) estimated the maximum daily heat release from human

metabolism in Greater Manchester to be 7.13 W m−2 This value was calculated for a 200-m grid square and is an uncommon extreme When averaged across Greater Manchester, the daily heat release from human metabolism was estimated to be less than 1 W m−2 (Smith et al., 2009) Similar to Grimmond’s (1992) study, Smith et al

(2009) also observed heat release from human metabolism to be the least important contributor to QF in Greater Manchester, representing about 8% of its total anthropogenic heat flux density It should also be highlighted that the census

population data used by Smith et al (2009) accounted for only resident population

numbers Temporary visitors in the study area were ignored and this has probably resulted in an under-estimation of the magnitudes of human metabolic heat release as

certain areas of Greater Manchester, particularly in its city centre and other commercial areas

Figure 2.5 Diurnal metabolic rates representative of Greater Manchester

(Source: Smith et al., 2009)

2.2.3 Inventory-based top-down modelling approach

In contrast to the bottom-up modelling approach, the top-down modelling approach requires data to be obtained at large aggregate scales (e.g yearly) for the purpose of downscaling to smaller scales of interest (e.g hourly) This particular approach was developed by Sailor and Lu (2004) and has been applied in a number of studies, such

required temporal scale It should be highlighted that traffic flow profiles are very useful data resources for estimating heat release from vehicles based on the top-down modelling approach Figure 2.6 shows the weekday hourly traffic flow profile data obtained by Sailor and Lu (2004) for various US cities As for the representative weekend traffic flow profile, they recommend that the dual peaks of the weekday profile be replaced by one that is uniform throughout the day and that the number of vehicles at night makes up for about 1% of the total daily number of vehicles Pigeon

et al (2007) and Lee et al (2009) have also developed mean hourly traffic flow

profiles representative of their study areas located in Toulouse and the Gyeong-In region of South Korea, respectively Their traffic profiles are also illustrated in Figure 2.6 In general, the traffic flow profiles given in Figure 2.6 exhibit similar patterns, with peak traffic volume occurring during the morning and evening rush hours

Figure 2.6 Daily normalised traffic flow profiles for (a) various US cities and states

on weekdays, (b) Toulouse and (c) the Gyeong-In region of South Korea The diurnal profiles are normalised by the average total daily traffic counts (Sources: Sailor and

Lu, 2004; Pigeon et al., 2007; Lee et al, 2009)

Pigeon et al (2007) further developed a weekly traffic flow profile for their study area,

which is provided in Figure 2.7 Based on estimated weekly traffic related data, they suggested that heat release from vehicles in Toulouse on weekends (i.e Saturday and Sunday) is about 25% lower than their weekday values In terms of diurnal variations, the mean daily vehicle related heat release in Toulouse remains relatively constant throughout the year with a magnitude of around 9 W m−2 (Pigeon et al., 2007)

(a) US cities and states

South Korea

Figure 2.7 Weekly traffic flow profile for Toulouse, normalised by the average total

weekly traffic counts (Source: Pigeon et al., 2007)

Sailor and Lu (2004) found heat release from vehicles to be the dominant term of QF

for every US city considered in their study (i.e Atlanta, Chicago, Los Angeles, Salt Lake City, San Francisco and Philadelphia) during the summer months, making up between 47% and 62% of the total anthropogenic heat flux density, which translates to magnitudes of between 14 W m−2 and 37 W m−2 The mean annual magnitude of heat release from vehicles for the South Korean cities of Seoul, Incheon and Gyeonggi were estimated to be 22 W m−2, 14 W m−2 and 10 W m−2, respectively (Lee et al.,

2009) These values represent about 39%, 26% and 35% of the total annual

anthropogenic heat flux density for Seoul, Incheon and Gyeonggi, respectively (Lee et

al., 2009)

A number of studies have used the top-down modelling approach to estimate heat release from buildings For example, Sailor and Lu (2004) obtained the total monthly building electricity consumption values for various US states and disaggregated them

to the city-scale using diurnally varying population densities The city-scale monthly consumption data was then mapped to representative national/regional diurnal

California ISO New England ISO New York ISO PJM ISO

Entergy Northwest Power Planning Commission National average

electricity load profiles for the summer and winter months Figure 2.8 shows these summer and winter electricity load profiles which were developed using data obtained from the online databases of various Independent System Operators (ISOs) in the US Estimates indicate that during the winter months, heat release from buildings constitutes the dominant component of the total anthropogenic heat flux density for Philadelphia (51%), Salt Lake City (52%) and Chicago (57%)

Figure 2.8 Representative (a) summer and (b) winter electricity load profiles for various service regions across the US The profiles are expressed as a fraction of mean total daily electricity consumption The solid line represents the average of

these profiles (Source: Sailor and Lu, 2004)

Heat release from human metabolism contributed to only 2%–3% of the total anthropogenic heat flux density for the US cities considered in Sailor and Lu’s (2004) study Similar to Grimmond (1992), Sailor and Lu (2004) divided each day into two time periods, daytime (07:00 h to 21:00 h) and night-time (23:00 h to 05:00 h) This was the first step towards estimating the magnitudes of metabolic related heat releases for their study areas The missing time periods represented transitional hours Daytime population densities were calculated using working and resident population

data whereas the night-time population densities were calculated using only resident population numbers Based on data from Fanger (1972), Sailor and Lu (2004) assumed a daytime and night-time metabolic rate of 175 W and 75 W, respectively; the daytime value was assumed to change linearly to the night-time value during the transitional hours It has been suggested that for cities with a daytime population density on the order of 10,000 persons km−2, their corresponding metabolic heat flux density should have a magnitude of about 2 W m−2 (Sailor and Lu, 2004)

2.2.4 Energy budget closure approach

The energy budget closure approach has been used to estimate the magnitude of QF in

very few studies The basis of this approach is the UEB, which is mathematically expressed in equation (1.1) Using various micro-meteorological techniques and ignoring heat advection, it is possible to obtain all fluxes of this equation except for

QF Therefore,QF can be estimated as a residual term of the UEB Offerle et al (2005) applied this approach to estimate the magnitude of QF for a downtown region

in Lodz Their findings indicate a summer time value of −3 W m−2 which is physically unrealistic, and a winter time value of 32 W m−2 Given that the residual term of the UEB incorporates not only all errors associated with the eddy-covariance method that is used to measure the sensible and latent heat fluxes, but also errors

owing to the omission of the net heat advection flux, the negative QF value obtained

by Offerle et al (2005) is expected Pigeon et al (2007) also applied the energy

budget closure approach in addition to the inventory-based approach in Toulouse Results from the study suggest that both approaches are in general agreement, with the

winter time QF value reaching as high as 70 W m−2 and the summer time value estimated to be about 15 W m−2

2.3 Discussion of Inventory-Based and Energy Budget Closure Approaches

The inventory-based approach has advantages and disadvantages compared to the energy budget closure approach In addition, the inventory-based bottom-up and top-down modelling approaches each have their own strengths and limitations The following sections provide an evaluation of these approaches in order to determine the

most suitable approach for quantifying QF in the present study

2.3.1 Comparisons between the inventory-based and energy budget closure

approaches

One of main issues associated with using the inventory-based approach to estimate QF

is the problem of obtaining the necessary energy consumption data at the required spatial and temporal scales As mentioned earlier, this data can be extremely difficult

to acquire due to the huge amount of effort and prohibitively high costs involved

Consequently, most of the past QF studies were only able to obtain consumption data

at the annual or monthly scales and at either the city or regional levels In order to

determine finer spatial and temporal scale variations of QF, these studies applied various mechanisms to map their data onto diurnal and/or building-level energy consumption profiles In comparison, the energy budget closure approach is theoretically simpler and more straightforward However, the unrealistic negative QF

value obtained by Offerle et al (2005) for downtown Lodz in summer highlights

several problems associated with this approach First, it has the inherent problem that

all measurement errors of the sensible and latent heat fluxes are accumulated in the QF

term Considering that the storage heat flux density cannot be directly measured and has to be modelled, additional errors resulting from the modelling process are

introduced into the residual QF These errors are further exacerbated by the omission

of the advection heat flux Therefore, in spite of the potential difficulties that might

be faced in the data acquisition process, the inventory-based approach is still favoured

over the energy budget closure approach in the majority of QF studies (e.g Grimmond, 1992; Sailor and Lu, 2004; Lee et al., 2009; Smith et al., 2009)

2.3.2 Comparisons between the bottom-up and top-down modelling approaches

The bottom-up and top-down modelling approaches are both inventory-based, but the former focuses on up-scaling small scale information to larger scales of interests while the latter downscales aggregate data to finer scales A major advantage that the top-down modelling approach has over the bottom-up modelling approach is its widespread applicability to any city in the US However, this advantage is limited to

US cities as the data required to estimate the magnitude QF and its variations from this approach is readily available in US databases only (e.g traffic and electricity load profiles) Thus, if the top-down modelling approach were applied to non-US cities, the amount of effort that has to be spent on acquiring the required data might not be less than that needed for using the bottom-up modelling approach Furthermore, the

estimates of QF obtained from the top-down modelling approach have limited spatial and temporal accuracies owing to several reasons, such as the issue of population

density formulation as pointed out by Sailor et al (2007) Specifically, population

density is a key feature of the top-down modelling approach; however, the inherently large uncertainties related to modelling diurnal and spatial population densities of US

cities as illustrated by Sailor and Lu (2004), implies that the accuracy of QF estimates

obtained from the top-modelling approach will somewhat be compromised It should also be added that the top-down modelling approach developed by Sailor and Lu

(2004) does not provide any means for mapping city-scale QF values to finer spatial resolutions (e.g building-level)