Sydney A Case Study

Power Efficiency Guide

Ultimate Guide to Power Efficiency

Get Instant Access

Cities are almost entirely dependent on the influx of resources from outside,18 and urbanization is proceeding rapidly around the world. Australia is one of the most urbanized countries in the world, and Sydney is its largest city, with roughly 4.5 million residents. In addition to growing personal incomes and expenditures - as described in the previous sections - we can expect the energy needs of Sydney residents to be influenced over time by a host of other societal factors: examples are the tendency for young people to move into single-person households, to marry late, and to have fewer children; more extensive child-care facilities and generous maternity-leave policies allowing women to enter the workforce; longer opening hours and more consumption possibilities; longer hours spent commuting to work and family social events;19 improved health and longevity, and increased leisure activity of the elderly; legislation such as tax rules, vehicle road worthiness criteria and building construction standards; regional population density and climate; and cultural traditions such as holiday cottages, expectations of comfort, etc. (Schipper 1998; Schipper et al. 1989). Unfortunately, data were not available on all of the above factors, but only on a set of 11 socio-economic-demographic variables (Table 4.4).20

The results presented in the following were obtained by (1) applying generalized input/ output analysis (Lenzen 2001) to the most recent Australian Household Expenditure Survey (Australian Bureau of Statistics 2000),21 (2) extracting the set of socio-economic-demographic data from the survey, and using these data as explanatory variables in a multiple-regression

18 This circumstance becomes drastically clear for residents of a city-territory, such as Hong Kong (Newcombe 1975; Part I and Part II).

19 Per unit of time, travel exhibits by far the highest energy intensity of all household activities (Schipper etal. 1989).

20Especially climate has been shown to significantly influence residential energy use in Australia ( Newman 1982), so future studies could add data on this variable. However, for a comparison of Sydney SLAs the omission of climate data is inconsequential, since the mean annual temperature is similar across all SLAs.

21 At the spatial level of Statistical Divisions (SDs) and Statistical Sub-Divisions (SSDs).

Table 4.4. Explanatory variables for energy consumption.

Explanatory variable

Symbol

Comments

Per capita household income

Inc

Weekly

Household size

Size

Number of people in the household

Average age

Age

Calculated from age bracket occupancies; in units of 100 years

Average qualification

Qual

Dummy variable: Postgraduate Degree = 5, Graduate Diploma and Graduate Certificate = 4, Bachelor Degree = 3, Advanced Diploma and Diploma = 2, Certificate = 1

Population density

Dens

Calculated as a weighted sum with population weights; in units of '000 people per km2

Tenure type

Ten

Dummy variable: Owners without a mortgage = 5, Owners with a mortgage = 4, Renters from state or territory housing authority = 3, Other renters = 2, Other = 1

Employment status

Empl

% of employed persons in household aged 18-64

Provenance

Prov

% of householders born overseas

Car ownership

Car

Number of cars per person

Travel to work by car

Trav

% of householders

Dwelling type

Dwel

Dummy variable: Separate house = 4, Semi-detached House = 3, Flat = 2, Other = 1

analysis (Lenzen et al. 2004), and (3) applying the regression formula to the same set of variables, but extracted for smaller spatial units from the Australian National Census (Australian Bureau of Statistics 2007).22 While task 1 is documented elsewhere (Lenzen 1998; Lenzen et al. 2006), tasks 2 and 3 are described in the following two subsections.

4.3.1 Explaining energy requirements

Before regressing energy data against potential explanatory variables, the latter have to be tested for pairwise correlation. Take, for example, the variables 'travel by car', 'dwelling type', and 'population density' which, in Australia, seem to be highly correlated (Table 4.5). This is probably because where population density is low (rural areas and city fringes), the percentage of people driving to work is high (correlation coefficient of —0.71), and the dwelling type index is high (—0.78), indicating a high percentage of separate houses instead of flats. In principle, any of these variables could explain part of the energy requirement, but not all of them simultaneously, since they really measure one and the same thing. Analysts often isolate the original causal variable; in this case it is likely that the availability of space enables the abundance of separate houses, but the sparse occupation of space also brings with it a lack of public transit. For this reason we have omitted the variables 'travel by car' and 'dwelling type' from our regression.23

22At the spatial level of Statistical Local Areas (SLAs).

23All other variables were considered in a step-up-type iterative selection process: first, a regression with one variable - the one showing the highest correlation with energy consumption - was carried out, then the next most-correlated variable added, and so on, until the adjusted R2 value of the regression did not increase anymore.

Table 4.5. Pearson product moment correlation coefficients between explanatory variables, extracted from about 6900 household samples.

Size

Age

Qual

Dens

Ten

Empl

Prov

Car

Trav

Dwel

Inc

-0.16

0.25

0.59

0.36

- 0.05

0.56

0.33

-0.22

- 0.46

- 0.50

Size

-0.58

- 0.09

-0.17

0.18

0.14

0.10

- 0.10

0.10

0.16

Age

0.23

0.23

0.29

-0.11

0.07

0.05

- 0.13

-0.21

Qual

0.54

- 0.12

0.11

0.52

- 0.18

- 0.60

-0.64

Dens

- 0.19

0.08

0.39

- 0.40

-0.71

-0.78

Ten

- 0.02

0.00

0.18

0.36

0.34

Empl

- 0.08

0.01

- 0.09

- 0.07

Prov

- 0.41

- 0.35

- 0.41

Car

0.62

0.51

Trav

0.84

Table 4.6. Results of a multiple regression of the per capita energy requirement against variables in Table 4.4. Rows marked 'Residential' etc. list regression coefficients.24 Rows marked t list results from a Students-t test for significance. Coefficients in bold font are significant at the 99% significance level, regular font - 95%, italic - 90%, underlined - not significant.

Table 4.6. Results of a multiple regression of the per capita energy requirement against variables in Table 4.4. Rows marked 'Residential' etc. list regression coefficients.24 Rows marked t list results from a Students-t test for significance. Coefficients in bold font are significant at the 99% significance level, regular font - 95%, italic - 90%, underlined - not significant.

Inc

Size

Age

Qual

Dens

Ten

Empl

Prov

Car

Residential

0.07

-0.10

1.41

-0.16

0.00

0.02

0.18

-0.29

0.92

t

0.48

2.00

1.98

1.29

0.08

0.17

0.51

0.89

1.87

Transport

0.32

0.02

1.03

-0.07

- 0.07

0.05

-0.01

0.01

0.33

t

3.61

0.54

2.33

0.95

2.54

0.78

0.05

0.06

1.09

Embodied

0.40

- 0.06

0.66

0.20

0.00

-0.01

0.04

-0.10

-0.34

t

6.48

2.84

2.19

3.65

0.03

0.20

0.27

0.76

1.63

Total

0.34

-0.06

0.83

0.12

0.00

0.00

0.07

-0.13

-0.09

t

6.41

2.96

3.15

2.55

0.23

0.02

0.51

1.13

0.47

Other relatively highly correlated pairs are car ownership and travel to work by car (+), employment status and income (+), and qualification and income (+), all for obvious reasons. People with high qualifications tend to live in densely populated areas, and hence in flats rather than separate houses. Interestingly, people born overseas tend to have higher qualifications.

A multiple regression yields that (Table 4.6)

• residential energy is best explained by household size: the more people occupy the household the lower the per capita residential energy consumption;

• increasing age also explains increasing per capita residential energy, independent of higher income or lower household sizes older households may have;

24We chose the functional form ln(E) = a + fi ln(Inc) + ^ fixi, where the x are Size, Age, Qual, etc., and a, fi and yi are regression coefficients (Wier et al. 2001). Using d ln(E)/LE = E_1, it can easily be shown that fi = (dE/E)/(dInc/Inc) is the income elasticity of the energy requirement, which describes the percentage change in the energy requirement as a result of a percentage change in per capita income. Similarly, ySize = (8E/E)/(8Size) describes the percentage change in the energy requirement as a result of the addition of one family member, yAge = (SE / E)/(8Age) the percentage change in the energy requirement as a result of an increase in the average age by one year, and so on.

• car ownership appears to explain residential energy, although the connection is unclear;

• per capita transport energy tends to be high when incomes are high, for households in less densely populated areas, and for older households;25

• per capita embodied energy is clearly and most strongly driven by income, followed by qualification (independent of income);

• per capita embodied energy decreases with increasing household size, which is probably due to people sharing things such as appliances;

• per capita embodied energy increases with age, independent of higher income or lower household sizes older households may have;

• tenure type, employment status and provenance do not appear to influence energy consumption;

• the variability of the total energy requirement is dominated by embodied energy.

We measure the income elasticity of the total energy requirement at 0.34 (Table 4.6), which means that for a 10% increase in expenditure, the energy requirement increases only by 3.4%. This is due to the fact that when income increases, the proportion of services compared to goods in the consumer basket increases, so that each additionally purchased commodity tends to be less energy intensive.26 The size and age elasticities of the energy requirement are -6% per additional household member, and +0.8% per year, respectively.

Applying the regression coefficients in Table 4.6 to Australian average socio-economic-demographic variables yields a per capita energy requirement of 180 GJ, which is in excellent agreement with the energy requirement of 176 GJ/cap calculated directly from household expenditure data (Australian Bureau of Statistics 2000).

4.3.2 A spatial view of direct and embodied energy consumption

There are a number of studies that attribute energy use or greenhouse gas emissions to residents of a city, and depict the result as shaded maps (Newcombe 1976' Kalma and Newcombe 1976' VandeWeghe and Kennedy 2007). This has even been done for Sydney, our own case study (Kalma et al. 1972). However, to our knowledge, these studies only deal with residential and not with embodied energy. The first publication to present embodied energy in a spatial context is Lenzen et al.'s analysis of Sydney (2004). The work presented in the following case study replicates this previous study's methodology, but updates the results, and significantly increases the spatial resolution by applying the regression coefficients in Table 4.6 to information on lifestyles contained in the Australian Census, which holds data on all of the variables in Table 4.6 at a detailed spatial level.

25 Interestingly, Schipper et al. (1989) finds that in the 1980s US, elderly people spent more time at home and drove less, so that their residential and transport energy requirements were above and below average, respectively. Already in 1989, Schipper conjectures that 'tomorrow' s energetic retirees . . . could carry with them their mobility patterns of younger years, while they continue to live in homes originally built to house families with two or three children'. In our regression we see exactly this phenomenon: Both residential and transport energy use are accelerated by average household age.

26The income elasticity of 0.34 reported here is lower than the expenditure elasticities in Fig. 4.5, which range between 0.7 and 1.0. This is because (a) a multivariate regression as in Table 4.6 involves explanatory variables in addition to income, and as such the income elasticity of 0.34 is lower than would have resulted from a univariate regression with only income as a variable; and (b) Fig. 4.5 depicts energy versus expenditure, while Table 4.6 analyses energy versus income. Expenditure elasticities are always higher than income elasticities because expenditure is a better proxy for the energy requirement than income (Newcombe 1979; Wier et al. 2001).

Labeled Map Asia Colouring

Fig. 4.6. Regions of Greater Sydney.

The maps shown in the following sections contain the city of Greater Sydney, which includes the Blue Mountains and the Lower Central Coast. Bordering are the Illawarra and Southern Highland regions, the Hunter Valley, and the Pacific Ocean (Fig. 4.6). These maps may suggest viewing the embodied energy issue as an issue of geographic location; however, we ask the reader to interpret spatial units as entities of certain socio-economic-demographic characteristics. Where the maps show 'hot spots' of high embodied energy, this is where the drivers of embodied energy - for example, affluence - are strong. Hence, we will investigate why certain areas feature certain characteristics, which then lead to consumption habits, and in turn to the embodied energy pattern.27

4.3.2.1 Residential energy

The majority of residential energy in Australia is consumed as electricity, natural gas, and firewood (Table 4.7). In our analysis of Sydney, firewood is likely to play a minor role compared to the whole nation, so that our multiple regression is most sensitive to households' electricity and gas bills.

Residential energy appears to be high in the SLAs around Sydney Harbour, and in rural areas to the north-west of the city (Fig. 4.7 [Plate 6]). Our regression (Table 4.6) shows that per capita residential energy is high for small and old households, so one would expect to see a spatial correlation of residential energy with these factors.

Household size exerts the strongest influence (Table 4.6( and is therefore shown for comparison in Fig. 4.8 [Plate 7]. Indeed, for the city itself, we find that residential energy is low where many people share a household, and vice versa. This could simply be because

27 Maps are available also for greenhouse gas emissions, water use and the Ecological Footprint for SLAs in all Australian States and Territories (http://www.acfonline.org.au/consumptionatlas).

Fig. 4.7. Per capita residential energy in Greater Sydney SLAs. [Plate 6] k/ ^

Residential energy GJ/cap

Fig. 4.7. Per capita residential energy in Greater Sydney SLAs. [Plate 6] k/ ^

Household size People/household

Table 4.7. Breakdown of Australian residential energy into fuel types (Australian Bureau of Agricultural and Resource Economics 2006).

Energy carrier

% of total

Electricity

44.1

Natural gas

30.1

Firewood

21.1

LPG, LNG

2.6

Solar energy

1.0

Heating oil

0.6

Diesel oil

0.4

Kerosene

0.1

Black and brown coal

0.04

these people share energy services such as light and heat that are largely independent of how many people live in a flat or house.

Sydney' s age structure (older households in the seaside suburbs, younger households towards the south-west) reinforces the spatial distribution in Fig. 4.7 [Plate 6]. The two outer high energy SLAs (Lithgow and Cessnock/Hunter Valley) are caused by the two remaining explanatory factors (high car ownership and low percentage of people born overseas), and are perhaps an artefact of the regression analysis.

4.3.2.2 Transport energy

Energy used for private passenger transport (almost all petrol) is clearly related to geographical location, with energy use increasing with increasing distance from the city centre (Fig. 4.9 [Plate 8]). People living in the centre and along major bus and train arteries have at their disposal various transport options, leading to below-average petrol consumption.

Population density is the strongest explanatory variable for transport energy (Table 4.6), and this relationship is well depicted by contrasting the maps for both factors (Figs 4.9 [Plate 8] and 4.10 [Plate 9]). Major train lines only reach west into the Blue Mountains, north toward the Central NSW Coast, and south to the Illawarra and Southern Highlands regions. Compared to urban rail transport in other megacities, these services are infrequent and slow. Frequent buses run mostly in SLAs denser than 500km-2' so that most people living well outside of the city proper rely entirely on the private car.

4.3.2.3 Embodied energy

By far the most significant relationship in our entire regression ties embodied energy to income (Table 4.6). This is why perhaps the most striking resemblance can be found between the maps depicting these two variables (Figs 4.11 [Plate 10] and 4.12 [Plate 11]).

First, note that the magnitude of the scale in Fig. 4.11 [Plate 10] is five to ten times higher than those in Figs 4.6 and 4.8 [Plate 7], once more emphasizing the dominance of embodied energy. Any visitor to Sydney can easily tell from the size and style of houses, and the water views, that the most affluent people congregate around Sydney Harbour, along the

Transport energy GJ/cap

19.5-20.0 ] 20.0-20.5 20.5-21.0 21.0-21.5 21.5-22.0

19.5-20.0 ] 20.0-20.5 20.5-21.0 21.0-21.5 21.5-22.0

Fig. 4.9. Per capita transport energy in Greater Sydney SLAs. [Plate 8]

Fig. 4.9. Per capita transport energy in Greater Sydney SLAs. [Plate 8]

Population density People/square km

0-86 87-276 277-505 ] 506-1185 1186-2369 2370-3338 3339-6595 6596-10 352

Embodied energy GJ/cap

106-114 115-120 121-128 129-138 139-152 153-170 171-188 189-237

Fig. 4.11. Per capita embodied energy in Greater Sydney SLAs. [Plate 10]

Fig. 4.11. Per capita embodied energy in Greater Sydney SLAs. [Plate 10]

<14000

Income $/cap/year

<14000

14 000-16 000 16 000-18 000 18 000-20 000 20 000-22 000 22 000-26 000 26 000-32 000 32 000-40 000

Eastern Suburbs beaches, and on the North Shore. This is where some of the nation's most wealthy and vigorous consumers reside, and where embodied energy accumulates.

In contrast, the south-west of Sydney and the rural areas to the west (except the Blue Mountains) are characterized by young, larger families with mostly low incomes and, as a consequence, low embodied energy budgets.

Sydney adheres strongly to the conventional measures of success implied by the last three centuries of industrialization and globalization. Its metabolism is based on the manner in which it attracts global citizens seeking modern lifestyles with the best-quality goods and services, and is facilitated by a well-connected port and airport. The city ' believes' in economic growth, affluence, population growth and a continuing real-estate boom. Occupancy rates for domestic and commercial building stocks remain high and reap good financial returns for their owners.

In terms of direct energy, Sydney faces many structural impediments caused by its settlement geography and history, which surface as a range of engineering difficulties, particularly for better urban transit. In energy terms it is mostly dependent on black-coal electricity generated in the Hunter Valley to the north of the city, thus locking in a high carbon content for its electricity. The strategic encouragement of car transport by continual freeway construction along with a massive underinvestment in efficient urban transit further imposes a higher carbon content of mobility than technology suggests is available. The relative age (in Australian terms) of the inner city means that retrofitting of building stock can be physically difficult and expensive. Finally, the marginalized and fragmented nature of local, state and federal representation means that energy politics has never mustered sufficient medium-term support and financing to achieve a ' great leap forward' in the city's energy metabolism. There are few politicians prepared to put their votes at risk for difficult transitions that might take two decades to return real energy savings, particularly given the dominant conventional view that equates embodied energy consumption with success.

In embodied energy terms, the residents of Sydney are generally aspirational of a lifestyle that implies style, fashion and a diversity of possessions. They have higher skills on average than the rest of Australia, get higher pay and thus consume more in embodied energy terms. In the wealth and consumption stakes Sydney is Australia's showcase city in world and national terms. Within its city boundaries it displays a strong gradient in per capita embodied energy requirements from the longer settled and more affluent areas around its harbour and coastlines, to the less affluent suburbs at the western and southern extremities. These embodied energy gradients reflect social equity barriers that have possibly locked in less affluent residents to energy dependence as a way of life. Since social disadvantage usually means poorer health and educational outcomes, any hypothetical energy transition will be doubly difficult because of lower levels of human capital and understanding of why these transitions should take place.

Given this hierarchy of global and functional influences, it seems unlikely that our city of interest, Sydney, or for that matter any first-world city, will be able to effect an energy transition with sufficient leverage to reformat its metabolism and thus meet its global environmental goals. Notwithstanding this dour conclusion, there are five policy strategies which Sydney's managers must work towards as follows. The first is to create new attributes of wealth, human development and status where social richness and community vitality replace volumetric consumption as the key driving forces and measures of success of advanced societies. The second is to cap absolute population numbers over the next two to three human generations, while achieving a population structure that is reasonably balanced by age (no booms and busts) and age by spatial location (no very old and very young suburbs). The third is to introduce carbon rationing and trading for consumers on a full production chain basis so that the full embodied greenhouse content is accounted for, and priced into consumer transactions at the point of sale. The fourth is to form a multimodal city well linked by low carbon transport options and to constrain single-person vehicle transport. The fifth is to mandate leading-edge energy regulations for all new buildings, and, in parallel, begin retrofitting suburb by suburb the existing built infrastructure to constrain energy use, while meeting reasonable economic and social requirements.

4.3.3 Forecasting energy requirements

Just before the manuscript for this book went to print, the 2006 Australian Census was released by the Australian Bureau of Statistics, prompting the main Sydney daily newspaper to describe Australians as 'richer, older, and lonelier than before'(Wade et al. 2007). These comments refer to increasing real incomes, an ageing society, and the trend towards single or couple households. Such developments have become daily topics of debate among politicians and the general public alike, and therefore it is intriguing to examine the energy implications of a few future scenarios.

The business-as-usual (BAU) scenario assumes that the historical 2% real economic growth (World Resources Institute 2006) continues until 2050. By that time the average Australian is assumed to live in suburbs that are as dense as city fringes are today (800 km—2(. Immigration and procreation are assumed to balance the age pyramid, and household sizes will stay the same (two generous assumptions). According to our lifestyle regression, under this scenario, even though affluence rises, the per capita energy requirement will stay about constant at 180 GJ/cap, thanks to (a) energy efficiency improvements of 0.8%/year or an overall 40%, (b) higher population densities that make better use of transport infrastructure, and (c) the ' consumer basket effect' described at the end of the previous subsection. Overall (national) energy consumption will be 70% higher than in 1999, because Australia's population will have grown by 70% from 20 million to just above 30 million.

The 'Fast' scenario assumes an accelerated real growth of 3%/year, which among other influences will drive the trend towards smaller households. An aged Australian population is assumed to live in families typical for today's North Shore (see Figs 4.8 [Plate 7 and 4.12

Table 4.8. Results for three simple future scenarios for 2050. All scenarios assume the continuation of historical trends for the overall energy intensity (—0.8%/year, technology driven) and population (+1%/year, immigration plus procreation) (Foran and Poldy 2002).

Scenario

Australia, average

BAU

Fast

Slow

Income growth

2%

3%

1%

Income

$337

$924

$1520

$559

Household size

2.8

2.8

2.2

3.6

Age

35

35

38

33

Population density

554

800

554

1000

Per capita energy requirement

180

180

225

141

Total energy requirement (1999 = 1)

1

1.7

2.1

1.3

Plate 11]; 2.2 people/household, 38 years), and new settlements are assumed to sprawl by copying the existing spatial structure (554 km-2). Under this scenario, socio-economic-demographic trends outrun efficiency gains, and per capita energy requirements will exceed 220 GJ. Australia's total energy use will more than double until 2050.

The ( Slow ( scenario features only 1%/year real growth, and consolidated urban and family structures. A younger Australian population is assumed to live in households typical for today's Fairfield-Liverpool (a low income Sydney suburb, 3.6 people/household, 33 years), but at densities of 1000 km-2.28 Under this scenario, per capita energy requirements will have significantly decreased to about 140 GJ. Australia (s total energy use will still increase by 30%.

Under every scenario, the growth of Australia's energy metabolism is underpinned by more people wanting a more comfortable and convenient life, travelling more often and further, and enjoying more material wealth. This shows once again, and drastically, that an urban energy transition cannot be viewed without considering embodied energy.

Was this article helpful?

0 0

Post a comment