reliability of building embodied energy modelling: an analysis of 30 melbourne case studies

This article was downloaded by: [Radboud Universiteit Nijmegen]On: 03 November 2014, At: 05:27Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Construction Management and EconomicsPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/rcme20

Reliability of building embodied energy modelling: ananalysis of 30 Melbourne case studiesYu Lay Langston a & Craig Ashley Langston aa Faculty of Science and Technology , Deakin University , School of Architecture andBuilding , 1 Gheringhap Street, Geelong, 3217 AustraliaPublished online: 26 Feb 2008.

To cite this article: Yu Lay Langston & Craig Ashley Langston (2008) Reliability of building embodied energymodelling: an analysis of 30 Melbourne case studies, Construction Management and Economics, 26:2, 147-160, DOI:10.1080/01446190701716564

To link to this article: http://dx.doi.org/10.1080/01446190701716564

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/rcme20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/01446190701716564

http://dx.doi.org/10.1080/01446190701716564

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

Reliability of building embodied energy modelling: ananalysis of 30 Melbourne case studies

YU LAY LANGSTON and CRAIG ASHLEY LANGSTON*

Faculty of Science and Technology, Deakin University, School of Architecture and Building, 1 Gheringhap Street,

Geelong, 3217 Australia

Received 11 May 2007; accepted 1 October 2007

Building design decisions are commonly based on issues pertaining to construction cost, and consideration of

energy performance is made only within the context of the initial project budget. Even where energy is elevated to

more importance, operating energy is seen as the focus and embodied energy is nearly always ignored. For the first

time, a large sample of buildings has been assembled and analysed in a single study to improve the understanding of

the relationship between energy and cost performance over their full life cycle. Thirty recently completed buildings

in Melbourne, Australia have been studied to explore the accuracy of initial embodied energy prediction based on

capital cost at various levels of model detail. The embodied energy of projects, elemental groups, elements and

selected items of work are correlated against capital cost and the strength of the relationship is computed. The

relationship between initial embodied energy and capital cost generally declines as the predictive model assumes

more detail, although elemental modelling may provide the best solution on balance.

Keywords: Accuracy, capital cost, embodied energy, capital cost, modelling.

Introduction

Energy has become a significant issue worldwide.

Greenhouse gas emissions (GGE) and the perceived

threat of climate change (caused by phenomena such

as global warming and ozone depletion) are identified

by Beggs (2002, p. 10) as driving ‘more than any

other issue’ change in energy consumption attitudes.

Since the energy crisis of the mid-1970s attention

has been directed towards strategies that lower operat-

ing energy demand (Robertson, 1991), yet it has

been only recently that the impact of energy embodied

in building materials themselves has come under

scrutiny.

Australia has the highest per capita GGE in the world

(NAEEEC, 1999; Department of Natural Resources

and Environment, 2000; ASEC, 2001). Without

targeted and effective action, these emissions are

projected to grow by 28% from 1990 to 2010

(NAEEEC, 1999). Bell and Fawcett (2000) indicate

that GGE from the Australian construction industry are

substantial and rapidly rising, particularly in the

commercial sector. Buildings consume 40–50% of the

energy and 16% of the water used annually worldwide

(Lippiatt, 1999; Hoglund, 1992; Lam et al., 1992).

Buildings comprise a combination of embodied energy

and operating energy. According to Boustead and

Hancock (1979), embodied energy is defined as the

energy demanded by construction plus all the necessary

upstream processes for materials such as mining,

refining, manufacturing, transportation, erection and

the like, while operating energy is related to issues of

recurrent occupancy and building use. It is now

realized that a focus on operating energy is insufficient

to address either national or international GGE

concerns (Fossdal, 1995). Embodied energy is steadily

increasing (AGO, 1999) owing to a greater use of

energy-intensive materials (such as aluminium, stain-

less steel, coated glass and high strength concrete),

larger buildings, more frequent refurbishment cycles,

more machine utilization in construction processes,

higher transportation energy and the introduction of

new technologies such as photovoltaic cells and

building management systems. Despite this trend, the

routine analysis of embodied energy remains absent

(Treloar et al., 2002).*Author for correspondence. E-mail: [email protected]

Construction Management and Economics (February 2008) 26, 147–160

Construction Management and EconomicsISSN 0144-6193 print/ISSN 1466-433X online # 2008 Taylor & Francis

http://www.tandf.co.uk/journalsDOI: 10.1080/01446190701716564

Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

Proper energy analysis during the design process can

no longer be simply overlooked. ASEC (2001, p. 100)

indicate that ‘total energy use has doubled [in

Australia] over the last 25 years […] at a faster rate

than GDP’. The rationale behind this paper firmly lies

with the perceived lack of integration of energy analysis

into current practice. Capital cost still remains the

primary criterion for building procurement decisions

(Brown and Yanuck, 1985; Langston, 1991; Bull,

1992), while other criteria are given less significance

either because of a narrow myopic focus (Ashworth,

1988) or because a suitable multi-criteria technique has

not been satisfactorily identified (van Pelt et al., 1990).

Energy analysis is costly, time consuming and, when

undertaken during the design phase, usually based on a

large number of assumptions (Verbeek and Wibberley,

1996). Even so, it is likely to produce advice that conflicts

with that generated from capital cost estimates (Arnold,

1993). This occurs because energy analysis takes a long-

term view, one that introduces multiple stakeholders and

wider social concerns, rather than merely reflecting

immediacy and profit-centred objectives. It has been

argued over many years (e.g. Stone, 1960; Kirk and

Dell’Isola, 1995; Flanagan and Norman, 1983; Langston

and Lauge-Kristensen, 2002) that costs should also be

accounted over a longer time span. Known as life cycle

costs (LCCs), these comprise both initial (capital) and

recurrent (operating) components that can be aggregated

to give a more realistic picture of the total expenditure

commitment (Fuller, 1982).

The problem essentially is how two criteria, one

measured in financial terms and the other in pure

energy terms, can ever be reconciled to provide clear

building design guidance. It is not commonly under-

stood that the lowest LCC solution will automatically

be the lowest energy solution. In fact, any comparison

of particular material choices will usually indicate that

cost and energy ratios vary widely (Irurah and Holm,

1999). Yet at the level of an entire building this

differential is expected to be less—a view that is

supported to some extent by the manner in which

embodied energy intensities are often determined (i.e.

from national input–output financial tables) and

operating energy interpreted (i.e. incurred cost).

If there is an inherent relationship between energy

and cost that can be exploited to enable better design

solutions to be identified, then it should be possible to

quantify energy directly from an LCC investigation.

The outcomes will naturally be dependent on the

chosen time horizon for the study but will simplify the

process of embodied energy calculation (in particular)

that to date has proved elusive to common practice.

The purpose of this paper is to explore the predictive

accuracy of initial embodied energy modelling based on

capital cost, using data derived from a range of building

types in Melbourne, Australia. From this information, a

better understanding of facilities performance can be

obtained, leading to further insight into the relationship

between energy and cost. The structure of this paper is to

review previous research findings, to outline the method

adopted in this research, to analyse the results, and to

make observations and draw conclusions for practice.

Background

Energy is a vital consideration for the effective design

and operation of the built environment, not only in

Australia, but globally. There is an increasing need for

society in general and the construction industry in

particular to calculate the expected energy performance

of buildings, in terms of both capital (embodied)

energy and operating (recurrent) energy. While operat-

ing energy is fairly simple to quantify, embodied energy

is time consuming, complex and highly dependent on

the chosen measurement algorithm. This problem has

hindered the uptake of energy considerations in

practice that in turn has made it difficult to make

judgments about the appropriate and sustainable

deployment of resources.

Thus if it were possible to routinely produce accurate

estimates of total embodied energy for individual

buildings, then a major impediment to the widespread

use of energy analysis would be overcome. Such a step

forward would lead to significant improvements in the

performance of the Australian construction sector and

in turn the economic and environmental performance

of the nation through the design, construction and

management of facilities that are energy optimized.

There are four types of embodied energy analysis

described in the literature. These comprise process

analysis, statistical analysis, input–output analysis and

hybrid analysis. No method is perfect. Incompleteness

in typical embodied energy analysis is estimated at

around 20% (Treloar, 1997).

The input–output-based hybrid method, as used in

this paper, is described by Crawford (2004; p. 130) as

‘the most sophisticated complete life cycle inventory

assessment method currently available for assessing

environmental impacts associated with building and

building-related products’. Energy analysis using this

method can be defined using a series of equations as

discussed below.

Total life cycle energy LEGJ~IEEGJzREEGJ

zOEGJ

ð1Þ

where:

IEEGJ 5 initial embodied energy;

148 Langston and Langston

Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

REEGJ 5 recurrent embodied energy over the

chosen time horizon (t);

OEGJ 5 operating energy over the chosen time

horizon (t).

Total life cycle energy excludes other energy involved

in reusing, recycling and final disposal. Fossil fuels such

as coal, oil and natural gas are responsible for over 90%

of global purchased energy (Chukwuma, 1996) and most

of Australian purchased energy (Lawson, 1995). Primary

energy is reduced by conversion and transmission losses,

resulting in delivered energy to the end user (Crawford

et al., 2002). Total life cycle energy is measured in

primary energy terms so that the total amount of fossil

fuels used directly and indirectly in the manufacture of

the delivered energy provides a complete assessment of

resource depletion and environmental impact (Treloar,

1996; Shen, 1981; Peet and Baines, 1986). The

estimated primary energy factor (primary energy needed

to create one unit of delivered energy) in Victoria is 3.4

for electricity and 1.4 for gas (Treloar, 1997; Treloar

et al., 2001).

Initial embodied energy IEEGJ~DEGJzIEGJ

zOIEGJ

ð2Þ

where:

DEGJ 5 direct energy in construction;

IEGJ 5 indirect energy (modified pathways);

OIEGJ 5 other indirect energy (remaining path-

ways).

Treloar et al. (2001) describe direct energy as a single

pathway related to the main process of construction while

indirect energy comprises upstream multiple energy

pathways. Those multiple pathways that can be quanti-

fied are called modified pathways. Other indirect energy

represents the embodied energy of the remaining path-

ways (sometimes referred to as ‘incompleteness’).

Direct energy DEGJ~DEIGJ=$1000|PP$ ð3Þ

where:

DEIGJ/$1000 5 direct energy intensity of I-O sector

(n);

PP$ 5 project price.

Direct energy is calculated using input–output data

where process analysis data are unavailable (Crawford,

2004). The total energy intensities and direct energy

intensities of the 106 sectors of the Australian economy

are compiled by Lenzen and Lundie (2002) based on

Australian Bureau of Statistics input–output data for

the financial year 1996–97.

Indirect energy IEGJ~QTYunit|EIGJ=unit ð4Þ

where:

QTYunit 5 delivered material quantity;

EIGJ/unit 5 hybrid material energy intensity per unit.

Indirect energy is a factor of delivered quantities of

materials and their energy intensities. Material quan-

tities are converted into delivered quantities using

wastage multipliers derived from Wainwright and

Wood (1981) and Fay (1999). The database of hybrid

material energy intensities compiled by Dr Graham

Treloar is in primary energy terms and comprises

various process data and input–output data also

obtained from the Australian Bureau of Statistics for

the financial year 1996–97.

Other indirect energy OIEGJ~REIGJ=$1000|PP$ ð5Þ

where:

REIGJ 5 energy intensity of remaining paths;

PP$ 5 project price.

Other indirect energy is the energy required to fill the

gap between total multiple pathways and modified

pathways (Crawford, 2005). Treloar (1998, p. 173)

emphasizes that the most crucial task in input–output-

based hybrid analysis is ‘not the modification of direct

energy paths with process analysis data but rather the

inclusion of the remainder of unmodified direct energy

paths to make the framework more comprehensive’. A

sector’s overall total energy intensity in the input–

output model consists of all the pathways’ energy

intensities. The energy intensity of the remaining

pathways is the gap between total energy intensity

and the sum of the modified pathways’ total energy

intensity and the direct energy intensity.

Recurrent embodied energy REEGJ~REGJ

zMEGJzOIEGJ

ð6Þ

where:

REGJ 5 replacement energy over the chosen time

horizon (t);

MEGJ 5 maintenance energy over the chosen time

horizon (t);

OIEGJ 5 other indirect energy over the chosen time

horizon (t).

Recurrent embodied energy consists of the embodied

energy of building materials and systems based on their

life expectancy and maintenance requirements and can

be estimated by sources such as Langston (1994),

Dell’Isola and Kirk (1995), AIQS (2002) and Langston

(2005).

Reliability of building embodied energy modelling 149

Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

Method

The selected research method is sampling via case

studies. Case study is an ideal methodology when a

holistic in-depth investigation is needed (Feagin et al.,

1991). It has been used in varied investigations,

particularly in sociological studies but increasingly in

construction. The procedures are robust, and when

followed the approach is as well developed and tested

as any in the scientific field. Whether the study is

experimental or quasi-experimental, the data collection

and analysis methods are known to hide some details.

Case studies, on the other hand, are designed to bring

out the details from the viewpoint of the participants by

using multiple sources of data (Tellis, 1997).

The data for the research are drawn from 30 actual

case studies obtained courtesy of the Melbourne office

of Davis Langdon Australia, a large national quantity

surveying practice. Case studies are located across

Greater Melbourne and are specifically intended to

reflect a broad range of functional purpose. They

represent a ‘random’ mix in that no selection decisions

were applied, including provision of office workspace,

health facilities, residential accommodation, teaching

and laboratory space, retail, hotel accommodation and

a number of specialist uses. Projects comprise both new

construction (73.3%) and redevelopment (26.7%). So-

called residential projects, comprising apartment build-

ings and aged care facilities, account for 23.3% of the

case studies, and the remainder are constructed for

various other commercial uses. One-third of the case

studies are hospitals or healthcare facilities. It should be

noted that no claims are being made that the case

studies are representative of the mix of building stock in

Melbourne (or Australia), yet the variety in the mix

adds confidence that the correlation between energy

and cost is robust and reliable.

Capital cost data and floor areas are obtained direct

from the elemental cost plans prepared by Davis

Langdon Australia. Embodied energy intensities are

estimated from the composite items of work listed in

these documents using the input–output-based hybrid

method (Treloar, 1998; Crawford, 2004). All costs are

adjusted and expressed in fourth quarter 2006 dollars

using published building price indices (BPI) also

supplied by Davis Langdon Australia.

Projects range from 1997 to 2004, and vary in floor

area from 249m2 to 18 821m2 gross floor area (GFA).

The mean floor area is 3749m2 (coefficient of variation

of 110.75%). They comprise a wide range of materials

and standards, some are air-conditioned and some not,

some have fire sprinkler systems, some have loose

furniture and special equipment, and some have

substantial external works.

This mix decreases the likelihood that projects

exhibit similarities in energy and cost performance.

Economies of scale also play a part in larger projects,

which tend to have lower unit costs than identical

designs of smaller size. The mix is therefore effectively

random, enabling a range of statistical techniques to be

applied to the sample.

Table 1 lists the case studies used in this research by

building type. Case studies are identified by a

numerical code, as the name and location of projects

needs to be kept confidential (this is a non-negotiable

agreement made between the researchers and Davis

Langdon Australia).

Data supplied by Davis Langdon Australia comprise

GFA and elemental capital costs based on abbreviated

measured quantities extracted from design cost plans.

The full project cost is presented, including prelimin-

aries, site works and external services, and special

provisions (such as allowances for loose furniture and

equipment), but excluding contingencies, professional

fees, land acquisition costs and goods and services tax.

Capital costs are converted to fourth quarter 2006

prices using a BPI provided by Davis Langdon

Australia. Otherwise no adjustment to capital costs is

undertaken and all unit rates are taken as correct and

reflective of the project given applicable market

conditions at the time. The BPI for fourth quarter

2006 is 175.0 (later indices were not used as they were

still forecasts at the time of analysis).

Embodied energy is determined using a sophisticated

spreadsheet model using an input–output-based hybrid

method that embraces both process analysis data

(where available) supplemented with the input–output

data from published government statistics (1996–97

financial year), extracted and compiled at Deakin

University by Dr Graham Treloar and Dr Robert

Crawford. The basis of this method is described earlier

and reflected in the equations provided. Examples of

the detailed calculations for embodied energy in this

study can be found in Langston (2006).

No allowance has been made for the embodied

energy of labour. This is common practice for

embodied energy analysis despite the fact that labour-

intensive activities have ‘upstream’ consequences.

Furthermore, the energy involved in reusing, recycling

and final disposal of materials is excluded from this

study. The introduction of labour potentially breaks the

close relationship between energy and GGE. This

situation can be traced to the assumption that the

traditional primary energy factors of economic produc-

tion—land, labour and capital—are independent

(Costanza, 1980). Yet a strong case can be made that

these input factors are not independent and that energy

is required for their production (Odum, 1971; Baird

and Chan, 1983). Hill (1978) concludes that human


Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

energy is ignored because the methods for assessing it

are not satisfactorily resolved. The treatment of labour

is therefore controversial.

Results

It is hypothesized that initial embodied energy is

reliably modelled by capital cost, but this reliability

decreases proportionally with the level of detail at

which the model is expected to operate. Linear

regression has been used in this study for two reasons:

to keep the model relatively simple so it can be

understood and applied by practitioners, and to adopt

a consistent approach at all levels of the investigation.

Adjusted r2 takes into account the number of data

points on the regression line. While 30 data points is a

reliable indicator of correlation (hence the difference

between r2 and adjusted r2 is small), fewer data points

see a bigger gap between the two variables. P-values are

calculated to confirm the existence of a relationship

(based on 95% confidence levels), so where p-value is

more than 0.05 no relationship exists. Similarly, where

the value of t-critical for one-tailed relationships is

significantly greater than zero, increased estimating

confidence in the slope coefficient or gradient (m) of

the regression line is provided.

The correlation between initial embodied energy and

capital cost is very strong (r250.9542) and the

discovered regression line is y50.0071x. The result is

presented in Figure 1. Furthermore, statistical analysis

shows a very robust relationship: adjusted r250.9179,

p-value51.12224 and t-statistic535.05.

Provided costs are expressed in fourth quarter 2006

terms, the embodied energy can be derived in minimal

time using the constant (gradient) of 0.0071. This

relationship can be used to help predict embodied

energy based on initial construction cost estimates. The

reliability of this constant is now investigated further.

Table 2 summarizes key correlation outcomes for

each elemental group. It should be noted that the

reliability of alterations and renovations (AR), external

services (XK-XS) and special provisions (YY) is poor

(in fact, coefficients cannot be calculated for YY), and

Table 1 Case study base information (Davis Langdon Australia, Melbourne Office)

Building

IDYear Building type Gross floor area (m2)

1 2003 Residence (new) 1409



4 2000 Office (new) 2543

5 2003 Health Centre (redevelopment) 528

6 2003 Hospital (new) 6761


8 2003 Information Centre (new) 1223

9 2004 Hospital (redevelopment) 3278


11 2004 Library (new) 249

12 2004 Civic Hall (new) 625

13 2004 Primary School (new) 2696




17 1999 Hotel (redevelopment) 652

18 2000 Car parking station (new) 5412


20 1998 Health Centre (new) 787


22 1999 Hotel (new) 12 930

23 1999 Residence (redevelopment) 18 821

24 1999 University building (new) 10 565

25 1999 Office (new) 4704



28 1998 University building (new) 2502




Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

external alterations and renovations (XX) has no

relationship (p-value50.220). Refer to AIQS (2002)

for full definition of elemental classifications used in

this research.

Table 3 increases detail further to the element level,

again confined to initial embodied energy and capital

cost. More variation in the strength of the energy–cost

relationship is evident. Sanitary plumbing (PD), water

supply (WS), ventilation (VE), special services (SS)

and all site-related elements except roads, footpaths

and paved areas (XR), outbuildings and covered ways

(XB) and external electric light and power (XE) are

unreliable, and external special services (XS) has no

relationship. It is argued that the more detailed

approach, based on elemental groupings or based on

elements, is less likely to reflect actual initial embodied

energy than a single line calculation performed at the

project level.

Values for m in the regression equation (y5mx) can

be used to predict initial embodied energy based on

construction cost (expressed in fourth quarter 2004

dollars) for all elements and groups except external

alterations and renovations (XX) at varying levels of

reliability. Nevertheless, caution should be exercised

when using these values to predict site-related ele-

ments, since the basis of the relationship is linked to the

extent of work involved compared to the GFA of the

building. This relationship can vary quite significantly,

and projects that possess more extreme ratios of work

extent to GFA will lead to over- or underestimation of

Figure 1 Embodied energy vs capital cost correlation


Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

initial embodied energy. But the values for m presented

do provide some indication of the significance of site-

related elements compared to the more usual building

elements.

The figures presented in Table 3 are important. They

underline the falling reliability of element correlations

compared to elemental group correlations, yet provide

support for the obvious link between energy and cost at

most levels of analysis. While the relationship is

undeniable, the strength of the relationship is variable.

Two elements (4.67%) show a very strong value for

adjusted r2 (.0.9), 20 elements (47.62%) show a

strong value (.0.7, but ,0.9), and five show a

moderate value (.0.5, but ,0.7). The remaining 15

elements (11.91%) show a weak value of adjusted r2

(,0.5). Therefore about half of the elements have a

strong energy–cost relationship or better. This is

compared to the 11 elemental groupings (five of these

are also elements included in the previous discussion).

Five groups (45.46%) show a very strong value of r2

(.0.9), two groups (18.18%) show a strong value

(.0.7, but ,0.9), and the remaining four groups

(36.36%) show weak values (,0.5). Bear in mind a

very strong correlation is produced at the overall

project level.

To further illustrate the declining energy–cost

relationship, elements in Building 24 are used as the

data points for regression. For example, one data point

is preliminaries, another is substructure, another is

columns, etc. It should follow that a relationship still

exists, but as can be seen from Figure 2, this relation-

ship is not strong, with an r2 of just 0.5217. Much more

‘scatter’ around the regression line is apparent.

This trend is repeated when the relationship between

energy and cost is tested for a range of common

materials used in construction (extracted from the

embodied energy model developed for this paper).

Table 4 lists these materials, their units of measure-

ment, energy/unit, cost/unit (based on the material

supply price including delivery but excluding installa-

tion) and ratio of cost to energy. There are 33 items in

the list, which provides a sufficient dataset for the

analysis. A Y-intercept of zero is again assumed.

The discovered relationship is weaker. Figure 3

shows the correlation between energy/unit and cost/

unit and determines that the derived r2 is quite low at

just 0.3560. While it is of course possible to collect

different materials and hence calculate different values

of r2, the point being made here is clearly evident from

the ratio of cost to energy shown in the last column of

Table 4.

Discussion

The focus in the hypothesis is the decreasing reliability

of the inherent energy–cost relationship as more detail

is employed. H0 is defined as the null hypothesis (i.e.

reliability does not decrease …) and H1 is defined as the

alternative hypothesis (i.e. reliability does decrease …).

Figure 1, Table 2, Table 3, Figure 2 and Figure 3

progressively demonstrate the decreasing reliability of

the relationship, and provide sufficient evidence to

conclude that H0 should be rejected in favour of H1 as

explained below.

Figure 1 indicates a strong relationship between

embodied energy and capital cost at the project level

across all 30 case studies. The derived r2 is 0.9542, but

even when a more stringent test of correlation is applied

by looking just at process analysis data, the value of r2 is

still strong at 0.8338. P-values are 1.12224 and 1.05217

respectively at 95% confidence levels.

Table 2 summarizes the correlation statistics at the

elemental group level across all case studies. Values of

r2 range down from 0.9712 (services) to 0.2188

(external alterations and renovations), and shows that

four (out of 11) elemental groups are well below the

selected test value of 0.7. The simple mean is 0.7120.

Table 2 Elemental group correlation statistics (Microsoft Excel Data Analysis Module)

Elemental

group

r r2 Adjusted r2 m p-value t-statistic

PR 0.9022 0.8140 0.7749 0.0051 5.99E-15 15.05

SB 0.9824 0.9651 0.9293 0.0200 1.27E-23 32.07

CL-ND 0.9741 0.9488 0.9125 0.0089 6.74E-24 32.83

WF-CF 0.9673 0.9357 0.9012 0.0060 7.59E-22 27.59

FT-SE 0.9717 0.9442 0.9079 0.0069 1.47E-21 26.93

SF-SS 0.9855 0.9712 0.9351 0.0024 5.39E-26 39.15

AR 0.5761 0.3319 0.1975 0.0016 1.43E-02 3.03

XP-XL 0.8645 0.7473 0.7062 0.0070 2.25E-12 12.66

XK-XS 0.4934 0.2434 0.1865 0.0041 3.02E-05 5.23

XX 0.4677 0.2188 0.0870 0.0004 2.20E-01 1.52

YY n/a n/a n/a 0.0226 4.12E-03 3.13


Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

Table 3 summarizes the correlations statistics at the

elemental level across all case studies. Here values of r2

vary between 0.9832 (external electric light and power)

and 0.0307 (water supply). Eighteen (out of 43)

elements again fall below the test value of 0.7. The

simple mean is 0.6863.

Figure 2 focuses solely on elements within a single

project (Building 24). The value of r2 is shown as 0.5217

(p-value51.3029). While a relationship still exists, it is

weaker, and below the selected test value of 0.7.

Finally, Figure 3 shows an r2 value of 0.3560 when

the energy–cost relationship is reduced to individual

materials. In this case, p-value is 2.1926.

The reason for the decreasing reliability is thought to

be a function of the level of the investigation. It is

already understood that the ratio of capital cost to

embodied energy varies widely for different materials.

Table 4 shows that this ratio ranges from 12 times

(cement) to 411 times (fabric reinforcement), with a

mean ratio of 101 times and a coefficient of variation of

Table 3 Element correlation statistics (Microsoft Excel Data Analysis Module)

Element r r2 Adjusted r2 m p-value t-statistic

PR 0.9022 0.8140 0.7749 0.0051 5.99E-15 15.05

SB 0.9824 0.9651 0.9293 0.0200 1.27E-23 32.07

CL 0.9899 0.9800 0.9398 0.0154 1.41E-24 41.49

UF 0.9491 0.9008 0.8478 0.0108 2.15E-13 18.00

SC 0.8247 0.6802 0.6279 0.0099 2.02E-09 9.72

RF 0.9179 0.8426 0.8078 0.0146 2.92E-17 18.54

EW 0.9515 0.9053 0.8688 0.0072 8.43E-19 21.23

WW 0.9660 0.9331 0.8891 0.0058 5.20E-20 28.48

ED 0.9254 0.8563 0.8083 0.0053 9.85E-15 16.14

NW 0.8856 0.7842 0.7470 0.0081 5.75E-14 13.74

NS 0.9219 0.8499 0.8170 0.0055 9.29E-16 17.26

ND 0.8862 0.7853 0.7470 0.0028 1.06E-13 13.40

WF 0.9099 0.8279 0.7923 0.0056 8.40E-15 15.27

FF 0.9440 0.8912 0.8596 0.0054 1.69E-18 20.68

CF 0.8852 0.7836 0.7438 0.0067 3.67E-14 14.36

FT 0.9631 0.9276 0.8912 0.0069 8.75E-20 23.12

SE 0.9569 0.9157 0.8552 0.0067 7.59E-12 16.38

SF 0.9486 0.8998 0.8544 0.0013 2.59E-14 19.37

PD 0.5492 0.3016 0.2141 0.0009 1.33E-03 3.99

WS 0.1752 0.0307 0.0021 0.0028 1.30E-02 3.30

AC 0.9519 0.9061 0.8653 0.0030 3.39E-18 21.73

FP 0.8649 0.7481 0.7096 0.0011 5.09E-13 13.18

LP 0.8968 0.8042 0.7634 0.0006 5.09E-16 16.59

CM 0.8004 0.6407 0.5378 0.0059 6.57E-05 6.22

GS 0.8182 0.6694 0.6625 0.0004 3.07E-05 7.16

VE 0.8062 0.6499 0.4715 0.0022 2.47E-03 4.99

TS 0.9633 0.9280 0.8008 0.0075 7.88E-07 13.67

SS 0.6223 0.3872 0.2745 0.0021 1.86E-03 4.34

AR 0.5761 0.3319 0.1975 0.0016 1.43E-02 3.03

XP 0.7314 0.5349 0.4720 0.0027 1.01E-04 5.04

XR 0.8542 0.7297 0.6812 0.0102 2.57E-10 10.59

XN 0.2585 0.0668 0.0257 0.0212 1.66E-04 5.38

XB 0.8868 0.7864 0.5954 0.0178 3.22E-03 5.29

XL 0.5425 0.2943 0.2317 0.0044 4.93E-05 5.21

XK 0.7016 0.4922 0.4389 0.0106 1.06E-05 5.83

XD 0.4822 0.2325 0.1248 0.0031 5.41E-05 5.56

XW 0.6701 0.4491 0.3671 0.0037 2.21E-05 6.23

XE 0.9916 0.9832 0.8410 0.0042 1.36E-08 29.38

XG 0.7674 0.5889 0.3308 0.0031 4.78E-03 4.82

XF 0.8305 0.6898 0.3263 0.0020 6.63E-03 4.46

XS n/a n/a n/a n/a n/a n/a

XX 0.4677 0.2188 0.0870 0.0004 2.20E-01 1.52

YY n/a n/a n/a 0.0226 4.12E-03 3.13


Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

101.32%. Yet at higher levels these discrepancies

become less important, and the mix of materials

introduces more stability.

A simple analogy can help to explain this phenom-

enon. A typical family shopping bill at the supermarket

might be $200 per week. This amount is reasonably

predictable when the shopping trolley is full. Within

that trolley is a range of commodities that vary

significantly in price (and volume). Yet, provided the

total volume is consistent per week, the total price is

also. The effect of variability in individual unit prices is

reduced by the mix. In much the same way construc-

tion cost is predictable in terms of cost/m2 GFA despite

the detailed decisions on materials and systems that

underlie it.

The relationship between embodied energy and

capital cost is interrupted to some extent by the cost

of labour. People obviously consume goods and

services to live, and therefore any process that involves

human labour involves indirect energy use. Peet and

Baines (1986) define labour in three ways: muscular

energy expended in the performance of the task; energy

content of the foods consumed by the worker for the

time spent on the task; and total quantity of food

consumed by the worker regardless of the time spent on

the task. As explained earlier, the embodied energy of

human capital is controversial and normally excluded

from energy analysis calculations. Nevertheless valuing

labour is routine in life cycle costing and therefore cost

modelling may provide a new and simpler approach to

deal with this problem.

This matter also provides some further explanation

of the decreasing relationship between embodied

energy and capital cost as the project is broken down.

Figure 2 Building 24 correlation by element


Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

At a project level, labour accounts for on average about

30–40% of the total cost. Excluding this component

would have no effect on the correlation. But as the

analysis is refined to elemental groups, elements and

ultimately items of work, so the ratio of labour to

material becomes more volatile, and hence the correla-

tion less reliable. This situation again advocates the use

of high-level data when modelling embodied energy

from capital cost.

So embodied energy and cost have little relationship

at the level of individual materials, but as the mix

within a typical building becomes more complex the

relationship appears to strengthen. This outcome

suggests that embodied energy is more reliably pre-

dicted at the project level than any selected sub-level,

which in turn is good news for people who wish to

estimate embodied energy impact. But as cost increases

so does energy, and this relationship raises serious

questions over whether the common-held view of

spending more money to realize energy efficiencies is

at all an achievable goal.

Conclusion

It is found that initial embodied energy is reliably

modelled by capital cost, but this reliability decreases

proportionally with the level of detail at which the

model is expected to operate. In other words, as the

level at which the model operates moves from the

overall project down to individual work items, so the

correlation between energy and cost drops.

Inconsistency in the ratio of embodied energy to capital

cost for individual materials is lost in the mix of

materials that typically comprise a building. This mix

strengthens the relationship, since high and low ratios

are offset against each other.

Table 4 Material correlation data (input–output-based hybrid EE model and Rawlinsons (2006))

Material item Unit GJ/unit $2006/unit Ratio

(cost: energy)

Aluminium kg 0.25 13.40 53

Brickwork (110mm thick) m2 0.56 23.20 41

Carpet, wool m2 0.74 71.54 97

Cement kg 0.02 0.20 12

Concrete roof tiles (25mm thick) m2 0.17 19.15 115

Clear float glass (6mm thick) m2 1.91 86.29 45

Concrete, 32MPa m3 5.46 144.18 26

Concrete block, hollow (200mm thick) m2 0.56 226.73 408

Concrete pipe (300mm diameter) m 0.57 61.08 107

Copper kg 0.38 28.88 76

Fibre cement sheet (6mm thick) m2 0.21 19.15 93

Fibreglass batts, R2.5 m2 0.17 5.18 31

Lead kg 0.15 6.67 44

MDF/particleboard (25mm thick) m2 0.53 28.46 53

Membrane m2 0.24 29.72 122

Paint, water based m2 0.05 12.88 273

Plasterboard (13mm thick) m2 0.16 15.32 97

Plastic kg 0.16 19.37 119

PVC water pipe (25mm diameter) m 0.16 15.19 95

Reflective foil m2 0.14 3.07 22

Reinforcement fabric, F62 m2 0.01 3.60 411

Sand m3 0.53 23.25 44

Screenings m3 0.47 48.87 104

Slate (25mm thick) m2 0.65 99.06 151

Steel kg 0.07 1.08 15

Steel, stainless kg 0.45 6.61 15

Terracotta roof tiles (20mm thick) m2 0.59 26.42 45

Tile, ceramic m2 0.29 38.52 131

Timber, hardwood (100 6 100mm) m 0.21 6.54 31

Timber, softwood (100 6 100mm) m 0.11 9.91 91

Toughened glass (6mm thick) m2 2.55 134.72 53

UPVC pipe (100mm diameter) m 0.16 46.23 289

Vinyl (2mm thick) m2 0.66 14.31 22


Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

The typical mix of elements is critical. The project

level equation (y50.0071x) requires all elements to be

present. Where this is not the case, such as no lifts or no

upper floors for example, the accuracy of the project-

based slope coefficient is reduced. Elemental groups

are the next best method, but the slope coefficients can

still be affected where key elements are absent.

Therefore the element-based slope coefficients provide

a more practical approach and are advocated as the

basis of a predictive model. Care must be taken,

however, with elements that have low reliability and

with external works where the ratio with building GFA

is subject to significant variation.

In all observed instances, whether at project,

elemental group or element levels, higher energy values

give rise to higher costs and vice versa. Cost is a strong

predictor of energy. While at the work item level it is

feasible to spend more capital to purchase materials or

systems with lower embodied and/or operating energy

demand, and thus introduce efficiency and value for

money, at the higher levels of aggregation this is

unlikely to be significant.

This finding leads to the conclusion that optimiza-

tion of energy and optimization of cost are not mutually

exclusive goals. The very strong correlations between

energy and cost, and particularly between embodied

energy and capital cost, suggest that this bond cannot

be upset by individual decisions made at more refined

levels of detail. It therefore may be more important to

look elsewhere for efficiencies. In the case of embodied

Figure 3 Common material item correlation


Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

energy, savings can be found by building less. This may

arise from more efficient floor plans (i.e. less non-

functional area), more efficient plan shape (i.e. lower

wall/floor ratios), more efficient internal layouts (i.e.

more open plan design), and higher floor densities (i.e.

people/m2). In the case of operating energy, savings can

be found, in addition to building less, by reducing

demand through use of natural lighting and ventilation,

proper orientation, reduced operating hours (where

possible) and intelligent control systems.

This research supports the proposition that energy

optimization and cost optimization are compatible

activities. The decisions that result from investigating

different material options do not significantly lower

total energy demand unless the cost of these options is

also proportionally reduced. Spending more initially to

save embodied energy, for example, is unlikely to be a

productive quest, although undoubtedly instances

where this might occur at a more detailed level within

a project can be imagined. A better strategy for

reducing embodied energy is simply building less.

Having said that, this study comprised case studies

that did not have significant components of recycled or

reused materials. Clearly where this is achieved, initial

embodied energy can be markedly reduced since the

upstream energy impacts would have been previously

counted on another project (i.e. double counting is not

valid). The cost of recycled and reused materials varies

widely. In some cases the cost can be higher than

comparable new materials, and in many other cases the

cost can be lower. The introduction of significant

amounts of recycled and reused materials may upset

the relationships found in this study.

This issue also has implications for embodied energy

value. In this study 73.3% of projects represented new

construction, and the remainder were redevelopment.

In many cases, redevelopment implies reuse of struc-

ture and other elements, so both embodied energy and

capital cost are reduced. There was no discernible

difference between the results for new and redeveloped

projects, indicating that the inherent energy–cost

relationship still applies.

The lack of public knowledge about embodied

energy has led to its absence from design decision

making and government policies. This situation has

arisen as a direct result of the complexity of the

calculations and the lack of reliable data upon which to

base them. Over time, embodied energy models have

become even more complex. This translates into

specialist knowledge requirements and a considerable

time commitment. The way forward is to develop

models that practitioners can use and apply routinely to

their projects. Only by putting useful tools into the

hands of building designers and managers can a wider

understanding of the importance of embodied energy

be gained. This also enables life cycle energy and GGE

calculation to be incorporated in the early design stages

to reduce unnecessary environmental degradation.

Acknowledgements

The authors of this paper would like to thank the kind

assistance of Associate Professor Graham Treloar from

the University of Melbourne for the use of his input–

output-based hybrid method for embodied energy

analyses and for his supervisory advice and guidance in

this research, Computerelation Australia Pty Limited for

making available LIFECOST TM software for operating

cost analyses, and Davis Langdon Australia for permis-

sion to use their data in the case studies. Without their

cooperation this research would not have been possible.

References

AGO (1999) Australian Commercial Building Sector Greenhouse

Gas Emissions 1990–2010, Australian Greenhouse Office,

Canberra.

AIQS (2002) Australian Cost Management Manual, Vol. 3,

The Australian Institute of Quantity Surveyors, Canberra.

Arnold, F. (1993) Life cycle doesn’t work. Environmental

Forum, 10(5), 19–23.

ASEC (2001) Australia 2001: State of the Environment,

Independent Report to the Commonwealth Minister for

the Environment and Heritage, Australian State of the

Environment Committee.

Ashworth, A. (1988) Making life cycle costing work.

Chartered Quantity Surveyor, 10(8), pp. 17–18.

Baird, G. and Chan, S. (1983) Energy Cost of Houses and Light

Construction, Report No. 76, New Zealand Energy Research

and Development Committee, November.

Beggs, C. (2002) Energy: Management, Supply and

Conservation, Oxford, Butterworth-Heinemann.

Bell, J.M. and Fawcett, A. (2000) Designing sustainability

into engineering education and practice: education for a

sustainable built environment. Presentation for Teaching

Sustainable Design Forum, 13 November.

Boustead, I. and Hancock, G.F. (1979) Handbook of Industrial

Energy Analysis, Ellis Horwood, Chichester.

Brown, R.J. and Yanuck, R.R. (1985) Introduction to Life Cycle

Costing, Fairmont Press, Atlanta.

Bull, J.W. (ed.) (1992) Life Cycle Costing for Construction,

Blackie Academic and Professional, London.

Chukwuma, C. (1996) Perspectives for a sustainable society.

Environmental Management and Health, 7(5), 5–20.

Costanza, R. (1980) Embodied energy and economic valua-

tion. Science, 210, 1219–24.

Crawford, R.H. (2004) Using input–output data in life cycle

inventory analysis, DPhil thesis, Deakin University,

Geelong, Australia.

Crawford, R.H. (2005) Validation of the use of input–output

data for embodied energy analysis of the Australian


Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

construction industry. Journal of Construction Research, 6(1),

71–90.

Crawford, R.H., Fuller, R.J., Treloar, G.J. and Ilozor, B.D.

(2002) Embodied energy analysis of the refurbishment of a

small detached building, in Luther, M., Dawson, A.,

Ham, J., Moore, M., Rollo, J. and Treloar, G. (eds)

Proceedings of 36th Annual Conference of the Australian and

New Zealand Architectural Science Association (ANZAScA):

The Modern Practice of Architectural Science from Pedagogy to

Andragogy?, Geelong, Victoria, Australia, November,

pp. 93–100.

Dell’Isola, A.J. and Kirk, S.J. (1995) Life Cycle Cost Data,

McGraw-Hill, New York.

Department of Natural Resources and Environment (2000)

Victorian greenhouse strategy. Discussion Paper,

Department of Natural Resources and Environment.

Fay, R. (1999) Comparative life-cycle energy studies of

typical Australian suburban dwellings, DPhil thesis,

University of Melbourne, Melbourne.

Feagin, J., Orum, A. and Sjoberg, G. (eds) (1991) A Case for

Case Study. University of North Carolina Press, Chapel

Hill.

Flanagan, R. and Norman, G. (1983) Life Cycle Costing for

Construction, Surveyors Publications, London.

Fossdal, S. (1995) Energy consumption and environmental

impacts of buildings in Norway: life cycle assessment.

International Energy Agency Energy Conservation News, 22,

1–3.

Fuller, R. (1982) Life cycle costing: a comprehensive

approach, in Brandon, P.S. (ed.) Building Cost Techniques:

New Directions, E&FN Spon, London, pp. 439–46.

Hill, R.K. (1978) Gross energy requirements of building

materials, in Proceedings of Conference on Energy Conservation

in the Built Environment, Department of Environment,

Housing and Community Department, Sydney, 15–16

March, pp. 179–190.

Hoglund, I. (1992) Energy conservation in the built

environment: the past, the future, in Proceedings of World

Building Congress, Montreal, Canada, 18–22 May,

pp. 376–7.

Irurah, D.K. and Holm, D. (1999) Energy impacts of

building construction as applied to South Africa.

Construction Management and Economics, 17, 363–74.

Kirk, S.J. and Dell’Isola, A.J. (1995) Life Cycle Costing for

Design Professionals, McGraw-Hill, New York.

Lam, J., Chow, T. and Yuen, R.K. (1992) Energy for space

heating in commercial buildings in Hong Kong, in

Proceedings of World Building Congress, Montreal, Canada,

18–22 May, pp. 352–3.

Langston, C.A. (1991) The Measurement of Life-Costs, Public

Works Department of New South Wales, Sydney.

Langston, C.A. (1994) The determination of equivalent value

in life-cost studies: an intergenerational approach, DPhil

thesis, University of Technology, Sydney.

Langston, C.A. (2005) Life-Cost Approach to Building

Evaluation, Elsevier, Oxford.

Langston, C.A. and Lauge-Kristensen, R. (2002) Strategic

Management of Built Facilities, Butterworth-Heinemann,

Oxford.

Langston, Y.L. (2006) Embodied energy modelling of in-

dividual buildings in Melbourne: the inherent energy–cost

relationship, DPhil thesis, Deakin University, Geelong,

Australia.

Lawson, W.R. (1995) Embodied energy of building.

Environment Design Guide, The Royal Australian Institute

of Architects, August, pro.2, pp. 1–6.

Lenzen, M. and Lundie, S. (2002) Input–output model of the

Australian economy based on published 1996–97

Australian input–output data, unpublished paper.

Lippiatt, B.C. (1999) Selecting cost-effective green building

products: BEES approach. Journal of Construction

Engineering and Management, 125(6), 448–55.

NAEEEC (1999) Future directions for Australia’s appliance

and equipment energy efficiency program. Discussion Paper

prepared by the National Appliance and Equipment Energy

Efficiency Committee, Commonwealth of Australia.

Odum, H.T. (1971) Environment, Power and Society, Wiley-

Interscience, New York.

Peet, N.J. and Baines, J.T. (1986) Energy Analysis: A Review

of Theory and Applications, Report No. 126, New Zealand

Energy Research and Development Committee, February.

Rawlinsons (2006) Australian Construction Handbook, 24th

edn, Rawlhouse Publishing, Perth.

Robertson, G. (1991) Energy efficient commercial buildings:

a realistic market objective. Architectural Science Review, 34,

139–42.

Shen, S. (1981) Energy and Materials Flows in the Production of

Primary Aluminium, Report for the Energy and

Environmental Systems Division Special Projects Group,

October.

Stone, P.A. (1960) The economics of building design. Journal

of the Royal Society, 123(3), 237–73.

Tellis, W. (1997) Application of a case study methodology.

The Qualitative Report, 3(3), available at www.nova.edu/

ssss/QR/QR3-3/tellis2.html (accessed 15 May 2007).

Treloar, G.J. (1996) A complete model of embodied energy

‘pathways’ for residential buildings, in Treloar, G., Fay, R.

and Tucker, S. (eds) Proceedings of the Embodied Energy: The

Current State of Play Seminar, Deakin University, Geelong,

Australia, 28–29 November, pp. 51–60.

Treloar, G.J. (1997) Extracting embodied energy paths from

input–output tables: towards an input-output based hybrid

energy analysis method. Economic Systems Research, 9(4),

375–91.

Treloar, G.J. (1998) A comprehensive embodied energy

analysis framework, DPhil thesis, Deakin University,

Geelong, Australia.

Treloar, G.J., Love, P.E.D. and Holt, G.D. (2001) Using

national input–output data for embodied energy analysis of

individual residential buildings. Construction Management

and Economics, 19, 49–61.

Treloar, G.J., Ilozor, B.D. and Crawford, R.H. (2002)

Modelling energy use and greenhouse gas emissions

associated with commercial building construction, in

Luther, M., Dawson, A., Ham, J., Moore, M., Rollo, J.

and Treloar, G. (eds) Proceedings of 36th Annual Conference

of the Australian and New Zealand Architectural Science

Association (ANZAScA): The Modern Practice of


Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

Architectural Ccience from Pedagogy to Andragogy?, Geelong,

Victoria, Australia, November, pp. 533–40.

Van Pelt, M.J.F., Kuyvenhoven, A. and Nijkamp, P. (1990)

Project appraisal and sustainability: methodological chal-

lenges. Project Appraisal, 5, 139–58.

Verbeek, I. and Wibberley, L. (1996) Data collection, in

Proceedings of 1st National Conference on LCA, 29

September–1 March, pp. 1–7.

Wainwright, W.H. and Wood, A.A.B. (1981) Practical

Builders’ Estimating, Hutchinson, London.


Dow

nloa

ded

by [

Rad

boud

Uni

vers

iteit

Nijm

egen

] at

05:

27 0

3 N

ovem

ber

2014

reliability of building embodied energy modelling: an analysis of 30 melbourne case studies

Documents