You are on page 1of 9

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/311767147

Reducing uncertainty in predicting embodied


energy of HVAC systems

Conference Paper June 2016

CITATIONS READS

0 5

5 authors, including:

Mike Medas Andrew Cripps


University of Reading AECOM, UK
3 PUBLICATIONS 2 CITATIONS 25 PUBLICATIONS 307 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

My PhD View project

Resource efficiency of building services View project

All content following this page was uploaded by Mike Medas on 20 December 2016.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
Reducing Uncertainty in Predicting
Embodied Energy of HVAC Systems

Mike Medas Dave Cheshire Andrew Cripps, PhD


Student Member ASHRAE

John Connaughton, PhD Michael Peters, PhD

ABSTRACT
An understanding of uncertainty in building services design can help optimize time, cost, technical quality and environmental impacts, but presents special
challenges when considering embodied energy or carbon, particularly for HVAC components composed of multiple raw materials. A review of relevant
literature suggests that analytical uncertainty propagation (AUP) may in certain conditions have advantages over sampling-based methods for the
uncertainty analysis of initial embodied impacts of composite building services components. A case study shows that the use of AUP combined with
parametric prediction could enable estimation of embodied energy of composite building services components incorporating uncertainty analysis. This
approach could support rapid estimation of whole life energy and carbon impacts of building services systems and components, implemented using BIM.

INTRODUCTION

Efficient design of buildings and their services should involve minimizing uncertainty in project costs and
time, material waste, environmental impacts and technical quality, if uncertainty is defined as unexpected variation
against target values. The progression from conceptual to detailed design removes some but not all uncertainty,
especially about environmental impacts. Therefore the ability during early building design to quantify uncertainty in
estimates of lifetime energy use and carbon emissions of HVAC systems has economic and environmental benefits.
Uncertainty about the operational energy impacts of a given building services design is a known challenge for
which current solutions include combining operational energy simulation with building information modelling (BIM)
(GSA, 2015). BIM is also expected to reduce uncertainty by reducing design errors and wastage of time, cost and
materials (HM Government 2013, 2015). Uncertainty about embodied energy (EE) or embodied carbon (EC) impacts
of building services is more challenging and less well understood. The significance of embodied energy and carbon
emissions for the built environment is growing, as modern buildings with greater operational energy efficiency have
relatively higher energy and carbon impacts outside the operational stage of building life (Dixit et al 2010).
Uncertainty about EE and EC is challenging firstly because of gaps in available data and methods for
estimating embodied environmental impacts of building services, especially for HVAC components composed of
multiple raw materials (Medas et al. 2015). Secondly, EE and EC studies and others based on environmental life cycle
assessment studies (LCA) have mainly considered the building structure and envelope, although building services may
represent up to 15% of initial EC and 25% of initial EE of a typical office building (Cole and Kernan 1996; Medas et
al. 2015). Gaps in data and methods make it harder to quantify the level of uncertainty around those estimates. We
have previously outlined a parametric method to estimate EC for composite building services components that avoids
Mike Medas is a Research Engineer in the School of the Built Environment, University of Reading, Berkshire, UK, funded by AECOM Ltd. UK and the
Engineering and Physical Sciences Research Council, UK. Prof. John Connaughton and Michael Peters are faculty members in the School of the Built
Environment at the University of Reading, Berkshire, UK. Dave Cheshire and Andrew Cripps are professionals working for AECOM Ltd, UK.
the cost and complexity of a full LCA (Medas et al. 2015). The addition of uncertainty analysis can make this method
more robust.
In this paper we review the use of uncertainty analysis in construction-based LCA and present a case study
into uncertainty analysis of embodied energy of composite building services components. While the study primarily
considers embodied energy, the proposed approach is equally applicable to the analysis of embodied carbon.

LITERATURE REVIEW

The embodied energy of a building or building element refers to energy consumed throughout its life cycle
stages from raw material extraction to final disposal, excluding operational use (Ramesh et al. 2010). EE studies face
the types of uncertainty associated with their parent discipline of LCA, namely parameter uncertainty, model
uncertainty and scenario uncertainty (Lloyd and Reis 2007). Uncertainty in LCA studies can either be reduced or
incorporated into the analysis (Heijungs and Lenzen 2014). Reduction can include obtaining more accurate data, using
expert views to define data quality (Heijungs and Huijbregts 2004) and standardizing LCA methods (BSI 2003,
2006(a), 2006(b), 2011). Incorporation uses statistical methods, including sampling-based Monte Carlo (MCS) and
Latin Hypercube simulation (LHCS), fuzzy set theory and analytical uncertainty propagation (AUP) (Groen et al.
2014).
According to a recent review article, uncertainty is measurable at several stages of the LCA model but most
often measured at the life cycle inventory stage; the complexity of LCA studies means that all types of uncertainty are
not usually covered within one study; and the most frequently used methods of uncertainty analysis are sampling
based (Lloyd and Reis 2007). Monte Carlo simulation (MCS) is the most widely used sampling method. Pseudo
random numbers are drawn from a set of input parameters with known probability distributions to produce a sampled
distribution of an output parameter (Groen et al. 2014). The benefits of MCS are its accuracy, its ability to compare
uncertainty distributions of impacts of alternative products and its availability within commercial LCA software. Its
main disadvantages are the large number of simulation runs needed, typically at least 1000 runs for each possible
parameter combination, and the difficulty in calculating the contribution of each input parameter to overall output
variance when a large number of input parameters exist (Heijungs and Lenzen 2014).
By contrast, AUP is based on a type of calculus, first-order Taylor Series approximation. Its advantage is that
only one calculation is needed to estimate the contribution of each input parameter to output variance (Morgan and
Henrion 1990). In Equation 1, the variance of output parameter y is approximately equal to the sum of the
contributions to variance of each input parameter x, which in turn is expressed as the product of the input variance of
x and the absolute value of the partial derivative of y with respect to x. In this form, this equation assumes there is no
covariance between input parameters.

! !" !
!!! ! (1)
!!! ! !

The disadvantages of AUP compared to sampling-based methods are firstly that it does not provide a
probability distribution for output uncertainty. Thus the mean environmental impacts of two products can be
compared, but not their probability distributions. Secondly, AUP is significantly less accurate than sampling if input
uncertainty levels exceed a threshold, estimated by recent studies at a coefficient of variation of either 5% (Groen et
al. 2014) or 10% (Heijungs and Lenzen 2014). Thirdly, if covariance exists between input parameters, a more
complex equation is needed. However, AUP has been used in an LCA study of input parameters for whole buildings
(Hoxha et al. 2014) and this approach can be adapted to input parameters of composite building services components.
How might uncertainty analysis help estimation of EE of building services? Arguably, the key challenge is
parameter uncertainty in the initial embodied, or cradle-to-gate stage of component life, because (a) estimation of
recurring EE is less complex and (b) model and scenario uncertainty have been reduced by recent ISO LCA standards
specific to buildings and building products (BSI 2011, 2013). Parameter uncertainty represents a gap because life cycle
inventory (LCI) databases generally exclude composite building services components and few Environmental Product
Declarations, or EPDs, exist for building services (EPD International 2015). This means that the material and energy
inputs at the initial embodied stage for most composite building services components are unknown. If an EPD or
bespoke LCA study of a component exists, a way is needed to measure the uncertainty involved in using that study as
a benchmark to estimate EE of other components of a similar class.
Ideally, uncertainty should be reduced via obtaining more accurate data, however this is more difficult for
embodied than operational energy. The advent of real-time data from smart buildings and appliances has lowered
costs and increased access to empirical data on operational rather than embodied energy use. Uncertainty reduction
using expert elicitation on data quality is already used to inform values in the Ecoinvent LCI database (Frishcknecht et
al. 2004). However the lack of data on material and energy inputs used by component manufacturers is a quantitative
rather than qualitative gap. This suggests that uncertainty should instead be incorporated via a statistical approach.
Here AUP has advantages over MCS for rapid estimation of EE during early building design due to its computational
simplicity, as long as the qualifying assumptions on covariance and input uncertainty are met. The analysis of
parameter uncertainty should begin with a broad contribution analysis to exclude parameters least likely to affect
output values. A sensitivity analysis to select parameters to which variations in output values are most sensitive is then
done by comparing the effects on total output of small variations in each input parameter. Next, uncertainty
propagation measures the effect of each parameter on output variation and uncertainty contribution analysis is used to
express that effect as a percentage of overall output variation. For initial EE of a composite building services
component, the relevant input parameters are found in Equation 2, which calculates EE for the product stage of raw
material supply, transport and manufacturing as defined in ISO 15804 (BSI 2013).

! ! !
EE!! = !!! ! + !!! ! + !!! !" (2)

In Equation 2, EECC = embodied energy of a composite component, m =mass, and kR, kT and kMF are the
energy intensity coefficients for raw material supply, transport and manufacturing respectively. kT is the product of
distance travelled and the energy intensity coefficient for the mode of transport, while kMF is the product of energy
used in manufacturing and the energy intensity coefficient for supply of electricity and/or heat.

CASE STUDY

In a previous study (Medas et al. 2015) we showed that variations in embodied carbon of a range of fan coil
units (FCU) were associated with variations in total cooling capacity and mass, indicating that EC can be predictable
for a generic type of FCU once mass and total cooling capacity are known. Using data from that case study for the
empirically calculated EE of a manufacturers range of horizontal, ducted, water-side FCUs, with coefficients kR, kT
and kMF shown in Table 1, the present study uses uncertainty analysis to answer three questions: (1) How uncertainties
in input parameters, namely the material mass and energy intensities of raw materials, affect the sensitivity and
uncertainty of initial EE across the available power and size range of a composite component; (2) how the results can
support estimation of EE of similar products made by other manufacturers for which more limited data is available;
and (3) how the results can inform product choice based on estimated whole-life energy use.
A broad contribution analysis excludes the input parameters representing the contribution to EE of transport
and final manufacturing (mkT and mkMF) from further consideration, because they only represent respectively around
1% and 3% of the total EE of a fan coil unit. This leaves 10 parameters, for the masses (m) and carbon coefficients
(kR) of five raw materials: Galvanized steel, copper, aluminium, general steel and polyurethane foam. A sensitivity
analysis shows that the value of m * kR for each raw material is in almost all cases worth more than 5% of total EE,
therefore as there are only five pairs of these parameters, all can be considered within the uncertainty analysis.
Uncertainty propagation is applied using Equation 1 under three scenarios. In Scenario 1, the mass of each
raw material m may vary by 24.5% while coefficient kR does not vary. Scenario 2 is similar to Scenario 1, except that
galvanized steel mass varies only by 12.25%. In Scenario 3, all raw material masses m vary by 24.5% and coefficient
kR also varies by 24.5%, except in the case of copper, for which kR is assumed to be deterministic with no variation.
To convert these ranges of input uncertainty into variances, it is assumed that nothing is known about their
probability distributions other than the upper and lower percentage limits for variation, while the nominal values for
each parameter are taken from a widely used LCI database for values of kR (Hammond and Jones 2011) and from
empirical manufacturers data for values of m. A triangular distribution is assumed using the upper, lower and nominal
values and Equation 3 to calculate the variance:

!! ! ! ! ! ! ! ! !"!!"!!"
= (3)
!"

In equation 3, a and c are the lower and upper values and b is the nominal value.
Table 1 shows the embodied energy coefficients used in the case study and their embodied carbon (Global
Warming Potential) equivalents. While the results discussed in the next section mainly consider energy impacts, similar
trends are obtained if calculated using the embodied carbon coefficients for the case study data.

Table 1. Carbon and Energy Coefficients Used for Case Study


Raw Material or Process Input Embodied Energy Embodied Carbon (GWP)
Fan coil unit casing - galvanized sheet
steel1 22.6 MJ/kg (19.43 Million Btu/ton) 1.54 kg CO2e/kg
FCU casing - polyurethane insulation foam1 102.1 MJ/kg (87.79 Million Btu/ton) 4.84 kg CO2e/kg
Coils - general aluminium1 155 MJ/kg (133.28 Million Btu/ton) 9.16 kg CO2e/kg
Coils - copper tube and sheet1 42 MJ/kg (36.11 Million Btu/ton) 2.71 kg CO2e/kg
Fan/motor assembly - general steel1 20.1 MJ/kg (17.28 Million Btu/ton) 1.46 kg CO2e/kg
Fan/motor assembly - general aluminium1 155 MJ/kg (133.28 Million Btu/ton) 9.16 kg CO2e/kg
Fan/motor assembly - copper tube & sheet1 42 MJ/kg (36.11 Million Btu/ton) 2.71 kg CO2e/kg
UK Electricity used in FCU manufacture2 0.65 MJ/kg (0.56 Million Btu/ton) 0.07 kg CO2e/kg
UK natural gas used in FCU manufacture3 0.04 MJ/kg (0.03 Million Btu/ton) 0.002 kg CO2e/kg
Transport-UK supplier to factory3 0.75 MJ/tonne/km (1038.7 Btu/ton/mile) 0.11 kg CO2e/tonne/km
Notes: (1) Hammond and Jones, 2011, (2) DECC 2015, (3) DEFRA 2015 (0.17 kg CO2e/ton/mile)

RESULTS

Figure 1 (a) compares the relative distribution of material mass with that of embodied energy and carbon for
a fan coil unit of average mass and cooling power. It shows that while galvanized steel casing makes up over 74% of
mass it only makes up 50.4% of EC and 46% of EE, because the non-ferrous materials are 1.8-5.9 times more carbon
intensive and 1.9-6.9 times more energy intensive than steel. Figure 1 (b) shows how embodied energy rises
proportionately with total cooling capacity across the FCU size range. The error bars represent output uncertainty at 2
standard deviations away from the calculated value for embodied energy, illustrating Scenario 1, in which input mass
of all raw materials varies by 24.5%. This level of uncertainty does not alter the close association between variations
in total cooling capacity, shown by the blue line, and variations in embodied energy, shown by the red columns.
Figure 2 breaks down this output uncertainty by contribution of each raw material over the FCU range. When
all material quantities are equally uncertain, the distribution of output uncertainty reflects the distribution of embodied
energy by material. Thus galvanized steel, while making up 72-79% of input mass, is responsible for only 48-57% of
output uncertainty, because it is less energy intensive than aluminium, copper and polyurethane.

Figure 1 (a) Material mass, initial EC and initial EE percentage breakdown of an average, water-side fan
coil unit of mass 69.1 kg (152.3 lb) and maximum total cooling capacity 5kW; and (b) Available range of fan coil units
for one manufacturer by total cooling capacity and estimated embodied energy with error bars showing uncertainty.

Figure 2 Uncertainty of embodied energy by raw material across FCU range, Scenario 1

In Figure 3(a) and (b), a comparison of Scenarios 1 and 2 shows that by halving the input uncertainty of
galvanized steel mass, overall uncertainty falls by 25% and the share of galvanized steel mass in total uncertainty falls
from 50% to 33%. Comparing Scenario 1 to 3 in Figure 3 (a), total uncertainty rises by 95% because for most
materials, the value of kR is now just as uncertain as that of m. The absolute uncertainty value of copper is the same in
all scenarios as shown in Figure 3(a) but its relative share varies from 6.5% to 8.6% to 3.3% as shown in Figure 3 (b).

Figure 3 Uncertainty of embodied energy in scenarios 1-3, showing (a) absolute and (b) relative shares
of output uncertainty, for an FCU of mass 46 kg (101 lb), mean EE of 2075 MJ (2 million Btu) and TCC of 2.5 kW.
The scenarios show that the AUP method can estimate the effects of varying levels of input parameter
uncertainty on EE output with minimal calculations when combined with information on EE sensitivity to input
parameters. How then might it help estimate the EE of a composite building services component when only the mass,
power rating and main raw materials are known?
We assume that the input parameters, mean EE value and EE uncertainty range are known for a generic
composite building services component C1 of rated output power X. Another component, C2, of the same type and
rated power has similar proportions of raw materials and energy intensities per material but 15% more mass.
Parametric EE estimation based on component C1 suggests that component C2 may have 15% more EE, while
sensitivity and uncertainty analysis using AUP quantifies the effect of each input parameter on the uncertainty of EE
of C2. This allows the prediction that if the difference in mean EE between C1 and C2 does not exceed the combined
output uncertainty ranges of C1 and C2, there may not be any significant difference in EE between C1 and C2.
How could this method assist selection of composite components with optimal whole life energy
consumption? In Figure 4 the three uncertainty scenarios for initial embodied energy are applied to three fan coil
units, (A. B and C) each with total cooling capacity of 2.5 kW but with varying mass, EE and operational energy
efficiency as shown in Table 2. A, B and C are assumed to have similar supply chains, so that the mean EE values of
FCUs B and C are estimated in proportion to their respective material masses based on the known EE value and mass
of FCU A, which is empirically calculated. For simplicity, the estimated operational energy use of each FCU excludes
any uncertainty because all three FCUs are assumed to have the same demand profile, operating period and
management conditions. Recurring EE from renewal, replacement and end of life of each FCU is also not included.
Figure 4 shows that for the most operationally energy efficient product, FCU A, mean EE exceeds OE by
17% over a 10 year life. Whilst OE of FCU A is 8% lower than that of FCU B, its mean EE is 43% greater, making
As whole life energy use 14% higher than that of B. However, in all 3 scenarios, the combined EE uncertainty range
of A and B exceeds the mean difference between them in whole life energy, therefore it is impossible to say whether
this difference is significant. Conversely, FCU C uses significantly more energy than A or B because its OE
consumption is more than double that of A or B and the difference between its whole life energy use and that of A or
B in all three scenarios exceeds the combined uncertainty range of either C+A or C+B. Therefore a component
chosen because its rated operational energy efficiency exceeds that of an alternative may not be significantly more
efficient in whole life energy use owing to the mean value and uncertainty range of its embodied energy.

Figure 4 OE and EE of three fan coil units of similar total cooling capacity but varying mean EE and OE
over ten years of operation, with error bars showing uncertainty at two standard deviations away from the mean value
of EE using Scenarios 1-3.
Table 2. Embodied and Operational Energy of Three Fan Coil Units
Parameter FCU A FCU B FCU C
Total mass - kg (lbs) 45.8 (100.9) 32 (70.5) 39 (85.9)
Mean embodied energy - kWh (MJ) 576.4 (2,075) 402.9 (1,450.6) 491.1 (1767.9)
Specific fan power - Watts/litres/sec 0.3 0.31 0.46
Fan motor type EC/DC EC/DC AC
Maximum input power - Watts 45 49 105
Operational energy use over 10 years - kWh (MJ)1 493 (1,774.8) 537 (1,933.2) 1,150.6 (4,142.2)
Note: (1) Notional OE consumption (kWh) = input power (W)*12 (hrs)*365.25 (days)* 0.25(utilization rate)*10(yrs)* 0.001 (conversion factor)

DISCUSSION

The proposed method adds an embodied energy dimension to the criteria used to select composite HVAC
components. If combined with data on the average mass and EE per floor area of distribution systems associated with
alternative HVAC systems, this method could allow system-level comparison of whole life energy impacts. It is
relevant then to consider how the method might feasibly be implemented in the construction industry.
In an era of smart buildings, clients and policy makers expect new technology to support cost-effective and
environmentally friendly results. Awareness is growing that this should embrace whole-life energy use, as shown by an
industry report on green BIM in the USA showing that the inclusion of LCA in BIM is seen as highly important by
nearly half of firms (McGraw Hill 2010). BIM can enable measurement of lifetime energy and carbon without the cost
or complexity of LCA software, as shown by academic studies (Capper et al 2012) and recent BIM-integrated LCA
applications (Tally 2015; BRE 2015). However, a strong case exists to improve the capability of these applications:

1. Input parameters on embodied impacts come from existing LCA databases that mainly cover raw materials
and not composite components. This could be improved by using a parametric method to estimate EE or EC
of composite components. Also, the design options generated do not yet include uncertainty ranges for EE,
EC or other environmental impacts, but this could be achieved using AUP-based uncertainty propagation.
2. Interoperability and data sharing between BIM, energy models, facilities management and the internet of
things (IOT) are all aims for Level 3 BIM (HM Government 2015). If the BIM object for a building services
component holds data on initial EE or EC and rated OE or OC, this can be augmented with real-time data
during building use on operational and recurring embodied energy or carbon. This data can assist post-
occupancy energy reviews or energy retrofits of existing projects by providing whole-life energy and carbon
data as well as help improve accuracy of estimating energy and carbon impacts of future projects.

CONCLUSIONS AND RECOMMENDATIONS

This study has shown that the uncertainty analysis using analytical uncertainty propagation can in principle be
combined with parametric estimation of embodied energy in order to generate useful predictions of embodied energy
values of composite building services components. A case study of horizontal, ducted fan coil units showed that the
sensitivity and uncertainty contribution of various input parameters could be quantified in order to explain the
possible range of cradle-to-gate embodied energy of a generic composite component based on limited empirical data.
Such an approach could in principle be integrated into BIM and augmented by real-time data of whole life
system and component perpormance, enabling rapid estimation of whole life energy and carbon impacts of alternative
design options combined with uncertainty analysis of these impacts.
In order to validate the AUP-based method proposed in this study, a comparison should be made with other
statistical methods of uncertainty propagation for embodied energy of composite components. This could determine
more precisely the range of conditions within which AUP can be sufficently accurate to produce robust results.
ACKNOWLEDGMENTS

The authors would like to thank the Engineering and Physical Sciences Research Council (EPSRC) and
AECOM Ltd., UK, for funding this project.

REFERENCES

BRE Impact - integrated material profile and costing tool. http://www.impactwba.com/


British Standards Institution:
BS EN 14040: Environmental management - Life cycle assessment - Principles and framework, London: BSI; 2006
BS EN 14044: Environmental management - Life cycle assessment - Requirements and guidelines, London: BSI; 2006
BS EN 15084: Sustainability of construction works - Environmental product declarations - Core rules for the product
category of construction products, London: BSI; 2013
BS EN 15978: Sustainability of construction works - Assessment of environmental performance of buildings - Calculation
method, London: BSI; 2011
Capper, G et al, 2012. Incorporating embodied energy in the BIM process. In CIBSE ASHRAE Technical Symposium, 18-19
April 2012, London, UK. London: CIBSE; 2012
Cole, R., and Kernan, C. 1996. Life-cycle energy use in office buildings. Building and Environment 31(4):307-317
Department for Environment, Food and Rural Affairs. 2015. UK government conversion factors for company reporting.
http://www.ukconversionfactorscarbonsmart.co.uk/
Department of Energy and Climate Change. 2015. Digest of United Kingdom Energy Statistics.
https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/450302/DUKES_2015.pdf
Dixit, M., et al., 2010. Identification of parameters for embodied energy measurement: A literature review. Energy and
Buildings. 2010; 42: 1238-1247
EPD International AB, Stockholm, Sweden. http://www.environdec.com
Frischknecht, R et al., 2004. The ecoinvent database: Overview and methodological framework. International Journal of Life
Cycle Assessment. 2005; 10(1): 3-9
Groen, E., et al, Methods for uncertainty propagation in life cycle assessment. Environmental Modelling & Software. 2014; 62:
316-325
Hammond G & Jones C., Inventory of Carbon and Energy (ICE). Bath, UK: Bath University, 2011.
http://www.circularecology.com/embodied-energy-and-carbon-footprint-database.html#.VlsWUmRsEu0
Heijungs, R and Lenzen, M., 2014. Error propagation methods for LCA - A comparison. International Journal of Life Cycle
Assessment. 2014; 19: 1445-1461
HM Government. 2013. Construction 2025.
https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/210099/bis-13-955-construction-
2025-industrial-strategy.pdf
HM Government. 2015. Digital built Britain: Level 3 building information modeling - strategic plan.
https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/410096/bis-15-155-digital-built-
britain-level-3-strategy.pdf
Hoxha, E, et al., 2014. Method to analyse the contribution of material's sensitivity in buildings' environmental impact.
Journal of Cleaner Production. 2014; 66:54-64
Lloyd, S and Ries, R., 2007. Characterizing, propagating and analyzing uncertainty in life-cycle assessment. Journal of
Industrial Ecology 2007; 11 (1):161-179
McGraw Hill Construction. 2010. Green BIM - How building information modelling is contributing to green design and
construction. Bedford, MA. http://www.aia.org/aiaucmp/groups/aia/documents/pdf/aiab085971.pdf
Medas, M., et al., 2015, Towards BIM-integrated, resource efficient building services. In CIBSE Technical Symposium.
London: CIBSE.
Morgan, M and Henrion, M., 1990. Uncertainty: A guide to dealing with uncertainty in quantitative risk and policy analysis.
Cambridge University Press, Cambridge.
Ramesh, T., Prakash, R., and Shukla, K. 2010. Life cycle energy analysis of buildings: An overview. Energy and Buildings 42:
15921600
Tally - LCA app for Autodesk Revit. http://choosetally.com/
U.S. General Services Administration. 2015. BIM Guide 05 - Energy performance.

View publication stats

You might also like