You are on page 1of 20

Energy & Buildings 201 (2019) 70–89

Contents lists available at ScienceDirect

Energy & Buildings


journal homepage: www.elsevier.com/locate/enbuild

Optimal renewable energy supply choices for net-zero ready


buildings: A life cycle thinking approach under uncertainty
Hirushie Karunathilake a,b, Kasun Hewage a,∗, Joshua Brinkerhoff a, Rehan Sadiq a
a
School of Engineering, University of British Columbia (Okanagan Campus), 1137 Alumni Avenue, Kelowna, BC V1V 1V7, Canada
b
Department of Mechanical Engineering, University of Moratuwa, Katubedda, Moratuwa 10400, Sri Lanka

a r t i c l e i n f o a b s t r a c t

Article history: The increasing concerns about the environmental and economic impacts of conventional centralised en-
Received 28 March 2019 ergy generation and fossil fuel usage have prompted an interest in renewable-based decentralised energy
Revised 28 June 2019
systems. Implementing such systems at building level can facilitate the development of net-zero energy
Accepted 16 July 2019
buildings. Energy system planning is a multi-faceted problem that involves technical, economic, envi-
Available online 17 July 2019
ronmental, and social dimensions, and affects multiple stakeholders at different levels. A multi-objective
Keywords: optimisation approach is needed to identify the optimal energy choices at building level, while paying at-
Hybrid renewable energy systems tention to stakeholder priorities and other constraints. The objective of this study is to develop a model to
Net-zero buildings identify the optimal mix of renewable energy (RE) while also accounting for uncertainties, which can be
Optimisation integrated at building level with life cycle thinking. A framework was proposed for planning an optimised
Fuzzy logic hybrid RE system at building level to support the net-zero development goals. The optimisation model
Life cycle assessment
was developed considering the objectives of minimising energy system cost, maximising operational cost
savings, minimising the life cycle environmental impacts, and maximising the RE fraction. A combinato-
rial optimisation approach was adopted to reflect the practical engineering aspects of energy planning
problems based on technologies available in the market. The developed framework was demonstrated
through a case study conducted for an average multi-unit residential buildings (MURB) located in British
Columbia, Canada. The results indicated that under the defined stakeholder priorities and constraints,
ground source heat pumps and solar photovoltaics (PV) are the optimal energy choices for MURB, and
the optimal energy system combination supplied 44% of the building’s energy demand through RE. The
findings will inform and guide community developers and other stakeholders with an interest in residen-
tial buildings, on the most suitable clean energy options for their building project during the pre-project
planning stage.
© 2019 Elsevier B.V. All rights reserved.

1. Introduction future resource scarcity and increasing energy prices have arisen
[3–4]. These conditions have a negative effect on the long-term en-
The drive for increased use of renewables in energy generation ergy security and economic growth of communities [2]. Localised
has been influenced by the rising global energy demand and re- energy initiatives based on renewable energy (RE) have been pro-
lated environmental concerns [1]. The use of fossil fuels is asso- posed as a solution to the above issues. A region’s energy de-
ciated with many environmental impacts such as climate change mand is created by several different sectors, including buildings
[2]. As a result, authorities all across the world have subscribed and households, transportation, industry, as well as agriculture,
to ambitious emissions reduction initiatives. Moreover, global fos- forests, and fisheries [5]. Among the above sectors, building opera-
sil fuel resources are being consumed rapidly, and concerns about tions are accountable for 32% of the world’s total final energy use
and 19% of the energy-related GHG emissions, as well as around
one-third of black carbon emissions. Thus, focusing on building-
Abbreviations and units: CAD, Canadian dollars; GJ, gigajoule; kW, kilowatt; kWh,
level clean energy initiatives is critical, during both new construc-
kilowatt-hours; MWh, megawatt hours; O&M, operations and maintenance; USD,
United States dollars. tion and renovation projects [6].

Corresponding author at: School of Engineering, The University of British The net-zero energy (NZE) concept has become popular in the
Columbia | Okanagan Campus, EME4287 - 1137 Alumni Avenue, Kelowna, BC, V1V recent times, with the growing interest in clean energy initiatives.
1V7, Canada. Under this concept, the energy needs of a community or a building
E-mail addresses: hirushiek@uom.lk (H. Karunathilake), kasun.hewage@ubc.ca (K.
are reduced through energy saving and conservation measures, and
Hewage), joshua.brinkerhoff@ubc.ca (J. Brinkerhoff), rehan.sadiq@ubc.ca (R. Sadiq).

https://doi.org/10.1016/j.enbuild.2019.07.030
0378-7788/© 2019 Elsevier B.V. All rights reserved.
H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89 71

a Northern region with a cold climate and seasonal variations. The


Nomenclature Canadian residential sector accounts for 17% of the energy use and
14% of the GHG emissions in the country [20]. At present, the en-
AOS annual operational savings ergy demand of Canadian residential and commercial building sec-
BC British Columbia tors are primarily supplied through grid electricity and natural gas.
COG centre of gravity In the residential sector, the shares of electricity and natural gas
DPP discounted payback period are 41.3% and 48.5% respectively [21]. For commercial and insti-
GHG greenhouse gas tutional sector, electricity and natural gas shares are 49.3% and
GSHP ground source heat pumps 47.3% respectively [22]. In addition to the above sources, wood,
LCA life cycle assessment coal, propane, and heating oil are used for residential energy ap-
LCC life cycle costing plications. The main end uses of the residential sector are space
LCI life cycle inventory and water heating, space cooling, lighting, and appliances. In the
LCIA life cycle impact assessment attempt to develop net-zero buildings, the above energy end uses
MODM multi-objective decision making can be supplied via renewable energy sources, such as solar, wind,
MOO multi-objective optimisation biomass, and geothermal energy. However, in Canada as well as in
MURB multi-unit residential buildings other parts of the world, it is necessary to identify the resource po-
NZE net-zero energy tential and supply reliability, as well as manage the economic chal-
PV photovoltaics lenges and risk perception effectively while ensuring environmen-
RE renewable energy tal benefits when implementing RE systems for buildings. With
RES renewable energy sources all of the above constraints and issues, planning RE-based energy
RET renewable energy technologies supply for NZE buildings is a multi-dimensional problem that is
ST solar thermal further complicated by the involvement of multiple stakeholders.
TEC total energy system cost Moreover, the parties who incur the costs and who reap the ben-
TFN triangular fuzzy number efits may not necessarily be the same. While studies have been
conducted on developing optimised energy systems at both build-
ing and community level, these studies have limitations in terms
the remainder is met using RE [7]. On-site renewable energy gen- of practical applicability [23,24]. The practical engineering implica-
eration is the preferred option, rather than purchasing off-site RE tions of integrating clean energy in buildings have not been con-
credits [7]. Hybrid renewable energy systems and their implemen- sidered in most studies, especially in combination with a decision
tation have been explored in several studies across the world, at making perspective based on life cycle thinking. At present, there
building and community levels [8–11]. Building-level RE generation is a lack of decision making frameworks for use during the pre-
can be used to replace the conventional energy supply in net-zero planning stages of community planning that consider the life cy-
energy buildings (NZEB). However, the possibility of reaching NZE cle economic and environmental impacts of energy use at building
status at building level is constrained by several factors, including level for optimal energy system design. In addition, the previous
limited resource availability and lack of funding [12,13]. In addition, studies and existing energy planning tools neglect the effect of un-
identifying the best energy supply solutions at building level is a certainties. Thus, the resulting energy mix solutions are determin-
challenging task due to the uncertainty surrounding RE supply and istic. Renewable energy system feasibility assessment models that
system performance, as well as the long term impacts of energy can be used during building and community project planning stage
systems throughout their life cycle [5,14]. Effective decision mak- will guide decision makers towards more effective energy choices
ing is required in assessing how much the building energy con- and will improve the penetration of RE at community level.
sumption can be reduced, and what fraction of the building’s en-
ergy demand can be supplied via RE, while maintaining economic, 1.2. Multi-objective optimisation and uncertainties
environmental, and social sustainability.
Currently, RE planning is mostly done through an ad-hoc deci-
1.1. Renewable based energy systems for buildings sion process without considering all the possibilities. . Most studies
conducted on developing RE systems have taken a scenario-based
Moving towards NZEB has been mandated by Energy Step Codes approach in energy planning [4,25–27]. Other studies have taken a
in different parts of the world [15]. While net-zero energy devel- single objective optimisation approach in energy planning [28,29].
opment has been mandated through various regulatory provisions, In scenario-based planning, the decision makers define a limited
the guidance on achieving these is limited. So far, there is a lack number of energy scenarios by selecting the RES, and defining
of decision tools and guidelines that treat energy system planning the mix based on their judgement [4]. The best alternative among
as a holistic, practical, and systematic process and eliminate the these scenarios is then selected through further analysis, consider-
requirement for specialised expertise from the developers and mu- ing different selection criteria such as economic performance and
nicipal decision makers [16]. There are multiple challenges present environmental impacts [26,27]. This method does not present a
in using RE technologies (RET) at building level. While the use of rational and evidence-based method for considering all available
renewable sources in energy systems is expected to reduce the as- technological options and selecting the most suitable RET and
sociated emissions of energy generation, it is necessary to assess cannot consider all possible variations in system planning, and
the overall life cycle costs and impacts of replacing conventional thus leads to many uncertainties. The “in-between” scenarios
energy sources with renewables [17]. The need for specialized in- are neglected, and it is not possible to identify the exact sizing
frastructure and the high capital and operational costs to accom- and capacity that will lead to the best results. Therefore, there is
modate RET are only a part of the problem. RE resources such as limited potential in identifying the truly optimal solutions. Energy
solar and wind are intermittent in nature, and therefore, the en- system planning needs to be conducted in a more systematic
ergy supply from these is subject to fluctuations [18]. manner identify the truly optimal solutions. For this, planning an
The heating load dominates the residential energy use in colder optimal energy system should be considered as a multi-objective
climates, and buildings located in such climates generally have decision making (MODM) problem. In MODM, conflicting crite-
high energy intensities [19]. Canada is a typical representation of ria are optimised to deliver the best solution to a given design
72 H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89

Table 1
Objectives of energy system optimisation.

Technical Economic Environmental & other

Minimise − Cost/efficiency ratio − Total cost of the system − Environmental impact


− Dumped energy − Capital cost / investment − Energy price volatility
− Total annual cost − Economic penalties due to insufficient energy
− Energy generation cost generation through RE
− Equipment cost
− Levelised cost of energy
− O&M costs
− Life cycle cost of system (LCC)
Maximise − Electricity production − Revenue of system − Local economic development
− Thermal energy production − Total benefits for a fixed investment − Job creation
− Energy supply − Benefit/cost ratio
− Power quality − Project lifetime economic return (PLER)
− Reliability of power supply

problem [30]. Energy system optimisation involves sizing and study is to investigate the feasibility of developing net-zero energy
developing the ideal “mix” of energy sources in the system buildings. While the demonstration is in the Canadian context,
[31–33]. This multi-objective optimisation (MOO) occurs under a the findings have global applicability in converting the residential
number of technical, economic, and environmental objectives and building stock to net-zero status through the use of RE. Further-
constraints. Optimisation techniques use objective functions to more, it addresses pre-project planning stage of community-level
define particular criteria of the system that are to be minimised or energy systems, to overcome the limitations of the previous
maximised [32]. Some of the factors defined as objective functions studies, and to extend renewable energy technology selection
in previous studies when optimising energy systems are listed in from an academic exercise to a practical engineering solution for
Table 1 [32–36]. While some studies have applied multi-criteria community-level decision makers through an uncertainty-based
decision making approaches to RE planning [37,38], more attention combinatorial optimisation approach. These findings will inform
needs to be paid to the identification of all relevant factors in and guide community developers and other stakeholders with an
devising community-level energy systems. When applying the interest in residential buildings on the most suitable clean energy
same concept to building-level energy systems, the economic options for building projects during the pre-project planning stage.
concerns of affordability are weighted higher than other aspects, The decision support framework developed in this study can be
as this factor directly affects stakeholders at all levels. used to plan the building energy supply based on the decision
Decision making pertaining to energy systems should consider makers’ priorities. Moreover, the findings can inform policy makers
uncertainties to make the project planning more relevant and prac- in developing building codes and energy codes targeted towards
tical. Data uncertainty is a critical issue faced in RE planning. The achieving net-zero buildings.
information obtained on environmental and socio-economic fac- The next sections of the paper describe the work carried out in
tors, locational parameters such as geography and climate as well developing the optimisation model and analysing the case study
as demand and price forecasts are usually vague, incomplete or for the selected site location. In the first part of the methodol-
imprecise [39]. Data uncertainty can be addressed through tech- ogy, the building energy use and the renewable energy potential in
niques such as Monte Carlo simulation and fuzzy based techniques the local area to match the above demand are analysed. The sec-
[40,41]. In addition to data uncertainties, stochastic uncertainties ond part of the methodology details the optimisation algorithm,
due to system and environmental variations can affect energy sys- impact assessment, and uncertainty analysis. The system perfor-
tem planning [42]. Fuzzy logic has the advantage of being able to mance evaluation and case study analysis is presented following
represent qualitative information and the vague or imprecise na- the methodology. The next section presents the optimisation re-
ture of human decision making, and has been commonly used in sults as well as the economic and environmental outcomes of the
energy planning problems [43]. proposed energy system, and these findings are explained in de-
The current energy planning tools is that they take a discretized tails in the discussion section that follows. Finally, the conclusions
approach to the different activities and aspects of energy planning, and recommendation on implementing optimal hybrid building en-
without having a holistic and practical vision. Moreover, the exist- ergy systems are given based on the study outcomes.
ing tools and studies neglect life cycle thinking and uncertainty in
the decision making for most part. Energy planning needs to be 2. Methodology
tackled as a multi-objective, multi-stakeholder, and multi-period
problem during the project planning stage of community develop- This this study, a framework was developed to identify the op-
ment and building construction project. timal energy mix and fraction of energy which can be supplied
The objective of this study is to develop a model to identify through RE in residential buildings and in small-scale residential
the optimal renewable energy mix while also accounting for communities. Therefore, the methodology is presented in two main
uncertainties, which can be integrated at building level with life sections for the building energy system development, and follow-
cycle thinking. The study aims to investigate the feasibility of net- ing that, the community level energy facility development.
zero building energy systems under different decision priorities, Building energy use was considered under the residential end
considering triple bottom line (TBL) impacts of energy use. The use categories. Different renewable energy technologies (RET)
life cycle economic and environmental impacts of energy use are which have the potential to be used at building level were identi-
considered for both conventional and renewable energy resources, fied, and the life cycle environmental impacts and costs pertaining
and the optimal energy mix for buildings are identified based on to these RET were evaluated. Energy supply mix variations were
this and the resource availability. A fuzzy logic-based approach is defined as combinations, and the overall impacts of each supply
taken to integrate the data uncertainties and potential variations in combination were assessed. The following sections detail how the
the current conditions. A case study analysis was conducted for a modelling was conducted for building energy demand and RE tech-
representative building in British Columbia, Canada. A goal of this nologies, and how the optimisation algorithm was developed to
H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89 73

Fig. 1. Overall methodology for building energy system optimisation.

identify the most suitable energy mix to meet the demand. The technologies, in order to assess the performance of different energy
overall study methodology is depicted in Fig. 1. Initially, data was system combinations. A fuzzy ranking method was used to sort the
collected on RE technologies and their energy generation poten- energy system combinations based on their overall performance, in
tial. A case study location was selected to apply and demonstrate order to identify the best energy supply mix for the community.
the developed framework. The energy technologies were further Detailed descriptions on the methods used are presented in the
analysed in order to select the most suitable options for the se- following sections under the Methodology. (A list of symbols used
lected given community. Building energy databases were analysed in the methodological equations are presented under Appendix I.)
to identify the average energy end uses, the end use intensities by The uncertainties and variable data regarding energy use, sup-
residential floor space, and monthly energy use variations across ply, and cost factors were represented in the form of triangular
different seasons. Similarly, seasonal variations of RE generation fuzzy numbers (TFN). For a triangular fuzzy number, x, the mem-
potential were also assessed. The above data was used as inputs bership function is represented by Eq. (1).
to the optimisation model, which was developed through a com- 
binatorial approach. The life cycle economic, environmental, and 0; xa, xc
human health impacts were evaluated for each of the selected RE μx˜ (x ) = x − a/b − a; a≤x<b (1)
c − x/c − b; b≤x<c
74 H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89

Here, a, b, and c are real numbers, and a < b < c. The notations of a triangular fuzzy number in the form of (a, b, c). EImi repre-
a and c represent the lower and upper bounds respectively. The sents the range of energy intensity values from all buildings in the
membership for the most likely value b, is denoted by μx˜ (b) = 1. cluster.
   
a P10 (E Imi )
2.1. Building energy use E Ib = b = P50 (E Imi ) (4)
c P90 (E Imi )
Both energy demand and availability of renewable energy sup- An urban developer using the proposed model to plan com-
ply in Canada change with seasonal variations. This variation needs munity energy systems can predict the energy performance of a
to be accounted for when identifying the best renewable en- proposed building cluster using the following method based on
ergy systems at building or community level. While the developed the average regional building data. Based on the median values in
model is applicable for any type of residential building, Multi-unit the building cluster information, the representative building can be
Residential Buildings (MURB) were selected as the focus of analy- defined considering the parameters of median number of storeys,
sis. A MURB is defined as multiple housing units contained within median gross floor area, and median ground floor area.
one building or several buildings, and the share of MURBs in the To define an alternative energy supply model for a residential
residential housing stock has grown with the increase in urban building, it is necessary to separate the heat and electric end uses,
population and the need for urban densification [44]. In Canada’s as the potential supply sources can vary based on the type of end
largest metropolitan areas, more than 50% of the newly planned use. The baseline load was accordingly split into the fractions of
residential construction is expected to be in the form of MURBs water heating and other electric load (appliances, lighting, and space
[44]. Due to the above reasons, it was identified that any planned cooling). An assumption was made that all appliances, lights, and
clean energy initiatives for the building sector need to especially cooling systems are powered by electricity.
focus on MURBs. The monthly space-heating load (SHImj ) was evaluated by sub-
The energy use in any residential building can be modelled con- tracting the baseline load of each month, and following the same
sidering the five main end uses. Out of them, the space and wa- procedure using the 10th percentile, the median, and the 90th per-
ter heating energy can be provided with either heat or electricity centile of the value to define the fuzzy number.
sources, while lighting and appliance energy can only be supplied
with electricity sources. Even though space cooling is generally E ISHm j = E Im − E Ib (5)
done via electricity supply, geothermal systems can also be used
   
as an alternative space cooling mechanism [45]. In hybrid building a P10 (E Im − E Ib )
energy systems, multiple energy sources (both conventional and SH Im j = b = P50 (E Im − E Ib ) (6)
renewable) are used to supply the energy demand of the build- c P90 (E Im − E Ib )
ing [27,46]. In Canada, similar to other colder climate with sea-
sonal variations, significant energy use variations can be observed SHB = SH Im j × AB (7)
during different times of the year. As space heating accounts for
the largest fraction of the total energy consumption in a residen-
BLEB = E Ib × AB (8)
tial building, the lowest energy consumption occurs in summer
months. In order to identify the behaviour of end use energy de-
mand on a monthly basis, the assumption of a non-variable base- W HB = FW H × BLEB (9)
line energy load is assumed [47]. A study conducted for BC, Canada
notes that space heating is either turned off or dormant during
OEB = BLEB − W HB (10)
mid-summer (July and August). The non-variable component of the
energy includes non-direct heating (e.g. domestic hot water), appli- AB = Building gross floor area (m2 )
ances, and lighting [47]. (This may be subject to minor variations SHB = Monthly space heating load of the building (kWh)
due to space cooling. However, as space cooling accounts for less BLEB = Monthly baseline energy load of the building (kWh)
than 1% of the energy demand in MURBs, this was assumed to be FW H = Fraction for water heating from the baseline load
negligible [21]. By removing the baseline loads from the total en- W HB = Monthly water heating load of the building (kWh)
ergy demand, it is possible to analyse the monthly heating loads. OEB = Monthly other electric load of the building (kWh)
Out of these monthly energy loads, The baseline energy de-
mand (Emb ) in a building was assumed to be the lowest monthly The above defined energy loads and building characteristics
energy demand, which occurs in July or August [47]. (Monthly en- were used for developing the energy model for an average MURB,
ergy demand is denoted by Emj .) to explore the feasibility of integrating building-level RE systems
  and achieving net-zero status for MURB.
Emb = min Em j ; f or 1 ≤ j ≤ 12 (2)
2.2. Renewable energy resources and technologies
The monthly energy intensity (EIm ) of a building, a quantity
which represents the energy use per unit area, was estimated us- The first step in developing a renewable based hybrid energy
ing the gross floor area (Af ). The energy intensity provides a mea- system for a MURB is to select the most suitable technologies. So-
sure of comparison amongst buildings of different sizes. lar technologies are widely used at building level. Wind resource
Em j availability for buildings is limited due to structural, noise, and
E Im = ; f or 1 ≤ j ≤ 12, (3) wind pattern considerations. Therefore wind power generators are
Af
not usually installed on buildings, especially as the resource avail-
The minimum monthly energy intensity is referred to as the ability is highly site specific [12]. Similarly, the biomass option
baseline energy intensity (EIb ) in each building. For the analysed was eliminated due to the challenges in a residential context re-
building cluster, a triangular fuzzy number (TFN) was defined for lated to ensuring a reliable and constant supply, and the need
the monthly baseline energy intensity using the 10th percentile, for additional transportation of supplies. In addition, building level
the median, and the 90th percentile of the EI value range respec- geothermal systems can be used to supply the heat demand in res-
tively as the lower bound, most likely value, and the upper bound idences [48].
H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89 75

2.2.1. Solar energy The performance ratio of a PV system can vary based on a num-
The solar energy potential of a residential building is limited ber of factors. Natural resources Canada has identified the most
by the rooftop and/or façade area available in the building [49]. In likely value as 0.75, and that the PR of most PV systems fall within
the current study, only rooftop area was considered for modelling a 15% range of this value [52]. This aligns with the findings of the
solar energy potential. The available roof area is utilised by solar National Renewable Energy Laboratory of U.S. Department of En-
PV panels, solar thermal collectors, or both. ergy, which states that PR usually varies between 0.6 and 0.8 [51].
Based on the above, a TFN was defined for the performance ratio
Asolar = APV + AST (11) as follows.
Asolar = Solar installed roof area (m2 ); P R = (0.6375, 0.75, 0.8625)
APV = PV installed roof area (m2 );
The electricity produced from PV system is primarily directed
AST = Solar thermal installed roof area (m2 )
towards fulfilling the other electric loads in the house, excluding
space and water heating.
A simple rule of thumb has been defined by the International
The equation for number of daylight hours (Nd ) is;
Energy Agency on the fraction of rooftop area that is utilisable
for solar installations, in terms of sufficient annual solar insolation 2
Nd = cos−1 (− tan φ tan δ ) (18)
and architectural suitability [49,50]. The rule can be summed up 15
as the solar utilisation factors for roofs and facades are 0.4 and where φ is the latitude, and δ is the declination angle on nth day
0.15 respectively, i.e. for every 1 m2 of building ground floor area of the year;
(AGF ), there is 0.4 m2 of solar suitable rooftop area [49]. This rule  284 + n

was used in defining the constraints for the building solar energy δ = 23.45 × sin 360
model. The area required for installing 1 kW of PV capacity1 is 365
6 m2 . The equation for the thermal energy output in ST collectors is
given below [49]. Since solar thermal water heating cannot be car-
AGF = AB/N (12)
B ried out on demand, and the heat energy generation can only be
carried out during a limited time period during the day, the effi-
Asolar, max = UF × AGF (13) ciency of heat energy storage has to be considered. An assumption
was made that the heat retention in the hot water storage is 50%
efficient, and only half the energy produced by the collectors is ef-
AST, avail = Asolar, − APV (14) fectively transferred to meet the water heating load [53]. The ISO-
max
efficiency equation is reported based on gross collector area and
not the net aperture area of the collector as per ASHRAE standards
APV = APV, unit × CPV (15) [54]. Therefore, the thermal energy output is calculated based on
the gross area, but the thermal energy output does not change due
to the above simplification [54].
AST = AST, unit × nST (16) EST = 50% o f EST,theoretical = 0.5 × A ×η×H (19)
ST,G

NB = Number of storeys of the building


UF = Roof utilisation factor η × AST,G = ηAP × A ST,A (20)
Asolar, max = Maximum rooftop area available for for solar instal-
lations (m2 ) ηAP = Efficiency in terms of net aperture area
APV, unit = Area for installing 1 kW of PV (m2 ) η = Efficiency in terms of gross collector area
AST, unit = Area for installing 1 solar thermal collector (m2 ) AST, G = Gross collector area
AST, avail = Total area available for installing solar thermal collec- AST, A = Net aperture area
tors (m2 )
In the developed optimisation model, the energy supplied from
CPV = Installed solar PV capacity (kW)
the solar thermal collectors was prioritised for catering the water
nST = Number of installed solar thermal collectors
heating demand in the residence.
The performance ratio (PR) of a system is a dimensionless pa-
2.2.2. Geothermal energy
rameter that represents the overall effect of system losses (due to
Ground source heat pumps (GSHP) are effective for heating pur-
inverter losses, wiring, mismatches, other power losses, tempera-
poses in cold climates [55]. These are more efficient than com-
ture, reflection, soiling or snow, system down time, and component
monly used air source heat pumps [56], and therefore, were se-
failures) on the rated output [51]. PR does not depend on module
lected as a suitable residential thermal energy source for the se-
efficiency (i.e. efficiency variation based on panel type) as it is de-
lected building location [48]. For urban or suburban areas which
fined based on the nominal (rated) power of the PV system. The
do not have access to a water body or a well, a vertical closed loop
following equation defines the energy generation of a PV system.
geothermal heat system is more suitable due to space restrictions
[48]. Therefore, this type of system was selected for the case study.
EPV = H × N × P R × APV , (17) For one ton of heat pump capacity (approximately equivalent
to 3.5 kW or 12,0 0 0 Btu/h), the length of piping required is ap-
EPV = Monthly energy generation of the PV system (kWh)
proximately 80–110 m [48]. When selecting the suitable heat pump
H = Monthly mean daily insolation in the plane of the PV array
capacity for the MURB, it was assumed to be constrained by the
(kWh/m2 )
maximum design power for space heating. This maximum design
N = Number of days in the month
power occurs in the month with the highest heating load. The
equations for heat energy generation from GSHP and design power
1
Based on the data published by Canadian Solar Inc. on their 275 W PV module, for space heating can be calculated from the equation given be-
the dimensions are 1650 × 992 mm, leading to 5.952 m2 [59]. low. In cold climates, it is recommended to size the pump to suit
76 H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89

the heating load [57]. A flat space heat load curve is assumed in economic and environmental impacts. This type of combinatorial
this analysis to reduce the computational complexity and optimis- optimisation approach reduces the computational time compared
ing time. to a continuous optimisation process. The use of a fuzzy-logic
based approach in the optimisation model and the resulting
EGSHP = CGSHP × Td × N (21)
decision framework helps to integrate the effect of uncertain and
variable data inputs into the decision making [40]. In this model,
SHB, max fuzzy numbers are used to represent data which do not have de-
PSH, max = = CGSHP, max (22)
Td × N terministic of crisp values or where the possible values vary within
a range, and similar approaches have been used in previous studies
EGSHP = Monthly energy generation from GSHP (kWh) to solve engineering problems with uncertain inputs [40,60].
CGSHP = Installed capacity of GSHP (kW)
Td = Hours operated per day
N = Number of days in the month 2.3.1. Performance objectives and weighting for building energy
PSH, max = Maximum design power for space heating (kW) model
Performance objectives were selected to represent the energy
The energy supplied from the GSHP system was prioritised for (EN), economic (EC), and environmental (EV) aspects related to
meeting the space heating load in the MURB. Space cooling aspect building energy systems. The economic performance objectives
of GSHP systems were neglected to the minimal impact in colder were to minimise the total energy system cost for the entire build-
climates. ing life cycle, and to maximise the operational energy cost savings.
As the ultimate target of renewable integration at building level is
2.3. Identifying the optimal building energy mix to develop net-zero energy buildings, maximising the RE fraction
in building energy supply was selected to be the energy perfor-
As the goal of the study is to identify the best energy supply mance objective. As RE integration in decentralised energy systems
mix and assess the feasibility of net-zero energy buildings, it was leads not only to reduced emissions but also to improved energy
necessary to evaluate the performance of the building energy sys- independence and energy security for communities [42], it is nec-
tem with a mixed supply from the renewable and conventional en- essary to maximise the RE fraction in the supply.
ergy sources. A discrete combinatorial optimisation approach was Reduction of operational emissions is commonly used as an ob-
adopted in evaluating the building energy system and identify- jective in energy system optimisations. However, it is necessary
ing the optimal energy mix. Combinatorial optimisation method is evaluate all possible environmental impacts (which occur in addi-
most suited to practical problems where activities and resources tion to emissions) with a life cycle perspective, in order to holisti-
(such as machines and number of trips) are indivisible [58]. The cally assess the energy system’s performance. Therefore, the envi-
reason for adopting the above approach in the current study is ronmental performance objective was to reduce the life cycle im-
to provide a practically feasible engineering solution to the en- pacts of the energy system. Life cycle impacts were evaluated un-
ergy technology selection and system planning problem during the der the impact categories of climate change, damage to ecosys-
pre-project planning stage of residential development. Assuming a tems, resource depletion, and human health risk [42].
continuous range of values for the energy technology system ca- The solar available roof area acts as a constraint to the solar
pacities is not a sensible approach in tackling practical engineer- energy installations as detailed in Section 2.2. As the goal of this
ing problems, as energy installations sizing is done with discrete study is to develop a framework for defining the best renewable
and indivisible capacity sizes based on standard system compo- based energy mix for buildings, it is important to consider the local
nents and equipment available in the market. conditions and user priorities in the decision making. To integrate
Solar PV installed capacity was incremented by steps of this aspect, the maximum levelised cost of energy (LCOE) and the
0.275 kW, to reflect the standard capacity of a commercial solar minimum fraction of RE in the energy supply were defined as flex-
panel available in the market [59]. For a commercially available ible constraints which can be changes based on user preferences.
solar thermal collector, capacity incrementing was done based on Levelised cost of energy is defined as the ratio of lifetime costs
the number of collectors. The installed capacity of GHSP was in- to the lifetime energy generation, is an important measure in es-
cremented by 1 ton (3.5 kW) steps [48]. Based on the information tablishing grid parity (whether a unit of RE can be provided at a
in RETScreen product database, a 3.37 kW commercially available comparable price to that of a conventional centralised energy sup-
ground source heat pump (heating COP of 3.1) has a cooling COP ply unit, i.e. unit of grid electricity) [42,61]. As the energy gen-
of 4.25, and a cooling capacity of 4.78 kW, and this capacity was erated at building level should be affordable, a constraint was de-
approximated in calculating the heating load supply.2 fined to represent the closeness to grid parity. Under this, the LCOE
The remainder of the space heating load (which cannot be of the developed energy system model should not exceed 1.2 times
catered by the GSHP) is supplied from grid electricity. While water the price of a grid electricity unit.
heating energy is primarily supplied from solar thermal collectors, n It +Mt +Ft

any remaining water heating energy requirement is supplied from LCOE = t=1 n (23)
(1+r )t Et
t=1 (1+r )t
electricity. Grid electricity is also used in fulfilling the remaining
electricity demand, if the PV electricity supply is insufficient to
It = Investment expenditure for year t;
meet the full electric load of the MURB.
Mt = O&M expenditure for year t;
With the above incrementing process, different energy mix
Ft = Fuel expenditure for year t;
combinations are defined for the building energy supply system.
Et = Electricity generated in year t;
The developed optimisation framework simulates all the possible
r = Discount rate;
combinations of alternative capacities for the identified energy
n = Facility lifetime (service life)
supply technologies, with the goal of supplying the total energy
demand of the building. The performance of each combination is
Another aspect in integrating user priorities and requirements
simulated based on a number of objective functions representing
to a decision making model is weighting. Weights represent the
relative importance of each performance objective. In the devel-
2
Addison DWPG017 model. oped decision model for the optimal energy mix, a standard fuzzy
H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89 77

Fig. 2. Life cycle impact assessment approach.

weighting scheme was used to represent the user-defined impor- environmental impacts in a quantitative form, and are more accu-
tance of the performance objectives, based on previous literature rate than endpoint results [63]. Thus, midpoint indicators are sub-
[62]. ject to less uncertainties in comparison to endpoint indicators that
As energy system planning is a multi-stakeholder problem, the are derived through much higher environmental mechanism mod-
performance objectives need to be weighed with reference to the elling [64]. The study was conducted using the SimaPro software,
interests of all parties in the given context. As an example, while a and the Ecoinvent database. Fuzzy logic is a method that is rec-
MURB involves property developers, building management, and oc- ommended for processing uncertainties in LCA [65]. In the current
cupants as key stakeholders, house owners are the primary stake- study, a 10% tolerance for data variability was assumed in defin-
holders in single-family detached housing. Therefore, the weights ing the upper and lower boundary values when converting the ob-
defined for the performance objectives were case-specific. tained life cycle impact assessment (LCIA) data into fuzzy numbers
[42]. The impact data from all categories were aggregated to four
2.3.2. Impact and outcome assessment for performance objectives main environmental impact indicators, as depicted in Fig. 2. Three
The life cycle environmental and economic impacts of the dif- indicators were defined based on the general midpoint to endpoint
ferent energy technologies (i.e. solar PV, solar thermal collectors, aggregation process in ReCiPe, namely damage to ecosystems, hu-
GSHP) were evaluated in order to quantify the total impacts of the man health impacts, and resource depletion [66]. The other indicator
developed hybrid energy system for the building [42]. A summary was global warming potential, which represents the carbon emis-
of the life cycle impact assessment method is provided in Fig. 2. sions related climate impact of the energy system. The following
The life cycle environmental impacts for different RET were equation represents the division by sum approach adopted in nor-
quantified using a life cycle assessment (LCA) approach. The sys- malising the impacts for aggregation [67].
tem boundary defined for the RET evaluation extended from the
raw material extraction to the end of life of the RE generation sys- I Mi j
Ni j = (24)
tems. A functional unit of 1 MWh of produced energy was used on j I Mi j
conducting the LCA, which was done under the ReCiPe midpoint
assessment method. Midpoint assessment method was selected to
where;
conduct the LCA as it provides information on emissions and other
78 H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89

Table 2
Renewable energy technology characteristics.

Technology Cost factor Unit L M H Std. Dev

Solar PV Installed cost $/kW 3008 3897 4786 ±889


Fixed O&M cost $/kW-yr 1 21 41 ±20
Fuel and/or water cost $/kWh 0 0 0 -
Lifetime Year 22 33 44 ±11
Solar thermal Installed cost $/ft2 71 162 253 ±91
Fixed O&M cost $/ft2 -yr 0.355 1.215 2.53 N/Aa
Fuel and/or water cost $/ton 0 0 0 -
Lifetime Year 17 31 45 ±14
GSHP Installed cost $/ton 3133 7765 12,397 ±4632
Fixed O&M cost $/ton-yr 15 109 203 ±94
Fuel and/or water cost $/ton 5 397 789 ±392
Lifetime Year 13 38 63 ±25
a
O&M cost is 0.5 to 1.0 % of the initial installed cost [70].

 
IMij = Impact value of the ith impact category for the jth RE  
technology − CSC + RF C (28)
t t
Nij = Normalised value for the indicator
j = number of compared items (RE technologies) SHB = Monthly space heating load of the building (kWh)
W HB = Monthly water heating load of the building (kWh)
Climate change effect of energy supply is a critical aspect which
OEB = Monthly other electric load of the building (kWh)
needs to be considered in energy system design, especially in
BLEB = Monthly baseline energy load of the building (kWh)
light of the Canadian climate action goals [68,69]. Therefore, global
AOS = Annual operational energy cost savings
warming potential was assigned a weight of 50% when aggregating
GEP = Grid electricity price
the impact indicators to evaluate the ultimate performance under
NGP = Natural gas price
life cycle environmental impact objective, and the other categories
CSC = Conventional supply costs
were assigned equal weights [42].
RF C = Renewable fuel costs
Eq. (25) was used in calculating the total energy system costs
over the building lifetime, considering all energy supply sources in Fraction of RE in the building energy supply (FRE) is calculated
the building energy system. The next two equations represent the as follows on an annual basis.
present value of a one-time and a recurring cash flow respectively 12
[40]. Each cost variable was defined in the form of a TFN, based (EPV + EST + EGSHP )
j=1
F RE = 12 (29)
j=1 (SHB + BLEB )
on literature-based cost data. The building lifetime and discount
rate were assumed to be crisp numbers, where building life was
50 years. Table 2 details the system life and cost factors associated with
      solar PV, ST collectors, and GSHP. The cost values were obtained
T EC = IC + FC + AOM + RC + DC − RV (25) from the distributed renewable energy generation costs published
t t t t t t by National Renewable Energy Laboratory [70]. The residual values
FV and salvage costs were assumed to be zero at the end of useful life.
PV = (26) The upper and lower limits fuzzy numbers were defined based on
(1 + r )t
the standard deviation of the cost values in the above database.
 
A (1 + r )t − 1 When developing the cost model for the total energy system
PVA = (27) cost throughout lifetime of the building, the system replacements
r (1 + r )t were considered based on the life expectancy of each technology.
T EC = Total discounted energy system cost
IC = Investment cost 2.3.3. Fuzzy-based building energy supply optimisation approach
F C = Fuel and conventional energy supply costs Algebraic computations for triangular fuzzy numbers are con-
RC = Rehabilitation and/or replacement cost ducted through fuzzy arithmetic principles, which are represented
AOM = Annual operations and maintenance cost in the equations given below [42]. A and B TFNs where A = (a1 , a2 ,
DC = Disposal cost a3 ) and B = (b1 , b2 , b3 ), assuming all a1 , a2 , a3 and b1 , b2 , b3 > 0;
RV = Residual value
PV = Present value of a future one time cash flow (FV) occuring Addition: A+B = (a1 +b1 , a2 +b2 , a3 +b3 )
at time t Subtraction: A-B = (a1 -b3 , a2 -b2 , a3 -b1 )
PVA = Present value of a future annually recurring cash flow (A) Multiplication: A(.)B = (a1 b1 , a2 b2 , a3 b3 )
over t number of years Division: A÷B = (a1 /b3 , a2 /b2 , a3 /b1 )
r = Annual discount factor Scalar multiplication: kA = (ka1 , ka2 , ka3 ); if k > 0
kA = (ka3 , ka2 , ka1 ); if k<0
The following equation was defined to calculate the total oper-
ational energy cost savings. (Conventional supply costs are calcu- For the optimisation, the performance scores under the selected
lated for the leftover building energy load which cannot be sup- objectives are evaluated for all energy supply combinations (so-
plied with RE under a given energy supply combination.) lutions). Then, the aggregate performance score for a particular
  solution is evaluated in the following manner. The performance
  scores under individual objectives such as TEC and AOS are nor-
AOS = [SHB + OEB ] × GEP + [W HB ] × NGP
malised, and then the normalised values are multiplied by the
t t
H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89 79

Table 3
Fuzzy weighting scheme of relative importance.

Linguistic Term Very Low (VL) Low (L) Medium (M) High (H) Very High (VH)

Membership function (0,0,0.3) (0,0.3,0.5) (0.2,0.5,0.8) (0.5,0.7,1) (0.7,1,1)


fuzzy weights defined in Table 3, based on the assigned relative where xmin = in f S; xmax = supS; S = m i=1 Fi ; Fi = {x| fFi (x ) >
importance levels. 0}; i = 1, 2, . . . , m
A normalised fuzzy number (under the “divide by sum” ap- The right utility value UM (Fi ), left utility value UG (Fi ), and to-
proach) is defined by the following equations [71]. If the perfor- tal utility value UT (Fi ) for a given (i) combination are defined as
mance score for the ith option under the jth performance criterion follows.
is represented by a fuzzy number Xi , then the normalised fuzzy
 
decision matrix P = [ pi j ] is represented as follows [72]. UM (Fi ) = sup fFi (x ) f M ( x ) , i = 1, 2, . . . , m (35)
   
I f Xi is a f uzzy number denoted by xi j1 , xi j2 , xi j3 ,
    UG (Fi ) = sup fFi (x ) f G ( x ) , i = 1, 2, . . . , m (36)
and xmax = max xi j3 and xmin = min xi j1
UM (Fi ) + 1 − UG (Fi )
 xi j1 −xmin  UT (Fi ) = (37)
xi j2 −xmin xi j3 −xmin 2
xmax −xmin
, xmax −xmin
, xmax −xmin
; for benefit criteria
pi j =  xmax −xi j1 xmax −xi j2 xmax −xi j3  (30) The highest performing (optimal) energy system combination
N xmax −xmin
, xmax −x , xmax −x ; for cost criteria (solution) under the given conditions and constraints is the one
min min

with the highest utility value, and all viable combinations will be
If the weight assigned to the jth performance objective is wj ,
ranked according to their respective UT (Fi ).
The weighted normalised value of the ith solution under the jth
performance objective is represented by; 2.4. Energy system performance assessment
vi j = ri j (. )w j (31)
After identifying the best energy system combinations, their
The overall performance score (pi ) of a particular solution (en- performance scores under different objectives were defuzzified
ergy supply combination) considering m number of performance to analyse the system characteristics further, using the following
objectives can be represented by the following. Here, pi is a TFN, equation representing the centre of gravity (COG) defuzzication
and a higher pi denotes a better performance in the solution. method [74].
 

m
I f Xi is a f uzzy number denoted by xi j1 , xi j2 , xi j3
pi = wi j ∗ ri j (32)
j=1  
xi j1 + xi j2 + xi j3
The constraints defined in Section 2.3.1 (i.e. grid parity, roof COGXi = (38)
3
area, minimum RE fraction) are used to filter out the non-viable
The performance of the selected optimal solution will be eval-
solutions out of all the defined energy system combinations.
uated based on its potential to reduce emissions, total life cycle
The likely values for the LCOE and RE generation is used in set-
cost, and payback time of the system investment. The emissions re-
ting the filtering criteria based on grid parity and RE fraction con-
duction potential is calculated based on the grid emissions factor,
straints. The likely value was established based on mean of max-
which is 10.67 tCO2 e/GWh in British Columbia [75]. The follow-
imum (MoM) method, where the defuzzified value is represented
ing equation is used in calculating the discounted payback period
by (a+b)/2 [40]. For a TFN, the mean of maximum corresponds to
(DPP) of the energy system.
the likely value based on the above equation. 
In order to identify the optimal solution out of the remain- 1
ing viable combinations, the TFNs for pi need to be ranked. One
DP P = ln 1− IC×r
AOS ln(1 + r ) (39)
method which has been widely used in literature for ranking fuzzy
The economic performance of the system was further analysed
numbers under similar contexts is the maximising set and min-
using the fuzzy DSW algorithm to aggregate the cost and rev-
imising set method [62,73]. This method can be used to select
enue factors. DSW algorithm applies fuzzy extension principle in
the most suitable solution out of the energy system combinations
a simplified manner for calculations involving fuzzy variables and
using the fuzzy appropriateness index (Fi ; i = 1, 2, . . . ..n), which
functions [40]. Under this technique, α -cut intervals are used in
represents the overall performance score. The TFN for Fi is Fi =
standard interval analysis to calculate complex output membership
[ai , bi , ci ]. The following equations define the maximising set (M)
functions made up of multiple variables (which are represented as
and minimising set (G) for the range of alternative combinations
fuzzy numbers). The following steps are used in the calculation.
[62]. (Here, fM (x) is the membership of x to M, and fG (x) is the
membership of x to G.) a. An “α ” value within the range 0 ≤ α ≤ 1 is selected.
b. The intervals that correspond to the selected α -cut level (in
M = {(x, fM (x ) )|x ∈ R}
the input membership functions) are found.
 x−xmin c. The output membership function’s interval for LCC is deter-
xmax −xmin
; xmin ≤ x ≤ xmax
f M (x ) = (33) mined using standard binary interval operations at the se-
0; otherwise lected α -cut level.
d. The above steps are repeated for different α -cut levels be-
G = {(x, fG (x ) )|x ∈ R} tween (0,1) membership
 x−xmax
xmin −xmax
; xmin ≤ x ≤ xmax In the results, the top-ranked energy system combinations were
f G (x ) = (34) analysed to assess their economic and environmental performance
0; otherwise compared to a conventional energy system.
80 H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89

Table 4
Energy end uses of the residential sector in BC.

Residence type Single family detached Single family attached Apartments (multi-unit)

Space heating 57.53% 42.90% 34.13%


Water heating 24.75% 35.89% 40.41%
Space cooling 0.89% 1.06% 0.64%
Lighting 4.77% 4.41% 3.26%
Appliances 12.06% 15.75% 21.56%

Fig. 3. Energy use intensity.

3. Case study of RE-based energy systems in BC. The alternative energy supply
model was analysed for this building to demonstrate the developed
As the developed framework is demonstrated for the average model. The energy intensities and end use fractions were derived
residential sector conditions in British Columbia (BC), Canada, the based on the median energy values of the building database on a
RE data was also compiled for a medium-scale municipality lo- monthly basis. It should be noted that the goal of this paper is to
cated in the Okanagan region of BC. To demonstrate the proposed present a customisable framework that is applicable to all residen-
building energy optimisation model, data was extracted from a re- tial buildings, and thus, if the inputs are adjusted to represent a
port published on the energy use in low-rise buildings to indicate specific building instead of a general case, the model outputs and
the average MURB energy demand in BC [76]. The location co- results will correspond to that particular building.
ordinates of the case study community were defined as 49.7711° Table 5 presents the fuzzy numbers for the monthly heating
N (latitude), 119.7275° W (longitude) to represent the District of load intensities of the case study building. (Here, a, b, and c re-
Peachland, a municipality in the Central Okanagan Regional district fer to the lower bound, most likely value, and the upper bound of
in BC. a triangular fuzzy number, as mentioned in Section 2.1.)
The main end use shares of the residential sector in British Table 6 presents the relevant mean daily insolation values for
Columbia (BC), Canada, are listed in Table 4 based on 2015 Natural the selected location under for different panel orientations, based
Resources Canada data [21]. on the photovoltaic and solar resource maps published by Natural
The energy intensity variation for the total building energy use Resources Canada [77]. For the purpose of the case study, the array
and baseline load of the selected building cluster is represented was assumed to be south facing with a tilt equalling latitude.
in Fig. 3 [76]. Here, the energy use above the baseline load is at- Liquid glazed type solar thermal collector is the most com-
tributed to space heating. The average energy end use ratios for monly used in Canada [78]. Therefore, this type of collector was se-
the apartment buildings (MURB) mentioned under Table 4 were lected for the case study. As the energy generation potential of so-
used in defining the split between water heating and other electric lar thermal collectors are dependent on the collector type, a glazed
loads. Accordingly, out of the baseline load, 61.35% is consumed for flat plate collector3 was selected from the ratings published by the
water heating, and the remaining 38.65% is for other electric loads. Solar Rating & Certification Corporation of the International Code
The fuzzy number defined for the MURB energy intensity by Council (ICC) [79]. The performance characteristics of the collector
floor areas as per Eq. (4) is given below. extracted from the collector datasheet are provided in Table 7 [80].
  (It was assumed that the panel orientation for PV applies similarly
kW h
m2 = (5.93, 8.70, 12.11)
EI to solar thermal for architectural purposes.) A, B, C, D, and E re-
fer to standard climate categories for solar collector operation as
For the MURB sector in BC, Canada, the representative MURB defined by manufacturer data sheets, and Climate category D was
was defined as a 4-storeyed building with a gross floor area of determined to be most applicable (space and water heating in a
7009 m2 and a ground floor area of 1752 m2 [76]. The defini-
tion of the representative MURB was based on the median values
of the building parameters obtained from a database of low-rise 3
OG-100 ICC-SRCCTM Certified Solar Collector #2012027A (Brand name: GOBI
MURB in British Columbia, to investigate the general applicability HT).
H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89 81

Table 5
Monthly space heating load intensity (kWh/m2 ).

Heating load (SHmj )

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

a 6.32 4.84 4.94 3.16 1.71 0.42 0.01 0.00 0.85 2.73 5.08 6.11
b 12.74 7.33 7.45 5.45 3.53 1.22 0.09 0.00 2.13 4.91 7.93 9.78
c 23.23 17.75 18.29 12.77 7.88 2.71 0.32 0.03 4.67 11.38 18.91 22.40

Table 6
Monthly mean daily insolation for the selected location.

Month Mean daily global insolation (kWh/m2 )

South-facing vertical South-facing South-facing South-facing tilt Two-axis Horizontal


(tilt=90°) tilt=latitude tilt = lat + 15° =lat − 15° sun-tracking (tilt = 0°)

January 1.8 1.8 1.89 1.62 2.11 1.03


February 2.74 2.97 3.02 2.76 3.54 1.92
March 3.63 4.5 4.37 4.39 5.76 3.34
April 3.47 5.11 4.68 5.29 7.23 4.79
May 3.15 5.28 4.62 5.69 8.06 5.66
June 3 5.44 4.63 5.98 8.97 6.17
July 3.21 5.75 4.94 6.28 9.49 6.48
August 3.57 5.67 5.07 5.96 8.57 5.51
September 4.06 5.38 5.12 5.33 7.33 4.01
October 3.5 4.01 4.01 3.77 5.02 2.42
November 2.17 2.23 2.32 2.01 2.65 1.15
December 1.57 1.53 1.62 1.36 1.79 0.81
Annual 2.99 4.14 3.86 4.21 5.89 3.62

Table 7
Solar thermal collector characteristics.

Kilowatt-hours (thermal) per panel per day)

Climate → High radiation Medium radiation Low radiation


Category (Ti –Ta) (6.3 kWh/m2 .day) (4.7 kWh/m2 .day) (3.1 kWh/m2 .day)

A (-5 ºC) 11.8 8.9 6.0


B (5 ºC) 11.2 8.3 5.3
C (20 ºC) 10.0 7.1 4.2
D (50 ºC) 7.6 4.8 2.1
E (80 ºC) 4.8 2.3 0.3

Collector gross area (AST, G ) = 2.503 m2


Collector aperture area (AST, A ) = 2.313 m2
Iso-efficiency equation for the collector (Based on gross area and (P = T i − Ta )

cool climate) to the selected building location. 4. Results

P  2

P Under the results of the life cycle impact assessment, it was
η = 0.763 − 2.40630 − 0.01420 identified that all three renewable energy options carry lower life
G G
cycle impacts per MWh of energy generation in comparison to grid
G = solar irradiance on the collector (W/m2 ); This is calculated electricity. While the BC grid carries lower life cycle impacts in
based on the irradiance data provided in Table 6. energy production due to the domination in hydropower, voltage
In the economic impact assessment, the discount rate was 3% transformations and transmission results in significant losses, lead-
(rate of inflation, f = 2%; interest rate, i = 5%; discount rate, r = i ing to a considerable increase in life cycle impacts per MWh of grid
– f = 3%), as these are aleatory rather than epistemic uncertain- electricity (Tables 9).
ties and thus cannot be reduced by fuzzy based modelling [81,82]. The present value of the lifetime costs of each RE technology
The total annual operational energy cost savings were calculated in for a unit installation is given below in Table 10.
comparison to the utility payments for a conventional energy sup- Based on the algorithm presented under the methodology, the
ply, i.e. the cost of using 100% grid electricity to cater the entire possible energy system combinations were simulated. After apply-
building energy demand. This is a reasonable assumption as elec- ing the constraints, 94,034 viable options were identified. The top-
tric heating is recommended over natural gas heating for new con- ranked combinations that were identified are provided below. The
structions in BC [83]. The price of grid electricity (CA$/kWh) was RE fraction was established by defuzzifying the fuzzy output using
defined as (0.042, 0.0936, 0.1872), and price of natural gas(CA$/GJ) the COG method (Table 11).
was defined as (5.4702, 6.0800, 6.6880) [40,42]. The maximum installation potential under the available rooftop
For the case study, it was defined that at least 40% of the en- area is 116.6 kW for solar PV, and 288 collectors for solar thermal.
ergy supply needed to be from RE sources. The rationale behind By analysing the top-ranked energy system combinations, it can
the importance weighting given in the case study for different per- be observed that while solar PV installations are maximised in the
formance objectives are described in Table 8. This logic can be top solutions, solar thermal installations are zero, thereby indicat-
used by decision makers who apply the proposed framework when ing that solutions with solar thermal energy are not optimal under
defining their own performance priorities. the given circumstances. The maximum potential for GSHP under
82 H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89

Table 8
Weights assigned to performance objectives for building energy system.

Performance objective Importance Rationale

EC1 Minimising the total life cycle Very high The total project costs represent the overall energy system related expenditure incurred by
energy system cost different stakeholders throughout the lifetime of the building. In a MURB, the total cost is a
significant factor in deciding whether the property developers and building management can
afford the implementation and upkeep of the energy system. A higher total system cost
means that developers have to incur a higher cost upfront, the management has to face high
facility maintenance during the operations, and residents are burdened with increased
housing prices. As this is a measure of the economic burden and financial feasibility of the
developed energy system, its importance was rated as VH.
EC2 Maximising the operational High The operational cost saving is most relevant to the occupants of a MURB, and it decides the
energy cost savings affordability of the energy supply for them. (In MURB context, the occupants will be
responsible for the energy bills, but not the major replacement activities of energy systems.)
Therefore, the importance of this performance objective was rated as H.
EN1 Maximising the RE fraction in Very low In most locations in BC, there is constant grid electricity and natural gas supply, and the grid
building energy supply mix is dominated by hydropower [81,84]. This means the emissions factor of the conventional
supply is lower. Therefore, maximising the RE fraction in the energy supply holds a lesser
importance compared to cost, emissions, and affordability. However, in remote/Northern
locations in Canada which lack grid connectivity, increasing the RE fraction to achieve
net-zero status is very important, in order to ensure energy independence and energy security
and reduce energy poverty in communities [42].
EV1 Minimising life cycle Medium A key goal of RE integration is to mitigate the negative environmental impacts associated with
environmental impacts conventional energy use [85]. While emissions and other related impacts are not as high in a
hydropower dominated location such as BC, other locations in Canada (e.g. Alberta) with high
grid emissions factors may find this to be a key goal in building energy planning [81]. For the
present case study, the importance of this objective was rated as M.

Table 9
Life cycle impacts of energy generation.

Per MWh of generated energy


Impact category Unit
PV ST GSHP GE

Climate change kg CO2 eq 7.82E+01 1.11E+01 2.44E+02 2.74E+02


Ozone depletion kg CFC-11 eq 1.67E-05 2.79E-06 1.58E-04 4.80E-05
Terrestrial acidification kg SO2 eq 4.88E-01 1.35E-01 1.10E+00 1.19E+00
Freshwater eutrophication kg P eq 7.18E-02 2.39E-02 1.18E-01 2.11E-01
Marine eutrophication kg N eq 3.57E-02 7.36E-03 4.78E-02 8.30E-02
Human toxicity kg 1,4-DB eq 1.18E+02 4.93E+01 9.19E+01 1.65E+02
Photochemical oxidant formation kg NMVOC 3.15E-01 5.13E-02 6.10E-01 9.45E-01
Particulate matter formation kg PM10 eq 2.51E-01 5.67E-02 7.96E-01 9.33E-01
Terrestrial ecotoxicity kg 1,4-DB eq 1.42E-01 4.95E-03 1.25E-02 2.85E-02
Freshwater ecotoxicity kg 1,4-DB eq 1.17E+01 9.13E-01 2.79E+00 6.19E+00
Marine ecotoxicity kg 1,4-DB eq 1.07E+01 9.59E-01 2.59E+00 5.59E+00
Ionising radiation kBq U235 eq 8.76E+00 1.05E+00 2.32E+01 3.56E+01
Agricultural land occupation m2 a 6.14E+00 1.31E+00 9.22E+00 5.96E+02
Urban land occupation m2 a 8.87E-01 2.19E-01 1.38E+00 3.93E+00
Natural land transformation m2 1.27E-02 5.31E-03 2.90E-02 5.27E-02
Water depletion m3 2.58E+00 2.30E-01 1.53E+00 6.81E+01
Metal depletion kg Fe eq 2.29E+01 1.58E+01 6.22E+00 9.64E+00
Fossil depletion kg oil eq 2.04E+01 2.77E+00 5.70E+01 7.94E+01

Table 10
Present value of RE system costs.

RE technology Unit of installed Present value of lifetime cost ($)


capacity
Low Likely High

Solar PV Per kW 3944 5769 7593


Solar thermal Per panel 2807 6769 11,141
Ground source heat pumps Per ton 4742 27,020 49,297

the current peak heating lead is 28 tons. In the top solutions, GSHP 4.1. Economic analysis of the optimal energy system combination
capacity varies from 17 to 20 tons. Therefore, the optimal GSHP in-
stallation capacity varies between 60–70% of the peak heating load The economic impacts of the top-ranked energy system solu-
of the building. The energy fraction that can be provided by RE tions were analysed based on the defuzzified values. The following
varies between 43–47% of the total energy demand. table lists the levelised cost of energy (LCOE), total energy system
The additional performance data for the optimal energy system cost for whole building life (TESC), annual operational cost savings
combination are presented below. This data is used in carrying out (AOS), initial investment cost (IC), simple payback period (SPP), and
extended analysis for the economic and environmental impacts of discounted payback period (DPP). From the discounted payback pe-
the proposed hybrid building energy system (Table 12). riod, it can be seen that all the top energy system solutions reach
Fig 4 depicts the most likely optimal mix of energy supply at their payback point before the replacement time for the RE as-
building level, based on the generation potential of each renewable sets, which is 33 years for solar PV and 38 years for GSHP systems
energy technology in the optimal energy system combination. (Table 13).
H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89 83

Table 11
Best energy system combinations.

Solution Solar PV Solar thermal GSHP UT (Fi ) RE fraction


(Ranked) kW # of collectors ton (defuzzified)

Combination 1 116.6 0 17 0.4865 44.15%


Combination 2 116.6 0 18 0.4864 45.43%
Combination 3 116.05 0 17 0.4863 44.10%
Combination 4 116.05 0 18 0.4862 45.37%
Combination 5 115.5 0 17 0.4862 44.04%
Combination 6 116.6 0 19 0.4861 46.70%
Combination 7 115.5 0 18 0.4860 45.31%
Combination 8 114.95 0 17 0.4859 43.98%
Combination 9 116.05 0 19 0.4858 46.64%
Combination 10 116.6 0 20 0.4857 47.97%
Combination 11 114.95 0 18 0.4857 45.25%
Combination 12 114.4 0 17 0.4856 43.92%
Combination 13 115.5 0 19 0.4856 46.58%
Combination 14 116.05 0 20 0.4856 47.91%
Combination 15 114.4 0 18 0.4855 45.19%
Combination 16 113.85 0 17 0.4855 43.86%
Combination 17 114.95 0 19 0.4855 46.52%
Combination 18 115.5 0 20 0.4854 47.85%
Combination 19 113.85 0 18 0.4854 45.13%
Combination 20 113.3 0 17 0.4853 43.80%

Fig. 4. Energy generation mix at building level.

Table 12 building life, under a linear depreciation for the optimal energy
Performance of the selected solution.
system combination, with the payback period marked on it.
Performance data Low Likely High The total life cycle cost of the energy system for the building
Energy generation Solar PV 112,443 132,286 152,129 life cycle was analysed in detail using the fuzzy DSW algorithm. In
(kWh) GSHP 253,418 337,107 401,173 the DSW algorithm, α -cuts were defined at 10% intervals between
Grid Electricity 199,229 684,833 1636,792 (0, 1) membership. The results are presented in Fig. 6.
GHG emissions Solar PV 7.50E+01 8.34E+01 9.17E+01 Based on the DSW algorithm results, the LCC ranges from pos-
(kgCO2 eq ) GSHP 2.36E+02 2.62E+02 2.89E+02
itive to negative, and therefore, could be a cost or a saving at the
Grid Electricity 2.10E+02 2.34E+02 2.57E+02
end of the building lifetime.
The economic performance of the optimal solution was also
analysed on a per capita basis (using the defuzzified values) to
From the above results, it can also be seen that the levelised present the results in a more generalizable form, as listed in
cost of energy of the system remains within an acceptable range. Table 14. These values can be in benchmarking building RE sys-
The likely value of the grid electricity price is $0.0936 per kWh. tems, to maintain the levelised cost and other economic indicators
While the grid parity constraint was set at +20%, the LCOE for at acceptable levels.
the best combinations vary from −0.45% to +0.05% from the grid
electricity price. Therefore, the above energy solutions do not 4.2. Environmental impact analysis of RE integration
pose an additional energy cost burden on the building occupants.
Moreover, an annual energy cost saving ranging from $45,089 to The life cycle environmental impacts of the conventional energy
$48,980 can be observed for the selected solutions. The following supply versus the hybrid renewable energy supply was compared
curves in Fig. 5 depict the variation in energy asset value over the based on normalised aggregated LCI for all impact categories, using
84 H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89

Table 13
Economic indicators for the best energy system combinations.

Combinations LCOE TESC AOS IC SPP DPP

Rank PV (kW) ST (#) GSHP (ton) $/kWh $ $/year $ Years Years

1 116.6 0 17 0.0937 $6310,887 $45,519 $762,314 16.75 23.61


2 116.6 0 18 0.0935 $6634,438 $46,673 $772,408 16.55 23.21
3 116.05 0 17 0.0936 $6307,777 $45,448 $759,527 16.71 23.54
4 116.05 0 18 0.0935 $6631,328 $46,602 $769,622 16.51 23.14
5 115.5 0 17 0.0935 $6304,667 $45,376 $756,741 16.68 23.47
6 116.6 0 19 0.0934 $6957,991 $47,827 $782,503 16.36 22.84
7 115.5 0 18 0.0934 $6628,218 $46,530 $766,836 16.48 23.07
8 114.95 0 17 0.0935 $6301,557 $45,304 $753,955 16.64 23.40
9 116.05 0 19 0.0934 $6954,881 $47,755 $779,716 16.33 22.77
10 116.6 0 20 0.0933 $7281,544 $48,980 $792,597 16.18 22.48
11 114.95 0 18 0.0934 $6625,108 $46,458 $764,049 16.45 23.00
12 114.4 0 17 0.0934 $6298,447 $45,233 $751,168 16.61 23.33
13 115.5 0 19 0.0933 $6951,771 $47,683 $776,930 16.29 22.70
14 116.05 0 20 0.0932 $7278,434 $48,908 $789,811 16.15 22.41
15 114.4 0 18 0.0933 $6621,997 $46,387 $761,263 16.41 22.94
16 113.85 0 17 0.0934 $6295,337 $45,161 $748,382 16.57 23.26
17 114.95 0 19 0.0932 $6948,660 $47,612 $774,144 16.26 22.63
18 115.5 0 20 0.0932 $7275,324 $48,837 $787,025 16.12 22.35
19 113.85 0 18 0.0933 $6618,887 $46,315 $758,476 16.38 22.87
20 113.3 0 17 0.0933 $6292,227 $45,089 $745,596 16.54 23.19

Fig. 5. Depreciation of energy asset value.

Fig. 6. Membership function for total energy system life cycle cost.
H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89 85

Fig. 7. Comparison of normalised life cycle impacts between conventional and renewable energy systems.

Table 14 improved, thereby increasing the utility and effectiveness of solar


Per capita economic performance for building energy systems.
thermal technology in the building energy model.
Parameter Unit Value The results indicate that under the current conditions and con-
Total energy system cost $/person 971 straints, an average MURB in British Columbia cannot reach net-
Annual operational cost savings $/person-year 7 zero status. For the best solution, the possible RE fraction of the
Investment cost $/person 117 total energy demand varies between 18–70%, with a likely (defuzzi-
Total LCC of system $/person 57 fied) value of 44%. One reason for this is the limited resource avail-
ability. MURBs face a special challenge in resource availability due
to the lower roof area to total floor area ratio. This essentially im-
the weighting scheme defined in Section 2.3.1. The results show plies that while the energy demand of the building increases with
that overall life cycle impacts reduce by 11.4% under the hybrid RE the number of storeys (and the corresponding increase in floor
system selected as the optimal energy solution. While generating area), the roof area does not increase proportionately, as would be
approximately 54% of the total energy supply, grid electricity con- the case in single family detached or attached housing. This leads
tributes to 76.21% of the overall life cycle impacts of the hybrid RE to a lower solar energy potential in MURBs compared to single
system as illustrated in Fig. 7. family housing stock. The RE potential also depends on the users’
A secondary analysis conducted only for climate change impact willingness to spend on the energy supply. Further analysis on
category revealed that annual carbon emissions can be reduced the potential RE supply by changing the optimisation model con-
from 167,277 kgCO2eq to 130,979 kgCO2eq , which is a reduction straints revealed that the the energy fraction supplied from RE can
of 21.70%. For every $ invested on the energy system, there is a be increased up to approximately 50% on the average with a higher
GHG saving equivalent to 0.0349 kgCO2eq . On the basis of building LCOE cut-off point (50% above the grid parity level). Under the
floor area, there is an annual reduction of 5.18 kgCO2eq /m2 of GHG present macro-economic conditions, it would be necessary to sur-
emissions. pass the grid parity point significantly for a building to reach net-
Based on the above results, it can be seen that implementing zero status. However, this may improve with the changing costs
the hybrid RE building energy system can deliver operational en- of RE technologies. RE deployment costs are reducing across the
ergy cost savings and life cycle environmental impact reductions. globe, and PV module prices was predicted to drop by 35% in 2018
[86]. This trend of decreasing costs will help to move the build-
ing stock closer to net-zero status in the coming years. Another
aspect to note is that this analysis estimated building energy de-
5. Discussion
mand based on the historical data from existing buildings, without
introducing the impact of energy efficiency and conservation mea-
Based on the above analysis, solar PV and GSHP are the rec-
sures that reduce building energy demand. For example, by reduc-
ommended RE technologies for MURBs in BC. Solar thermal en-
ing a building’s vertical surface area to floor area ratio to 0.5 based
ergy generation has been excluded from the optimal energy sys-
on the BC Energy Step Code recommendations, energy use inten-
tem combinations due to the lower energy generation potential
sity can be reduced by 7%, and thermal energy demand intensity
compared to solar PV. As PV and ST are competing for the same
can be reduced by 20% [87]. The construction of code complying
rooftop area, PV has been determined to be the more lucrative in-
buildings may make the net-zero status more achievable for the
vestment based on the cost and energy generation potential. One
BC building stock.
reason for the above is that ST energy generation potential reduces
The optimal energy system combination leads to a reduction
significantly during the winter months, and is zero in December
of 11.4% in the life cycle environmental impacts at building level.
(see Appendix II). Another factor is that the heat retention was
In British Columbia, the central electricity grid is “greener” com-
assumed to be only 50% efficient, and therefore, only half of the
pared to other Canadian provinces such as Alberta, where there
theoretical energy production potential of the collectors was con-
is a much higher reliance on fossil fuel. The environmental ben-
sidered as the available ST supply in the optimisation model. With
efits of substituting the conventional energy supply with on-site
better hot water storage mechanisms, this available energy can be
86 H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89

renewable energy generation will be much higher in such regions options at building and community levels. The developed tools can
with a less “green” energy supply. A point of interest is that build- also be used in assessing the feasiblity of achieving net-zero status
ing level hybrid energy systems lead to on-site energy genera- in small and medium scale communities.
tion, thereby reducing the environmental impacts related to volt-
age conversion, transmission, and transport of conventional energy
supplies. In fact, in BC electricity supply, only 11.1% of the life cy- 5.1. Limitations and future work
cle GHG emissions are related to energy production based on the
life cycle impact assessment. The rest of the emissions are due to A limitation in the developed model is that it considers the
voltage transformation and transmission. daily heat load as a flat curve, with no hourly variations. To fur-
The economic analysis reveals that operational energy cost sav- ther refine the model and size the geothermal heating system to
ings are achievable by implementing a hybrid RE system for the meet the exact peak heating load of the building, the hourly heat
MURB. The economic impact assessment should be viewed through demand needs to be taken into account. A bin method using the
a multi-stakeholder lens to identify the true implications. The en- hourly heating degree days variation (based on the outside tem-
ergy cost savings indicate an economic benefit for the occupants perature and the required inside temperature of the building) can
during the operational phase of a MURB. However, the invest- help to address this limitation, by breaking down the daily thermal
ment necessary to implement this energy system is incurred by the load into hourly values and thereby identifying the actual peak de-
building developers. Therefore, subsidies and other financial incen- mand.
tives are required to motivate the building developers to install RE The results can be further refined by obtaining more exact data
based energy systems in MURB. This problem is highly relevant in on the energy use and RE generation potential. The solar perfor-
the MURB context in BC, as MURBs are not constructed by the oc- mance ratio provides a general assessment of the PV system losses
cupants, and the building management is carried out by the prop- and PV module efficiency. However, the exact efficiency details of
erty owner (in the case of rental apartments) or the building strata a selected panel type and system configuration can be used for
corporations. Based on the ownership type (rental or owned), the a more accurate prediction of the PV electricity generation. The
replacement costs for energy assets during building lifetime can building energy demand analysis depends on the baseline energy
be incurred by the building managers or the homeowners. This method described under Section 2.1. The actual energy demand
may lead to a principal agent problem, where the bearer of the en- and end use distribution across the months may change from the
ergy system cost may not necessarily be the party who is benefit- predicted model for real-world building applications. However, if
ting from the outcomes [20]. The discounted payback period analy- the developed model is used for planning the energy system of an
sis indicates that the energy system investment reached breakeven actual building, the developers will be able to provide more reli-
point before the assets have to be replaced, thereby supporting the able and specific information on their building, thereby eliminating
economic viability of the proposed system. The LCOE does not vary this limitation.
significantly from the conventional energy prices, thereby leading The RE planning model needs to be combined with an energy
to a minimal financial impact on the occupants due to the shift efficiency and conservation planning model to investigate the true
from conventional to renewable energy. feasibility of developing net-zero energy buildings in Canada. The
Under the LCC analysis using DSW algorithm, there is 0.8 mem- goal of net-zero energy planning is to meet the energy demand re-
bership to an LCC of zero. By carrying out a fuzzy based analysis duced through efficient practices with locally available RE. There-
instead of a deterministic analysis, the effect of uncertainties and fore, retrofitting and other demand reduction measures need to be
variations on the final outcome has been demonstrated. A deter- investigated to identify the potential for reducing building energy
ministic analysis would only have calculated the LCC as a single demand. Moreover, the model can be extended to other types of
value, either a cost or a saving. However, the fuzzy-based possi- residential buildings such as single family detached or single fam-
bilistic analysis shows that there is a possibility of either a cost or ily attached housings. It can also be adapted to commercial and
a saving in the total LCC for the energy system. institutional buildings.
The above obtained results are based on the user priorities and The scope of the current study was limited to the standard end
constraints selected for the case study demonstration. Changing use categories with the goal of supplying the building energy de-
the importance weights for the performance objectives and the mand under the present conditions. However, electric vehicle (EV)
constraints can change the results of the optimisation considerably. energy demand will be a major concern in building-level energy
While the minimum possible RE fraction was set as 40%, increasing supply planning in the future, as the EV penetration gradually in-
this level will eliminate many of the currently viable solutions, and creases. This aspect should be taken into account when indicat-
reducing the minimum RE level will make many more combina- ing the energy demand and end uses in the future phases of this
tions viable. A similar phenomenon will occur when the LCOE cap study.
constraint (grid parity factor) is changed. Therefore, this building The next step to be tackled is the development of commu-
energy optimisation model takes stakeholder priorities and prefer- nity level RE systems. While building level RE generation supply
ences into account in the decision making. a fraction of the community energy demand, centralised (commu-
The developed model can be used as an energy system plan- nity level) RE generation facilities are necessary for developing net-
ning tool during the pre-project planning stage of MURB construc- zero energy communities. Net-zero energy communities can be de-
tion. While the model has been demonstrated for a representative veloped through a combined approach of building level and cen-
building for the BC MURB stock, the results can be further refined tralised RE generation. Optimal energy planning can be extended
and made relevant to a particular case study by providing the spe- to this context for community level RE projects. This is especially
cific information (e.g. floor area, roof area, roof configuration, en- important in planning a better energy future for off-grid remote
ergy demand and end uses) for a selected building as model in- communities.
puts. Alternatively, it can also be used in urban energy planning Since the cost factors, performance characteristics of RE tech-
based on the average data for a region, to analyse the RE poten- nologies, and building energy use change significantly over time,
tial for the regional building stock, and to assess the feasibility of these results need to be subjected to a time-based analysis for
net-zero communities. further clarity. The future viability of the planned energy systems
Based on the findigs of this study, Excel-based decision sup- should be assessed by considering the temporal variability of envi-
port tools were developed to identify the optimal energy supply ronmental and system aspects.
H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89 87

6. Conclusions ipal and provincial planners in net-zero energy community plan-


ning. This work will be further supported by the implementation
This research proposed a framework for planning an optimised of a federal renewable energy plan which can act as a long-term
hybrid RE system at building level, to support the net-zero devel- roadmap for sustainable community planning.
opment goals of Canada and British Columbia. The optimisation
model was developed considering the objectives of minimising en- Conflict of interest
ergy system cost, maximising operational cost savings, minimis-
ing the life cycle environmental impacts, and maximising the RE None.
fraction. A combinatorial optimisation approach was adopted to re-
flect the practical engineering aspects of energy planning problems Appendix I. Symbols used in equations
based on technologies available in the market. The results indi-
cated that under the defined stakeholder priorities and constraints, EIb = Baseline energy intensity
ground source heat pumps and solar PV are the optimal energy Af = Gross floor area
choices for MURBs in the selected site location in British Columbia, EIm = monthly energy intensity
Canada, and the optimal energy system combination supplied 44% Emb = baseline energy demand
of the building’s energy demand through RE. Therefore, at build- Emj = Monthly energy demand
ing level, reaching net-zero status is not a feasible goal with the SHImj = monthly space-heating load
current resource potential and economic conditions. However, sig- AB = Building gross floor area (m2 )
nificant emissions and operational cost reductions can be achieved SHB = Monthly space heating load of the building (kWh)
with the hybrid RE system. Overall life cycle impacts can be re- BLEB = Monthly baseline energy load of the building (kWh)
duced by 11.4%, and annual carbon emissions can be reduced by FW H = Fraction for water heating from the baseline load
21.70%. For the optimal solutions, the levelised cost of energy does W HB = Monthly water heating load of the building (kWh)
not vary above ±0.05% from the normal grid electrcity price, and OEB = Monthly other electric load of the building (kWh)
the system is paid back within 22–23 years. Moreover, the per Asolar = Solar installed roof area (m2 );
capita annual energy cost saving with the proposed system is $7. APV = PV installed roof area (m2 );
The uncertainties in the decision variables result in the LCC of the AST = Solar thermal installed roof area (m2 )
system varying over a range of possible values from positive to NB = Number of storeys of the building
negative in the fuzzy DSW analysis. UF = Roof utilisation factor
Planning for net-zero ready buildings with on-site RE genera- Asolar, max = Maximum rooftop area available for for solar instal-
tion would align with the BC Energy Step Code requirements, and lations (m2 )
will assist communities to become more energy secure and en- APV, unit = Area for installing 1 kW of PV (m2 )
ergy independent. However, while RE based energy systems can AST, unit = Area for installing 1 solar thermal collector (m2 )
deliver environmental benefits and energy security, it carries an AST, avail = Total area available for installing solar thermal collec-
economic burden, particularly for small communities and neigh- tors (m2 )
bourhood level applications. The investment made on RE should CPV = Installed solar PV capacity (kW)
be balanced with the economic interests of the various stakehold- nST = Numbe of installed solar thermal collectors
ers ranging from property developers to residents. The priorities PR = Performance ratio
of the stakeholders, the building energy demand, planning con- EPV = Monthly energy generation of the PV system (kWh)
straints, and requirements all vary based on the local conditions H = Monthly mean daily insolation in the plane of the PV array
for a given building. Technical, economic, environmental, and so- (kWh/m2 )
cial aspects all need to be considered together in a multi-objective N = Number of days in the month
decision making process to make the best energy choices for build- Nd = number of daylight hours
ings. This study demonstrates a method which can also consider φ = latitude
the uncertainties and variations inherent to energy planning prob- δ = declination angle
lems in selecting the optimal energy mix for a MURB. The adop- ηAP = Efficiency in terms of net aperture area
tion of a fuzzy-based method produced the results in the form of η = Efficiency in terms of gross collector area
a range of possible values, which would be more useful to deci- AST, G = Gross collector area
sion makers in coming to conclusions as opposed to a single crisp AST, A = Net aperture area
data point. A fuzzy range represents all different possibilities that EGSHP = Monthly energy generation from GSHP (kWh)
may occur, thereby providing more a comprehensive and complete CGSHP = Installed capacity of GSHP (kW)
set of information to the decision makers. It should be noted that Td = Hours operated per day
the results in urban planning problems are highly subjective to the N = Number of days in the month
available data and the selected decision making technique. By con- PSH, max = Maximum design power for space heating (kW)
sidering the life cycle environmental impacts instead of limiting It = Investment expenditure for year t;
the study to operational emissions, a more holistic approach was Mt = O&M expenditure for year t;
adopted in determining the best energy mix. This is especially im- Ft = Fuel expenditure for year t;
portant in British Columbia or other similar regions with “green” Et = Electricity generated in year t;
electricity grids. In such contexts, the low environmental impacts r = Discount rate;
of energy generation (due to higher reliance on hydropower) ob- n = Facility lifetime (service life)
scures the relatively high environmental impacts of energy conver- IMij = Impact value of the ith impact category for the jth RE
sion and transmission. Furthermore, life cycle impact assessment technology
reveals that even “zero-emission” energy supply technologies are Nij = Normalised value for the indicator
not necessarily free of environmental impacts, as they also lead to j = number of compared items (RE technologies)
other impacts due to facility construction and disposal. The devel- T EC = Total discounted energy system cost
oped framework will assist in uncertainty based holistic decision IC = Investment cost
making for building energy systems, and it can facilitate munic- F C = Fuel and conventional energy supply costs
88 H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89

RC = Rehabilitation and/or replacement cost [11] B.O. Ariyo, M.F. Akorede, I.O.A. Omeiza, S.A.Y. Amuda, S.A. Oladeji, Optimisa-
AOM = Annual operations and maintenance cost tion analysis of a stand-alone hybrid energy system for the senate building,
university of Ilorin, Nigeria, J. Build. Eng. 19 (2018) 285–294 August 2017.
DC = Disposal cost [12] P. Torcellini, S. Pless, M. Deru, D. Crawley, Zero energy buildings: a critical look
RV = Residual value at the definition, 2006 ACEEE Summer Study on Energy Efficiency in Buildings,
PV = Present value of a future one time cash flow (FV) occuring 2006.
[13] R. Ruparathna, K. Hewage, R. Sadiq, Rethinking investment planning and op-
at time t timizing net zero emission buildings, Clean Technol. Environ. Policy 19 (6)
PVA = Present value of a future annually recurring cash flow (A) (2017) 1711–1724.
over t number of years [14] M.K. Dixit, Life cycle embodied energy analysis of residential buildings: a re-
view of literature to investigate embodied energy parameters, Renew. Sustain.
SHB = Monthly space heating load of the building (kWh)
Energy Rev. 79 (October 2016) (2017) 390–413.
W HB = Monthly water heating load of the building (kWh) [15] BC Hydro, Build energy-efficient homes, 2018. [Online]. Available:
OEB = Monthly other electric load of the building (kWh) https://www.bchydro.com/powersmart/business/programs/new-home.html.
[Accessed: 17-Jul-2018].
BLEB = Monthly baseline energy load of the building (kWh)
[16] Z. Huang, H. Yu, Z. Peng, M. Zhao, Methods and tools for community energy
AOS = Annual operational energy cost savings planning: a review, Renew. Sustain. Energy Rev. 42 (4800) (2015) 1335–1348.
GEP = Grid electricity price [17] I.K.Bhat Varun, R. Prakash, LCA of renewable energy for electricity generation
NGP = Natural gas price systems—a review, Renew. Sustain. Energy Rev. 13 (5) (Jun. 2009) 1067–1073.
[18] F. deLlano-Paz, A. Calvo-Silvosa, S.I. Antelo, I. Soares, Energy planning and
CSC = Conventional supply costs modern portfolio theory: a review, Renew. Sustain. Energy Rev. 77 (March
RF C = Renewable fuel costs 2016) (2017) 636–651.
w j = weight assigned to the jth performance objective [19] N. Nord, Building energy efficiency in cold climates, in: Encyclopedia of Sus-
tainable Technologies, Elsevier, 2017, pp. 149–157.
ri j = normalised fuzzy number [20] H. Karunathilake, K. Hewage, R. Sadiq, Opportunities and challenges in energy
vi j = weighted normalised value of the ith solution under the demand reduction for Canadian residential sector: a review, Renew. Sustain.
jth performance objective Energy Rev. 82 (February 2018) 2005–2016.
[21] Natural Resources Canada, Residential sector: British Columbia and
pi = overall performance score Territories, Comprehens. Energy Use Database (2018) [Online]. Avail-
DPP = discounted payback period able: http://oee.nrcan.gc.ca/corporate/statistics/neud/dpa/menus/trends/
AOS = Annual operational energy cost savings comprehensive/trends_res_bct.cfm . [Accessed: 30-May-2018].
[22] Natural Resources Canada, Commercial/Institutional Sector: British Columbia
COGXi = Center of gravity
and Territories, Comprehens. Energy Use Database (2018) [Online]. Avail-
able: http://oee.nrcan.gc.ca/corporate/statistics/neud/dpa/menus/trends/
comprehensive/trends_com_bct.cfm . [Accessed: 30-May-2018].
Appendix II. Energy generation potential for a unit installation
[23] O. Erdinc, M. Uzunoglu, Optimum design of hybrid renewable energy systems:
of solar energy technologies overview of different approaches, Renew. Sustain. Energy Rev. 16 (3) (2012)
1412–1425.
[24] F. Harkouss, F. Fardoun, P.H. Biwole, Multi-objective optimization methodology
for net zero energy buildings, J. Build. Eng. 16 (October 2017) (2018) 57–71.
Technology Solar PV Solar thermal
[25] M. Arriaga, C.A. Canizares, M. Kazerani, Renewable Energy Alternatives for Re-
Unit kWh/ kWh/ kWh/ kWh/ kWh/ kWh/ mote Communities in Northern Ontario, Canada, IEEE Trans. Sustain. Energy 4
kW kW kW panel panel panel (3) (Jul. 2013) 661–670.
[26] A. Petrillo, F. De Felice, E. Jannelli, C. Autorino, M. Minutillo, A.L. Lavadera,
Fuzzy number a b c a b c Life cycle assessment (LCA) and life cycle cost (LCC) analysis model for a
January 35.57 41.85 48.13 2.15 2.39 2.63 stand-alone hybrid renewable energy system, Renew. Energy 95 (Sep. 2016)
February 53.01 62.37 71.73 22.84 25.38 27.91 337–355.
March 88.93 104.63 120.32 56.68 62.97 69.27 [27] T. Hong, C. Koo, T. Kwak, H.S. Park, An economic and environmental assess-
April 97.73 114.98 132.22 60.63 67.37 74.10 ment for selecting the optimum new renewable energy system for educational
May 104.35 122.76 141.17 58.17 64.64 71.10 facility, Renew. Sustain. Energy Rev. 29 (2014) 286–300.
[28] A.A. Hassan, F.H. Fahmy, A.E.S.A. Nafeh, M.A. Abu-elmagd, Genetic single ob-
June 104.04 122.40 140.76 55.83 62.04 68.24
jective optimisation for sizing and allocation of renewable DG systems, Int. J.
July 113.63 133.69 153.74 68.12 75.69 83.26
Sustain. Energy 36 (6) (2017) 545–562.
August 112.05 131.83 151.60 73.85 82.06 90.27
[29] Y. Liu, H. Li, and Y. Chen, Development of a single-level optimization model for
September 102.89 121.05 139.21 73.69 81.87 90.06 energy planning - a case study of Shanxi, China, 2017, p. 040 0 07.
October 79.25 93.23 107.22 49.91 55.46 61.00 [30] S.D. Pohekar, M. Ramachandran, Application of multi-criteria decision making
November 42.65 50.18 57.70 11.09 12.32 13.55 to sustainable energy planning - A review, Renew. Sustain. Energy Rev. 8 (4)
December 30.24 35.57 40.91 0.00 0.00 0.00 (2004) 365–381.
Total 964.35 1134.53 1304.70 532.96 592.18 651.40 [31] C. Hung, Environmental Impacts of Renewable Energy, Norwegian University
of Science and Technology, 2010.
References [32] R. Siddaiah, R.P. Saini, A review on planning, configurations, modeling and op-
timization techniques of hybrid renewable energy systems for off grid applica-
[1] O. Ellabban, H. Abu-Rub, F. Blaabjerg, Renewable energy resources: Current sta- tions, Renew. Sustain. Energy Rev. 58 (2016) 376–396.
tus, future prospects and their enabling technology, Renew. Sustain. Energy [33] M. Gitizadeh, M. Kaji, J. Aghaei, Risk based multiobjective generation expan-
Rev. 39 (2014) 748–764. sion planning considering renewable energy sources, Energy 50 (2013) 74–82.
[2] M. Höök, X. Tang, Depletion of fossil fuels and anthropogenic climate [34] Y.P. Cai, G.H. Huang, Q. Tan, Z.F. Yang, Planning of community-scale renewable
change—A review, Energy Policy 52 (Jan. 2013) 797–809. energy management systems in a mixed stochastic and fuzzy environment, Re-
[3] S. Bilgen, K. Kaygusuz, A. Sari, Renewable energy for a clean and sustainable new. Energy 34 (7) (Jul. 2009) 1833–1847.
future, Energy Sour 26 (12) (2004) 1119–1129. [35] C. Koroneos, M. Michailidis, N. Moussiopoulos, Multi-objective optimization in
[4] S. Thompson, B. Duggirala, The feasibility of renewable energies at an off-grid energy systems: the case study of Lesvos Island, Greece, Renew. Sustain. En-
community in Canada, Renew. Sustain. Energy Rev. 13 (9) (2009) 2740–2745. ergy Rev. 8 (1) (2004) 91–100.
[5] Intergovernmental Panel on Climate Change, Renewable energy sources and [36] A.T.D. Perera, M.P.G. Sirimanna, R.A. Attalage, K.K.C.K. Perera, V.P.C. Das-
climate change mitigation: special report of the Intergovernmental Panel on sanayake, Multi objective optimization and multi criterion decision making
Climate Change, Choice Rev. Online 49 (11) (Jul. 2012). 496309–496309. in expanding existing standalone energy systems combining renewable energy
[6] Intergovernmental Panel on Climate Change, Climate Change 2014 Mitigation sources, Eng. Appl. Sci. (2012) 234–238.
of Climate Change, Cambridge University Press, Cambridge, 2014. [37] D.A. Haralambopoulos, H. Polatidis, Renewable energy projects: structuring a
[7] N. Carlisle, O. Van Geet, S. Pless, Definition of a “Zero Net Energy” Community, multi-criteria group decision-making framework, Renew. Energy 28 (6) (2003)
2009. 961–973.
[8] D. Mazzeo, G. Oliveti, C. Baglivo, P.M. Congedo, Energy reliability-constrained [38] J.R. San Cristóbal, Multi-criteria decision-making in the selection of a renew-
method for the multi-objective optimization of a photovoltaic-wind hybrid able energy project in spain: the Vikor method, Renew. Energy 36 (2) (Feb.
system with battery storage, Energy 156 (2018) 688–708. 2011) 498–502.
[9] V. Khare, S. Nema, P. Baredar, Solar-wind hybrid renewable energy system: a [39] Y.P. Cai, G.H. Huang, Z.F. Yang, Q.G. Lin, Q. Tan, Community-scale renewable en-
review, Renew. Sustain. Energy Rev. 58 (2016) 23–33. ergy systems planning under uncertainty—An interval chance-constrained pro-
[10] H. Dagdougui, R. Minciardi, A. Ouammi, M. Robba, R. Sacile, Modeling and op- gramming approach, Renew. Sustain. Energy Rev. 13 (4) (May 2009) 721–735.
timization of a hybrid system for the energy supply of a ‘green’ building, En- [40] R. Ruparathna, K. Hewage, R. Sadiq, Economic evaluation of building energy
ergy Convers. Manag. 64 (2012) 351–363. retrofits: A fuzzy based approach, Energy Build 139 (2017) 395–406.
H. Karunathilake, K. Hewage and J. Brinkerhoff et al. / Energy & Buildings 201 (2019) 70–89 89

[41] Ü. Şengül, M. Eren, S. Eslamian Shiraz, V. Gezder, A.B. Şengül, Fuzzy TOPSIS [63] R. Hischier, et al., Implementation of Life Cycle Impact Assessment Methods
method for ranking renewable energy supply systems in Turkey, Renew. En- Data v2.2 (2010), Dübendorf, 2010.
ergy 75 (Mar. 2015) 617–625. [64] M. Goedkoop, M. Oele, J. Leijting, T. Ponsioen, E. Meijer, Introduction to LCA
[42] H. Karunathilake, K. Hewage, W. Mérida, R. Sadiq, Renewable energy selection with SimaPro, San Francisco (2016).
for net-zero energy communities: Life cycle based decision making under un- [65] R. Heijungs, M.a J. Huijbregts, A review of approaches to treat uncertainty in
certainty, Renew. Energy 130 (Jan 2019) 558–573. LCA, iEMSs 2004 Int. Congr. (2004) 8.
[43] R. Sadiq, M.J. Rodriguez, Fuzzy synthetic evaluation of disinfection by-product- [66] PRé, SimaPro Database Manual: Methods Library, PRé, San Francisco, 2019.
s—a risk-based indexing system, J. Environ. Manage. 73 (1) (Oct. 2004) 1–13. [67] T.T. Muthukumarana, H.P. Karunathilake, H.K.G. Punchihewa, M.M.I.D. Manthi-
[44] Statistics Canada, Evolution of housing in Canada, 1957 to 2014, lake, K.N. Hewage, Life cycle environmental impacts of the apparel industry in
2016. [Online]. Available: https://www150.statcan.gc.ca/n1/pub/11- 630- x/ Sri Lanka: Analysis of the energy sources, J. Clean. Prod. 172 (2018) 1346–1357.
11- 630- x2015007- eng.htm. [Accessed: 26-Jan-2017]. [68] Environment Canada, Government of Canada announces 2030 emissions
[45] United States Environmental Protection Agency, Geothermal Heating and target, 2015. [Online]. Available: https://www.canada.ca/en/news/archive/2015/
Cooling Technologies, Renew. Heat. Cool. (2016) [Online]. Available: https: 05/government- canada- announces- 2030- emissions- target.html. [Accessed:
//www.epa.gov/rhc/geothermal- heating- and- cooling- technologies . [Accessed: 17-Sep-2016].
30-May-2018]. [69] The Canadian Chamber of Commerce, Climate Change and Canadian Business:
[46] S. Zhang, P. Huang, Y. Sun, A multi-criterion renewable energy system de- What Does COP21 Mean for Canadian Business ?, 2015.
sign optimization for net zero energy buildings under uncertainties, Energy [70] National Renewable Energy Laboratory, Distributed generation renewable en-
94 (2016) 654–665. ergy estimate of costs, Energy Anal. (2016) [Online]. Available https://www.
[47] G. Finch, E. Burnett, W. Knowles, Energy consumption and conservation in mid nrel.gov/analysis/tech- lcoe- re- cost- est.html . [Accessed: 06-Jul-2018].
and high rise residential buildings in British Columbia, Building Enclosure Sci- [71] P.-T. Chang, E.S. Lee, The estimation of normalized fuzzy weights, Comput.
ence & Technology (BEST2) Conference, 2010. Math. with Appl. 29 (5) (Mar. 1995) 21–42.
[48] Natural Resources Canada, Ground-Source Heat Pumps (Earth-Energy Sys- [72] V. Anandan, G. Uthra, Defuzzification by area of region and decision mak-
tems), 2017. [Online]. Available: http://www.nrcan.gc.ca/energy/publications/ ing using Hurwicz criteria for fuzzy numbers, Appl. Math. Sci. 8 (63) (2014)
efficiency/heating- heat- pump/6833?attr=4. [Accessed: 06-Jul-2018]. 3145–3154.
[49] N. Badea (Ed.), Design for Micro-Combined Cooling, Heating and Power Sys- [73] S. Salahshour, S. Abbasbandy, T. Allahviranloo, Ranking fuzzy numbers using
tems, Springer London, London, 2015. fuzzy maximizing-minimizing points, in: EUSFLAT-LFA 2011, 2011, pp. 763–769.
[50] S. Pelland, Y. Poissant, An evaluation of the potential of building integrated [74] J. Krejčí, Pairwise Comparison Matrices and their Fuzzy Extension, 366,
photovoltaics in Canada, 31st Annual Conference of the Solar Energy Society Springer International Publishing, Cham, 2018.
of Canada (SESCI)., 2006. [75] BC Ministry of Environment, 2016 BC best practices methodology for quantify-
[51] B. Marion, et al., Performance parameters for grid-connected PV systems, in: ing greenhouse gas emissions, Victoria (2016).
Conference Record of the Thirty-first IEEE Photovoltaic Specialists Conference, [76] (RDH) RDH Building Science Inc, Energy Consumption in Low-Rise Multi-Fam-
2005., 31, 2005, pp. 1601–1606. February. ily Residential Buildings in Bc, 2017.
[52] S. Pelland, et al., The development of photovoltaic resource maps for Canada, [77] Natural Resources Canada, Photovoltaic And Solar Resource Maps, 2017 [On-
31st Annual Conference of the Solar Energy Society of Canada, 2006. line]. Available: http://www.nrcan.gc.ca/18366 [Accessed: 10-Jul-2018].
[53] L. Hughes and T. Wood, Solar energy and multi-storey residential buildings, [78] ClearSky Advisors Inc., Survey of Active Solar Thermal Collectors, Industry and
2007. Markets in Canada (2012), Ottawa ON, 2015.
[54] American Society of Heating Refrigerating and Air-Conditioning Engineers Inc., [79] Solar Rating & Certification Corporation - International Code Council, Rat-
2015 ASHRAE Handbook - Heating, Ventilating, and Air-Conditioning Applica- ings summary page, 2018. [Online]. Available: https://secure.solar-rating.
tions (SI Edition), American Society of Heating, Refrigerating and Air-Condi- org/Certification/Ratings/RatingsSummaryPage.aspx?type=1. [Accessed: 10-Jul-
tioning Engineers, Inc., 2015. 2018].
[55] BC Hydro, Should you get a heat pump?, 2915. [Online]. Available: https: [80] Solar Rating & Certification Corporation - International Code Council, OG-
//www.bchydro.com/news/conservation/2015/heat-pump.html. 100 ICC-SRCC Certified Solar Collector #2012027A, 2018. [Online]. Available:
[56] National Energy Board of Canada, Market Snapshot: Steady growth for heat https://secure.solar-rating.org/Certification/Ratings/RatingsReport.aspx?device=
pump technology, 2018. [Online]. Available: https://www.neb-one.gc.ca/nrg/ 1691&units=METRICS. [Accessed: 10-Jul-2018].
ntgrtd/mrkt/snpsht/2018/02- 03htpmps- eng.html?=undefined&wbdisable=true. [81] H. Karunathilake, P. Perera, R. Ruparathna, K. Hewage, R. Sadiq, Renewable en-
[Accessed: 11-Jul-2018]. ergy integration into community energy systems: A case study of new urban
[57] Natural Resources Canada, RETScreen Software Online User Manual: Ground– residential development, J. Clean. Prod. 173 (Feb. 2018) 292–307.
Source Heat Pump Project Model, 1, RETScreen® International, September. [82] R. Ruparathna, et al., Climate conscious regional planning for fast-growing
Varennes, QC, 2005. communities, J. Clean. Prod. 165 (Nov 2017) 81–92.
[58] K.L. Hoffman, Combinatorial optimization: Current successes and directions for [83] D. Heerema, Gas vs. electricity? Comparing home heating costs in B.C, Pem-
the future, J. Comput. Appl. Math. 124 (1–2) (20 0 0) 341–360. bina Institute Blog (2017) [Online]. Available: http://www.pembina.org/blog/
[59] Canadian Solar Inc., Solar modules, 2018. [Online]. Available: https://www. gas- vs- electricity . [Accessed: 17-Jul-2018].
canadiansolar.com/solar- panels/all- black.html. [Accessed: 27-May-2018]. [84] BC Hydro, Generation system, 2016. [Online]. Available: https://www.bchydro.
[60] B. Reza, R. Sadiq, K. Hewage, A fuzzy-based approach for characterization of com/energy- in- bc/operations/generation.html. [Accessed: 12-Jul-2018].
uncertainties in emergy synthesis: an example of paved road system, J. Clean. [85] G.S. Denis, P. Parker, Community energy planning in Canada: The role of re-
Prod. 59 (2013) 99–110. newable energy, Renew. Sustain. Energy Rev. 13 (8) (2009) 2088–2095.
[61] D. Gu Choi, S. Yong Park, N.-B. Park, J. Chul Hong, Is the concept of ‘grid par- [86] Bloomberg LP, Chinese Burn Will Only Make the Solar Industry Stronger, 2018.
ity’ defined appropriately to evaluate the cost-competitiveness of renewable [Online]. Available: https://www.bloomberg.com/opinion/articles/2018- 06- 05/
energy technologies? Energy Policy 86 (Nov. 2015) 718–728. chinese- burn- only- makes- the- solar- industry- stronger. [Accessed: 17-Apr-
[62] H.Y. Lin, P.Y. Hsu, G.J. Sheen, A fuzzy-based decision-making procedure for data 2017].
warehouse system selection, Expert Syst. Appl. 32 (3) (2007) 939–953. [87] BC Housing, Energy Step Code Building Beyond the Standard: 2017 Metrics Re-
search Full Report, 2017 Vancouver.

You might also like