Energy 102 (2016) 416e426

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

An integrated multi-objective optimization model for determining the

optimal solution in the solar thermal energy system
Jimin Kim a, Taehoon Hong a, *, Jaemin Jeong a, Myeonghwi Lee a, Choongwan Koo a, b,
Minhyun Lee a, Changyoon Ji a, Jaewook Jeong a
a
Department of Architectural Engineering, Yonsei University, Seoul 03722, Republic of Korea
b
Division of Construction Engineering and Management, Purdue University, West Lafayette, IN 47906, USA

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

Article history: The STE (solar thermal energy) system is considered an important new renewable energy resource.
Received 14 November 2015 While various simulations are used as decision-making tools in implementing the STE system, it has a
Received in revised form limitation in considering both diverse impact factors and target variables. Therefore, this study aimed to
15 February 2016
develop an integrated multi-objective optimization model for determining the optimal solution in the
Accepted 16 February 2016
Available online 12 March 2016
STE system. As the optimization algorithm, this study utilizes GA (genetic algorithm) to select optimal
STE system solution. Using crossover and mutation, GA investigates optimal STE system solution. The
proposed model used GA based on the software program Evolver 5.5. The proposed model presents high
Keywords:
Solar thermal energy system
available and efficient results as decision-making tools. First, to determine the optimal solution, a total of
Multi-objective optimization 30,407,832 possible scenarios were generated by considering various factors in terms of their high
Generic algorithm availability. Second, in terms of efficiency, an average of 131 s were used to determine the optimal so-
Economic and environmental assessment lution out of the previously proposed various scenarios. The proposed model can become a tool for
consumers to decide on the optimal solution for the design of the STE system.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction implementation of NRE, and is planning to expand the NRE share in

Due to the global increase in the use of fossil fuels, there have
been various climate issues, such as global warming. To solve such
environmental problems, the world has taken greater interest in
diverse NRE (new renewable energy) technologies. Among NREs,
the world has taken the greatest interest in solar energy [1e8].
Particularly, the STE (solar thermal energy) system is considered a
highly efficient system that converts solar radiation to heat. It
economically reduces CO2 emissions by reducing the use of fossil
fuels. It is one of the most widely available NRE technologies today
that is applicable to households or industries [9e12].
According to the second energy master plan of South Korea, the
country's final energy consumption rate from 2000 to 2012
increased at the yearly average of 2.8%, and energy consumption of
households accounted for 38% of the total energy consumption.
Therefore, the South Korean government is expanding the
* Corresponding author. Yonsei University, 50 Yonsei-ro, Seodaemun-gu, Seoul
120-749, Republic of Korea. Tel.: þ82 2 2123 5788; fax: þ82 2 365 4668.
E-mail address: hong7@yonsei.ac.kr (T. Hong).
the country's total energy portfolio to 11% by 2035 [13]. To promote
such expanded implementation of NRE, the South Korean govern-
ment supports the initial investment cost through incentive pol-
icies such as housing support or building support [14]. The total STE
system area in South Korea is 48,473 m2 as of 2013, producing
27,812 TOE (tons of equivalent) energy [15]. In the country's fourth
basic plan for NRE, the South Korean government aims to increase
the monthly primary energy production ratio of the STE system to
the total energy production of NRE technologies from 0.3% in 2012
to 7.9% by 2035 (i.e., 21.2% of the yearly average increase rate),
which is the maximum increase rate among NRE technologies [16].
Meanwhile, in South Korea, hot water heater of energy con-
sumption has accounted for 17% of the total energy consumptions.
In order to solve this problem, Korean government recommends
high efficiency boiler to consumers, however most of them use near
90% efficiency boiler. In order to make up for what they lack at hot
water demand, Korean government has promoted the imple-
mentation of the STE system [17]. Therefore, in this paper, the
target use of the STE system has been set to solar hot water system.
In South Korea, the STE system is expected to show the biggest
increase potential. Although the STE system has several advantages

http://dx.doi.org/10.1016/j.energy.2016.02.104
0360-5442/© 2016 Elsevier Ltd. All rights reserved.
 J. Kim et al. / Energy 102 (2016) 416e426
Abbreviations
AHG Annual Heat Generation
AHG/Unit Annual Heat Generation per Unit
AoC Azimuth of the installed Collector
BEP Break-Even-Point
GA Genetic Algorithm
GL minimum electricity Generation Limit
IIC Initial Investment Cost (with the government subsidy)
iMOO integrated Multi-Objective Optimization
LCC Life Cycle Cost
LCCO2 Life Cycle CO2
NoC Number of the installed Collector on the rooftop area
(i.e., the government's support policies and its high efficiency), it is
difficult to widely implement due to its high IIC (initial investment
cost) and its unstable performance against climate conditions. To
overcome these challenges, the full life-cycle cost of the STE system
must be analyzed before it is implemented. In other words, through
the overall analysis of the life-cycle cost, the economic feasibility of
the STE system should be confirmed with decision-makers. Toward
this end, one of the most important steps is to estimate the amount
of heat generated by the STE system. There are several impact
factors that should be considered in estimating the amount of heat
generated by the STE system. Those factors include (see Table S1,
Supplementary material (SM)): (i) the regional factors and (ii) the
building characteristics. Previous studies have analyzed the target
variables of the STE system (i.e., heat generation, economic effect,
and environmental effect) by considering the aforementioned
impact factors [2e4,9e12,18e35], [36e55]. There are many impact
factors affecting solar collection cycle, however aforementioned
impact factors (i.e. regional factors, building characteristics) are
sufficient technical factors to considered for the design of STE
systems [10,11,23,24,40]. If decision maker wants to adopt the STE
system, the proposed model could be decision support tool as
intuitive model in the early design phase. Using proposed model,
easily, decision maker can easily select the STE system by consid-
ering heat generation, economic and environmental analysis from
life cycle perspective.
First, some studies have analyzed the heat supply demand of a
building by considering the heat generation of the STE system. The
analysis of the heat generated by the STE system considered the
regional factors and the physical information of the STE system [i.e.,
the ToC (type of collector) and the ToS (type of storage)]
[2,4,21,27,33]. Allouhi et al. [4], aimed for technical evaluations of
the STE system in Morocco, used the flat plate collector and the
apricus collector to analyze the energy generated by the STE system
in six regions. Hobbi and Siddiqui [27], proposed an STE system that
could satisfy the hot water demand of single households in Canada.
For maximum heat generation, they set the characteristic of the
collector as the design variable. Other studies have focused only on
the heat supply and demand of a building with the installation of
the STE system, but have not analyzed the economic and environ-
mental effects.
Second, recent studies have analyzed the economic effect of the
installation of the STE system. These studies selected the optimal
ToC and ToS by considering mainly the IIC and heat generation of
the STE system [9,10,31,40]. Bornatico et al. [9] calculated the
optimal STE system that would satisfy the minimum initial cost and
maximum energy generation through an optimization algorithm.
Szargut and Stanek [31] determined the optimal cost by setting the
417
NoC_L Number of installed Collector along the Length of the
rooftop area
NoC_W Number of installed Collector along the Width of the
rooftop area
NPV Net Present Value
RL Rooftop Length
RW Rooftop Width
SIR Savings-to-Investment Ratio
SoC Slope of the installed Collector
ToS Type of the Storage
ToC Type of the Collector
VBA Visual Basic for Applications
design value as the characteristic of the solar collector in the eco-
nomic analysis. However, their analysis did not consider the full life
cycle and the environmental effect.
Third, recent studies have also analyzed the environmental ef-
fect of the installation of the STE system. These studies mainly
considered the regional factor and heat generation of the STE sys-
tem [17,41,43,51,54]. Zambrana-Vasquez et al. [41], performed the
life cycle assessment for the environmental assessment based on
the characteristics of the STE system. Bessa and Prado [51], con-
ducted a comparative analysis of the CO2 emissions of the electrical
heating system and of the STE system in 24 regions in Brazil to
propose an optimal solution. Though they analyzed the environ-
mental effect based on the regional factor, they did not simulta-
neously consider the life cycle cost. Based on the previous
researches, it can be concluded that several types of target variables
should be simultaneously considered to analyze the economic and
environmental feasibility of the STE system [56e62].
Therefore, this study aimed to develop an iMOO (integrated
Multi-Objective Optimization) model that would satisfy several
target variables by simultaneously considering various impact
factors. The proposed model was developed in six steps: (i) estab-
lishment of a database, (ii) generation of the STE system's instal-
lation scenarios, (iii) analysis of the STE system's installation
scenarios, (iv) economic and environmental assessment from the
life cycle perspective, (v) establishment of an iMOO process using a
GA (genetic algorithm), and (vi) systemization of the proposed
model using a Microsoft-Excel-based VBA (Visual basic application)
(see Fig. 1). Using the optimal solution, these five objectives can be
attained: (i) minimization of the IIC (with government subsidy); (ii)
maximization of the AHG (annual heat generation), (iii) maximi-
zation of the NPV (net-present value), (iv) maximization of the SIR
(savings-to-investment ratio), and (v) maximization of the AHG/
unit. In addition, Fig. 2 showed the decision-making process in the
iMOO model.
2. Materials and methods
2.1. Step 1: establishment of database
To select the optimal STE system solution, the research team
first considered the impact factors associated with the PV (photo-
voltaic) system [63,64], which are very similar to those of the STE
system, except that modules are used in the PV system whereas
solar collectors are used in the STE system [65,66]. Toward this end,
the study established a database of several impact factors based on
previous studies (see Table S1) and the Korea Public Institution and
Geographical Information [63e71]. As shown in Table S2, the
418 J. Kim et al. / Energy 102 (2016) 416e426
Fig. 1. Research framework.
parameters were divided into three categories: (i) regional
geographical information, (ii) regional meteorological information,
and (iii) physical information on the STE system. The established
database will be used in Step 2 (generation of the STE system's
installation scenarios) and Step 3 (analysis of the STE system's
installation scenarios).
As the regional geographical information, the latitudes of 16
major cities in South Korea were collected from Google EarthTM,
which offers geographic information. The meridian altitudes (at
noon of the winter solstice) were calculated as follows [67]. (refer
to Eq. (1) and see Table S3, SM):

t ¼ 90  4  ε (1)
where, t is minimum incident angle of the sun, 4 is the region
latitude, and ε is the tilt angle of the Earth's axis.
As the regional meteorological information, the monthly inci-
dent solar radiation and the monthly diffuse solar radiation of the
16 cities in South Korea were collected from the New & Renewable
Energy Data Center, through which the monthly total solar radia-
tion was calculated [68] (see Tables S4 and S5, SM). The monthly
ambient temperature and the monthly water supply temperature
in Korea were collected from the most updated data of the Korea
Meteorological Administration [69] (see Tables S6 and S7, SM). The
monthly water supply temperature is the temperature of the water
supplied to buildings through the water pipes from the waterworks
headquarters [69].
As the physical information on the STE system, the following
information on the ToC and ToS were collected from the Public
Procurement Service: (i) ToC: width, length, area, intercept
efficiency, and negative slope of the efficiency curve; and (ii) ToS:
diameter, height, and volume (see Tables S8 and S9, SM) [70].
2.2. Step 2: generation of the STE system's installation scenarios
To acquire the optimal STE system solution, the aforementioned
impact factors were considered (refer to Table S1). Based on the
previous studies [63,71], which conducted a sensitivity analysis
from the unit panel to the rooftop PV system, the complex corre-
lations among the impact factors affected the amount of electricity
generated by the rooftop PV system. Because the impact factors of
the STE system are very similar to those of the PV system [65,66],
they were divided into three categories: (i) defined parameters, (ii)
adjustable parameters, and (iii) constraint parameters.
Defined parameters are fixed parameters that cannot be
adjusted while the STE system is installed and signify the following
information on the target building: (i) the regional information on
the target building, (ii) the types of facilities and systems, and (iii)
the AoC. The AoC is the installation azimuth of the solar collector
whose efficiency is highest when it faces south (0 ) [63].
Adjustable parameters can be considered when installing the
STE system. Various STE system scenarios can be generated by
combining the adjustable parameters: (i) the SoC, (ii) the ToC, (iii)
the ToS, (iv) the number of installed collectors along with the
length of the rooftop area (NoC_L), and (v) the number of installed
collectors along with the width of the rooftop area (NoC_W).
Constraint parameters are the limitations that should be
considered in installing the STE system. They filter the scenarios
that need to be excluded from all the possible STE system scenarios:
(i) the rooftop length (RL), (ii) the rooftop width (RW), (iii) the
minimum AHG limit, and (iv) the maximum budget limit.
 J. Kim et al. / Energy 102 (2016) 416e426 419

Fig. 2. Decision-making process in the iMOO model.

Considering the above three parameter categories, scenarios diffuse solar radiation on horizontal surfaces (kWh/m2/day), b is
were generated to calculate the optimal STE system for the target the tilt angle ( ), H is the daily total radiation on horizontal surfaces
building. Using the produced scenarios, the heat generated by the (kWh/m2/day), and rg is the ground reflectivity (0.2).
STE system was analyzed in Step 3. H, the daily total radiation, is calculated using Eq. (3) [72,73]:

H ¼ Hb þHd (3)
2.3. Step 3: analysis of the STE system's installation scenarios
where, Hb is the daily incident solar radiation on horizontal sur-
The STE system scenarios generated in Step 2 were analyzed to faces (kWh/m2/day) and Hd is the daily diffuse solar radiation on
determine the optimal solution. The heat generated by the STE horizontal surfaces (kWh/m2/day).
system can be calculated from the impact factors such as the effi- Rb is calculated using Eqs. (4)e(6) [72,73]:
ciency of the solar collector, the total collector area, and the solar
h i.
radiation [refer to Eqs. (2)e(8)] [72,73]. 0 0
Rb ¼ cosðf  bÞcos d sin WS þ p 180  WS sinðf  bÞsin d
=
Eq. (2) calculates the solar radiation based on the angle [72,73]:
h i
 cos f cos d sin WS þ p 180  WS sin d
=
HT ¼ Hb Rb þHd ð1 þ cosbÞ=2 þ Hrg ð1  cosbÞ=2 (2)
(4)
where, HT is the daily average radiation on tilted surfaces (kWh/m2/
day), Hb is the daily incident solar radiation on horizontal surfaces where, 4 is the latitude angle ( ), b is the tilt angle ( ), d is the
(kWh/m2/day), Rb is the ratio of the monthly radiation incident on declination angle ( ), WS is the sunset hour angle for horizontal
an inclined plane to that on a horizontal plane, Hd is the daily surfaces ( ), and W’S is the sunset hour for tilted surfaces ( ); and
420
WS ¼ cos1 ð  tan 4 tan dÞ and
h i
0
WS ¼ min cos1 ð  tan4 tandÞ; cos1 ð  tanð4  bÞtandÞ
where, ‘min’ refers to the smaller of the two items in the brackets.
The efficiency of the collector is evaluated by applying the ra-
diation data with the installation angle, determined using Eq. (2), to
Eq. (7) [11]:
h ¼ Q u =Ap IT ¼ FrðtaÞ  Fr UL  ðTi  Ta Þ=IT
where, h is the efficiency (%), Q u is the heat generation per day
(kWh/day), AP is the total collector area (m2), IT is the incident ra-
diation (kWh/m2/day), Fr (ta) is the intercept efficiency, FrUL is the
negative slope of the efficiency curve [(W/m2)/ C], Ti is the tem-
perature of the fluid-entering collector ( C), and Ta is the ambient
temperature ( C).
Finally, the heat generated by the STE system was calculated
using Eq. (8) [72,73]:
Q u ¼ hAP IT
where, Q u is the heat generated per day (kWh/day), h is the effi-
ciency (%), AP is the total collector area (m2), and IT is the incident
radiation (kWh/m2/day).
2.4. Step 4: economic and environmental assessment from the life
cycle perspective
For the economic and environmental assessment, LCC and
LCCO2 analyses were performed. From the results, the following
four assumptions were established [67e70,74e90]: (i) the starting
point of the analyses and the analysis period, (ii) the real discount
rate, and (iii) the significant costs of ownership. Table S10, SM
shows the assumptions for the key elements of the LCC and LCCO2
analyses.
The starting points of the analysis and the analysis period were
set as 2015 and 25 years, respectively. The analysis period was
determined through interviews with STE system experts.
Data on nominal interest rates and inflation rates were collected
from the Bank of Korea Economic Statistics System and the Korean
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Fitness function ¼ ðSA  0Þ2 þ ð1SB Þ2 þ ð1SC Þ2 þ ð1SD Þ2 þ ð1SE Þ2
Statistical Information Service to calculate the real discount rate
[refer to Eq. (9)] [The real discount rates are the interest growth rate
(2.68%), the electricity price growth rate (0.98%), the gas price
growth rate (1.97%), and the KCERs (Korea certified emission re-
ductions) growth rate (2.61%)]:
i ¼ ð1 þ in Þ=ð1 þ f Þ  1
where, i is the real discount rate, in is the nominal interest rate, and
f is the inflation rate.
J. Kim et al. / Energy 102 (2016) 416e426
The costs (i.e., the initial construction cost, and the operations
(5) and maintenance cost) and benefits [i.e., the gas bill savings, gov-
ernment subsidy, and KCERs (reduction of CO2 emissions due to
energy consumption savings)] generated during the life cycle of the
STE system were considered as the significant costs of ownership.
(6)
2.5. Step 5: establishment of an iMOO process using a genetic
algorithm
The iMOO model, a more intuitive, simplified, applicable, and
flexible model, presents the optimal solution set based on the
Pareto front according to the following six steps (see model
(7)
implementation part, SM) [91]: (i) problem statement, (ii) defini-
tion of the optimization objective function, (iii) establishment of
the data structure, (iv) standardization of the optimization objec-
tive function, (v) definition of the fitness function, and (vi) imple-
mentation of the genetic algorithm.
In previous researches, the STE system presents the optimal
solution as follows (see Table S1): (i) heat generation; and (ii)
economic and environmental effect. Thus the objective functions
related to the energy, economic and environmental effect were
considered in this model.
(8) For the optimization objectives, the objective functions were
defined by energy and economic and environmental effect. There-
fore, the final objective functions of this model were as follows: (i)
minimization of the IIC and (ii) maximization of the AHG, NPV, SIR,
and AHG/unit.
To calculate the STE system's optimal solution set, the regional
information, and the information on the types of collectors and
storage, were collected. For the regional information, the ambient
temperatures, radiation levels, and latitudes of the 16 regions in
South Korea were collected (see Tables S3eS7, SM). From the Public
Procurement Service, 26 types of collectors and 27 types of storage
information were collected (see Tables S8 and S9, SM). The pro-
posed objective functions were standardized to the two extreme
points -the maximum extreme point (Zþ) and the minimum
extreme point (Z)- using the SM Eqs. (S3) and (S4) (i.e., the IIC,
AHG, NPV, SIR, and AHG/unit) (see Table S11, SM).
To standardize the optimization objectives, the objective func-
tions were standardized. The optimization objectives were mini-
mization of the IIC and maximization of the AHG, NPV, SIR, and
AHG/unit. The STE system's optimal solution, which is the scenario
with the lowest iMOO score, was calculated using the fitness
function [refer to Equation (10)]:
(10)
where, SA, SB, SC, SD, and SE are the standardized values for the
optimization objectives of the IIC, AHG, NPV, SIR, and AHG/EA,
respectively; and the iMOO score is the value of the fitness function.
Developed in 1970 by J. Holland, GA is an optimization algo-
rithm that plays an important role in the search for solutions using
crossover operators [92,93]. The proposed model used GA based on
the software program Evolver 5.5. To apply GA, the study used SoC,
(9) ToC, ToS, NoC_L, and NoC_W, which are the adjustable parameters
considered in Step 2, as well as the limitations of RL and RW, the
minimum AHG limit, and the maximum budget limit, which are the
constraint parameters.
 J. Kim et al. / Energy 102 (2016) 416e426
2.6. Step 6: systemization of the proposed model using a Microsoft-
Excel-based VBA
As mentioned earlier, the study used Microsoft-Excel-based VBA
to develop the STE system's iMOO model [94]. Fig. S1 shows the
graphical user interface (GUI) of the iMOO model. The GUI has
seven parts.
 Part (A), the optimization goal: The final decision-maker can
select the target variables that can establish the fitness function
(i.e., the IIC, AHG, NPV, SIR, and AHG/unit) (refer to ‘B’ of Fig. 2
and ‘Part (A)’ of Fig. S1)
 Part (B), the defined parameters: This part sets the region for the
target building and the information on the AoP, and collects data
from the database from Step 1 (refer to ‘B’ of Fig. 2 and ‘Part (B)’
of Fig. S1).
 Part (C), the adjustable parameters: In Step 2, the adjustable
parameters are presented as follows: SoC, ToC, ToS, NoC_L, and
NoC_W. The adjustable parameters of the STE system can differ
according to the characteristics of the target building. Using GA,
the optimal solution for the adjustable parameters is proposed
(refer to ‘B’ of Fig. 2 and ‘Part (C)’ of Fig. S1).
 Part (D), the constraint parameters: In Step 2, the constraint
parameters are presented as follows: the rooftop length, rooftop
width, minimum AHG limit, and maximum budget limit. The
potential installation scenarios of the STE system can be initially
selected based on the constraint parameter (refer to ‘B’ of Fig. 2
and ‘Part (D)’ of Fig. S1).
 Part (E), Standardization: According to Step 5, to set the fitness
function of the iMOO model, the maximum extreme point and
the minimum extreme point of the objective functions are
determined. After the standardization of the objective functions,
the fitness function is calculated using Eq. (10) (refer to ‘Part (E)’
of Fig. S1).
 Part (F), Economic and Environmental Assessment from the Life
Cycle Perspective: To enable intuitive decision-making by the
final decision-maker, the results of the economic and environ-
mental assessment are presented in a graph, which shows the
following: (i) the initial investment cost, (ii) the replacement
cost, (iii) the repair cost, and (iv) the heat generation benefit
(refer to ‘Part (F)’ of Fig. S1).
 Part (G), Trade-off analysis: To enable intuitive decision-making
by the final decision-maker, the relations between NPV, AHG,
and SIR, which are the objective functions, are expressed in a
two-dimensional chart (refer to ‘Part (G)’ of Fig. S1).
3. Model applications
3.1. Model validation
In order to validate the reliability of the proposed model, this
study compared Retscreen program which has been used all over
the world as the NRE system analysis program with the proposed
model. Retscreen program was developed by Natural Resources
Canada's CANMET Energy Diversification Research Laboratory with
over 70 experts from industries, governments and academics [95].
In order to conduct case studies, this study selects ‘A’ company
building, ‘A’ university, ‘G’ elementary school, and ‘S’ high school.
The proposed model was compared with Retscreen program in
terms of the annual energy generation. The CV(RMSE) (coefficient
of variation of the root mean square error) is one of statistical
indices, as a method for determining the similarity of the simula-
tion model [85,86]. As the CV(RMSE) is closer to 0%, the building
energy simulation is closer to that of the another energy simulation
[85,86]. As shown in Fig. 3, the error rate between the simulation-
421
Fig. 3. The error rate between the proposed model and the Retscreen program.
based data of the proposed model and Retscreen program was
presented as minimum 4.07%. ASHRAE guideline 14 presented the
CV(RMSE) as criteria by the similarity between the model [85]. The
margin of error of the CV(RMSE) is plus or minus 20%. The proposed
model met the criteria of the ASHRAE 14, therefore the reliability of
the prosed model was validated.
3.2. Model application
To secure the feasibility of the developed iMOO model, the
following model application was performed. The case study process
for acquiring the optimal solution is as follows: (i) selection of the
target buildings, (ii) setting of the analysis parameters, and (iii)
optimization of the iMOO model and retrieval of the optimal STE
system.
The target buildings were selected by considering the type of
building and the regional factor. In South Korea, based on the RPS
(Renewable Portfolio Standard), public institutions should use NRE
systems for 15% of their total energy consumption in 2015
[13,96e99]. Facilities with the highest energy consumption,
excluding residential buildings, are office facilities and educational
facilities, each of which consumes 15.7% and 14.2% of the total
energy, respectively [100]. Therefore, the case in this study was
selected based on the following criteria: (i) a facility whose energy
consumption ratio is high, (ii) a facility to which RPS is applied, and
(iii) facilities from diverse regions. As shown in Table 1, ‘A’ company
building, ‘A’ university, ‘G’ elementary school, and ‘S’ high school
were selected and analyzed.
Refer to Table 1, the annual heat load of each target building is
used from 74,487 kWh to 1,171,800 kWh. In South Korea, most of
heat loads is distributed through the domestic hot water using the
high efficiency boiler and the STE system. In this proposed model,
the STE system was design to satisfy the demand of the rest of hot
water [17].
The iMOO model was applied by considering the objective
functions (i.e., the IIC, AHG, NPV, SIR, and AHG/unit) from the life
cycle perspective. As shown in Table 2, detailed parameters were
set for the performance of the iMOO model: (i) defined parameters,
(ii) adjustable parameters, and (iii) constraint parameters.
As one of the defined parameters, the regions of each target
building were Seoul, Suwon, Busan, and Asan. Basic information on
each region were collected from the regional geographical infor-
mation (i.e., the latitude, meridian altitude, monthly incident ra-
diation, and monthly diffuse radiation) and the regional
meteorological information (i.e., the monthly ambient temperature
and monthly water supply temperature) (refer to Table 2). As
shown in Table 1, the azimuth of the target buildings is to the south,
422 J. Kim et al. / Energy 102 (2016) 416e426

Table 1
Information on the target buildings.

Target building Region Azimuth Length of roof (m) Width of roof (m) Annual heat load (kWh)

‘A’ company building Seoul 52.7 41.40 12.00 195,733

‘A’ university Suwon 2.25 14.80 23.00 1,119,925
‘G’ elementary school Busan 21 19.00 11.35 74,487
‘S’ high school Asan 0 15.73 40.00 1,171,800

Table 2
Establishment of the optimization process using a genetic algorithm.

Classification Variables Detailed description

Objective functions IIC (Initial Invest Cost) Minimization

AHG (Annual heat Generation) Maximization
NPV (Net Present Value)
SIR (Savings-to-Investment Ratio)
AHG/unit (Annual heat Generation per unit)
Defined parameters Region Refer to Table 1
The AoC (azimuth of the installed collector) Refer to Table 1
Adjustable parameters SoC (Slope of the installed Collector) Dividing the range from 0 to 90 by 1 (South: 0 )
ToC (Type of the Collector) Profile of Solar thermal Collectors
ToS (Type of the Storage) Profile of Solar thermal Storage
Constraint parameter Rooftop area limit (RL and RW) Table 1
Minimum heat generation limit (GL) 15% of ‘G’ elementary school heat consumption (11,173 kWh)
Maximum budget limit 1,000,000 US$
Basic information Latitude Regional value (see Table S3, SM)
Meridian altitude Regional value (see Table S3, SM)
Monthly incident radiation Regional value (see Table S4, SM)
Monthly diffuse radiation Regional value (see Table S5, SM)
Monthly ambient temperature Regional value (see Table S6, SM)
Monthly water supply temperature Regional value (see Table S7, SM)

except that of ‘S’ high school. Therefore, the AHG can be maximized
by setting the AoC to the south.
Since the SoC affects the performance of the STE system, it was
adjusted in the optimization process by considering the regional
factor of each region. The 26 types of ToC and the 27 types of ToS
were collected from the Public Procurement Service (see Tables S8
and S9, SM) and adjusted in the optimization process based on the
regional factor of each region.
In Table 1, the rooftop area limits (RL and RW) of each target
building are presented. From the constraint parameters (i.e., RL, RW,
and GL), the possibility of installation of NoC_L and NoC_W was
determined (see Table S12, SM). According to RPS in 2015, public
institutions should use an NRE system for over 15% of their total
energy consumption. In ‘G’ elementary school, 15% of its energy
requirement should be supplied by an STE system, and therefore,
the minimum heat generation limit (GL) was limited to 15% of the
heat consumption (11,173 kWh). The maximum budget limit was
set at US$1,000,000 for the proposed model. Based on these pa-
rameters, possible installation scenarios were generated.
4. Results and discussion
To generate the possible installation scenarios, the defined
parameters, adjustable parameters, and constraint parameters
were determined. As mentioned, the defined parameters were
determined as shown in Table 2, and the adjustable parameters
were defined within a certain range: (i) the SoC, 0 e90 ; (ii) the
ToC, from No. 1 to No. 26; and (iii) the ToS, from No. 1 to No. 27. As

Table 3
Optimal STE system solution for each target building.
l
Target building Adjustable parameters Objective function iMOO score
a  b c d e f g h i j k 2
SoC ( ) ToC ToS NoC_L NoC_W NoC IIC (US $) AHG (kWh) NPV (US $) SIR AHG/unit (kWh/m )

‘A’ company building 26 No.20 No.13 14 7 98 45,810 145,465 142,151 2.355 1484 0.3391
‘A’ university 27 No.20 No.1 5 16 80 37,800 126,077 121,633 2.405 1576 0.3474
‘G’ elementary school 28 No.20 No.2 7 8 56 36,660 57,705 87,386 2.041 1030 0.4989
‘S’ high school 32 No.20 No.10 5 18 90 43,290 141,775 26,933 1.162 1575 0.4547
a
SoC stands for the slope of the installed collector.
b
ToC stands for the type of the collector.
c
ToS stands for the type of the storage.
d
NoC_L stands for the number of the installed collectors on length side.
e
NoC_W stands for the number of the installed collectors on width side.
f
NoC stands for the number of the installed collectors.
g
IIC stands for the initial investment cost.
h
AHG stands for the annual heat generation.
i
NPV stands for the net present value.
j
SIR stands for the savings-to-investment ratio.
k
AHG/Unit stands for the annual heat generation per unit collector.
l
iMOO score stands for the value of the fitness function.
 J. Kim et al. / Energy 102 (2016) 416e426 423

Table 4
Correlation analysis of the objective function.

Variables IIC AHG NPV SIR AHG/Unit

b
IIC Pearson correlation 1 .739a .351a .300a .479a
Sig (2-tailed) .000 .000 .000 .000
N 4283 4283 4283 4283 4283
c
AHG Pearson correlation .736a 1 .892a .851a .733a
Sig (2-tailed) .000 .000 .000 .000
N 4283 4283 4283 4283 4283
d
NPV Pearson correlation .351a .892a 1 .977a .696a
Sig (2-tailed) .000 .000 .000 .000
N 4283 4283 4283 4283 4283
e
SIR Pearson correlation .300a .851a .977a 1 .646a
Sig (2-tailed) .000 .000 .000 .000
N 4283 4283 4283 4283 4283
f
AHG/Unit Pearson correlation .479a .733a .696a .646a 1
Sig (2-tailed) .000 .000 .000 .000
N 4283 4283 4283 4283 4283
a
Correlation coefficient is significant at 0.01 level (both sides).
b
IIC stands for the initial investment cost.
c
AHG stands for the annual heat generation.
d
NPV stands for the net present value.
e
SIR stands for the savings-to-investment ratio.
f
AHG/Unit stands for the annual heat generation per unit collector.

shown in Table 1, the RL and RW of each target building were

presented, from which the number of possible installation sce-
narios was calculated. Finally, the optimal STE system solution was
selected.
The iMOO model generated a total of 3,513,510e11,498,760
possible installation scenarios, which considered adjustable pa-
rameters and constraint parameters, and calculated the optimal STE
system solution (see Table S12, SM). The computational time for
calculating the optimal solution was only 131 s, using a computer
(Processor, Intel® CoreTM i7-3770 CPU @ 3.40 GHz, and RAM,
15.90 GB available).
From the possible installation scenarios, the optimal iMOO
model solution was proposed (see Tables S13eS16, SM). The most
minimized iMOO score was that of the ‘A’ company building
(0.3391), which showed the following results: SoC, 26 ; ToC, No. 20;
ToS, No. 13; NoC_L, 14 EA; and NoC_W, 7 EA (refer to Table 3).
To determine the high availability of the optimal solution, a
trade-off analysis between the objective functions was conducted
[101]. Before conducting the detailed trade-off analysis, a correla-
tion analysis of the following five objective functions was per-
formed (using the ‘A’ company building as an example) (refer to
Table 4).
The results were as follows: (i) the IIC at the AHG (0.739)
Fig. 4. Surface plot with the interpolated iMOO score between the IIC and the SIR.
showed a very negative correlation, and the NPV (0.351), SIR
(0.300), and AHG/unit (0.479) showed a low negative correlation;
and (ii) the AHG at the NPV (0.892), SIR (0.851), and AHG/unit Fig. 4 shows the trade-off relation between IIC (the x axis) and
(0.733) showed a very high positive correlation. Through the SIR (the y axis), which confirms how the iMOO score between IIC
optimization, the closer the IIC moved to the minimum value, the and SIR changes during the optimization process. To minimize the
closer the other objective functions (AHG, NPV, SIR, and AHG/unit) iMOO score, the following two conditions should be met: (i) the IIC
moved to the maximum value. That is, the IIC was negatively should be minimized and close to 0, and (ii) the SIR should be
correlated with the other objective functions, which shows that maximized and close to 1. Therefore, the optimal solution should
there is a trade-off relation between IIC and the other objective satisfy the values (0, 1).
functions. Thus, to perform the trade-off analysis, only two of the Finally, to more easily explain the results of the multi-objective
five objective functions should be selected. The analysis results of optimization, a correlative analysis of the three objective functions
the IIC and the SIR of the ‘A’ company building were shown as and the iMOO scores was conducted (refer to Fig. 5): (i) the IIC (Part
sample trade-off analysis results in Fig. 4. As shown in Fig. 4 and A in Fig. 5), (ii) the AHG (Part B in Fig. 5), (iii) the NPV (Part C in
Table 4, IIC and SIR are negatively correlated, showing a Pareto- Fig. 5), and (iv) the iMOO score (Part D in Fig. 5). As shown in Fig. 5
front shape. As a result, they satisfied (0.210, 0.820) through the and Table S17, the optimal solution and two alternative scenarios in
optimization, based on which the iMOO score was 0.3391. the ‘A’ company building were analyzed. The final decision-maker
Figs. S2eS10 show the other surface plots with the interpolated can decide on alternatives based on Fig. 5. If he or she considers
iMOO score between two variables, among the other objective more carefully the economic aspects, he or she may prefer the
functions. optimal alternative (the blue line in Fig. 5). However, if the IIC
424 J. Kim et al. / Energy 102 (2016) 416e426
Fig. 5. Comparison chart for intuitive decision-making.
budget is insufficient, he or she may prefer Alternative 1 (the red
line (in web version) in Fig. 5) instead.
First, a total of three alternative scenarios were compared and
analyzed [i.e., the optimal alternative (the blue line), Alternative 1
(the red line), and Alternative 2 (the green line)]. The less the IIC
value was and the greater the NPV and AHG were, the closer the
scenario being considered moved to the optimal solution.
Compared to the other alternatives, the optimal solution was
disadvantageous to the IIC (Part A). However, the NPV (Part B) and
AHG (Part C) were superior to the other two alternatives. Second, as
shown in the results of the trade-off analysis of the two alternative
scenarios, the optimal alternative (blue line) was more disadvan-
tageous to the IIC (Part A) than was Alternative 1 (red line). How-
ever, it was superior to the NPV (Part B) and the AHG (Part C). The
closer the iMOO score is to 0, the more likely it is that the alter-
native will move to the optimal solution. Since the iMOO score of
the optimal alternative (blue line) is closer to 0 than that of Alter-
native 1 (red line), the optimal alternative is believed to be closer to
the optimal solution.
5. Conclusions
In conclusion, this study developed the integrated multi-
objective optimization model for determining the optimal STE
system solution. The proposed model was developed in six steps:
(i) establishment of a database, (ii) generation of the STE system's
installation scenarios, (iii) analysis of the STE system's installation
scenarios, (iv) economic and environmental assessment from the
life cycle perspective, (v) establishment of an integrated multi-
objective optimization process using a genetic algorithm, and (vi)
systemization of the proposed model using a Microsoft-Excel-
based VBA. This model selected five objectives and solved the
multi-objective optimization problem using the iMOO scores. To
guarantee the feasibility of the developed iMOO model, model
applications were conducted on four target buildings, which veri-
fied the high availability and efficiency of the developed model. The
proposed model considers sufficient technical factors for the design
of STE systems such as the region factors and the building charac-
teristics. The iMOO model could enable a potential STE system user
to easily, efficiently, and accurately determine the optimal solution
for the implementation of the STE system in the early design phase.
This could be achieved using a visual chart for intuitive decision-
making. In addition, it could be useful for contractors in competi-
tive bidding processes to analyze the alternatives.
This study had some limitations. First, it used only a GA to
establish the multi-objective optimization process. Thus, in future
studies, more advanced optimization approaches should be used.
Second, as combinations of NRE systems are applied to buildings
nowadays, the iMOO model should be applied to hybrid NRE sys-
tems (i.e., the STE system and the ground source heat pump sys-
tem) and could be extended to any other country or sector in the
global environment.
Acknowledgments
This work was supported by the National Research Foundation
of Korea (NRF) grant funded by the Korea government (MSIP;
Ministry of Science, ICT & Future Planning) (No. NRF-
2015R1A2A1A05001657).
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.energy.2016.02.104.
References
[1] U.S. climate action report 2014. U.S. Department of State; 2014.
[2] Haillot D, Franquet E, Gibout S, Be
d e

carrats JP. Optimization of solar DHW
system including PCM media. Appl Energ 2013;109:470e5. http://dx.doi.org/
10.1016/j.apenergy.2012.09.062.
[3] Abdelhafez EA, Salem AM, Hamdan MA, Al Aboushi AR. Simulation of solar
thermal hybrid heating system using neural artificial network. 2015.
[4] Allouhi A, Jamil A, Kousksou T, El Rhafiki T, Mourad Y, Zeraouli Y. Solar do-
mestic heating water systems in Morocco: an energy analysis. Energ Convers
Manage 2015;92:105e13. http://dx.doi.org/10.1016/
j.enconman.2014.12.055.
[5] Zeng Y, Klabjan D, Arinez J. Distributed solar renewable generation: option
contracts with renewable energy credit uncertainty. Energ Econ 2015;48:
295e305. http://dx.doi.org/10.1016/j.eneco.2014.12.013.
[6] Pillai U. Drivers of cost reduction in solar photovoltaics. Energ Econ 2015;50:
286e93. http://dx.doi.org/10.1016/j.eneco.2015.05.015.
[7] Hirth L. The market value of variable renewables: the effect of solar wind
power variability on their relative price. Energ Econ 2013;38:218e36. http://
dx.doi.org/10.1016/j.eneco.2013.02.004.
[8] Scarpa R, Willis K. Willingness-to-pay for renewable energy: primary and
discretionary choice of British households' for micro-generation technolo-
gies. Energ Econ 2010;32(1):129e36. http://dx.doi.org/10.1016/
j.eneco.2009.06.004.
[9] Bornatico R, Pfeiffer M, Witzig A, Guzzella L. Optimal sizing of a solar thermal
building installation using particle swarm optimization. Energy 2012;41(1):
31e7. http://dx.doi.org/10.1016/j.energy.2011.05.026.
[10] Kim YD, Thu K, Bhatia HK, Bhatia CS, Ng KC. Thermal analysis and perfor-
mance optimization of a solar hot water plant with economic evaluation. Sol
Energy 2012;86(5):1378e95. http://dx.doi.org/10.1016/
j.solener.2012.01.030.
[11] Shariah AM, Lo €f GOG. The optimization of tank-volume-to-collector-area
ratio for a thermosyphon solar water heater. Renew Energ 1996;7(3):
289e300. http://dx.doi.org/10.1016/0960-1481(95)00132-8.
[12] Pillai IR, Banerjee R. Methodology for estimation of potential for solar water
heating in a target area. Sol Energy 2007;81(2):162e72. http://dx.doi.org/
10.1016/j.solener.2006.04.009.
[13] The 2nd energy master plan. Ministry of Trade, Industry and Energy; 2014.
[14] Korea Energy Management Corporation new renewable energy center;
http://www.knrec.or.kr/.
[15] Yearbook of energy statistics. Korea Energy Economics Institute; 2014.
[16] The 4th basic plan for NRE. Ministry of Trade: Industry and Energy; 2014.
[17] Yoo JH. Evaluation of solar hot water heating system applications to high-
rise multi-family housing complex based on three years of system opera-
tion. Energ Build 2015;101:54e63. http://dx.doi.org/10.1016/
j.enbuild.2015.04.037.
 J. Kim et al. / Energy 102 (2016) 416e426
[18] Kulkarni GN, Kedare SB, Bandyopadhyay S. Determination of design space
and optimization of solar water heating systems. Sol Energy 2007;81(8):
958e68. http://dx.doi.org/10.1016/j.solener.2006.12.003.
[19] Cardinale N, Piccininni F, Stefanizzi P. Economic optimization of low-flow
solar domestic hot water plants. Renew Energ 2003;28(12):1899e914.
http://dx.doi.org/10.1016/S0960-1481(03)00070-3.
[20] Farahat S, Sarhaddi F, Ajam H. Exergetic optimization of flat plate solar
collectors. Renew Energ 2009;34(4):1169e74. http://dx.doi.org/10.1016/
j.renene.2008.06.014.
[21] Salazar JPLC, Abreu SL, Borges TP, Colle S, Reguse W. Optimization of a
compact solar domestic hot water system for low-income families with peak
demand and total cost constraints. In: ISESeInternational Solar Energy So-
ciety; Gotemburg, Sweden; 2003. p. 1048e57.
[22] Krause M, Vajen K, Wiese F, Ackermann H. Investigations on optimizing large
solar thermal systems. Sol Energy 2002;73(4):217e25. http://dx.doi.org/
10.1016/S0038-092X(02)00111-1.
[23] Kulkarni GN, Kedare SB, Bandyopadhyay S. Optimization of solar water
heating systems through water replenishment. Energ Convers Manage
2009;50(3):837e46. http://dx.doi.org/10.1016/j.enconman.2008.09.019.
[24] Shariah A, Al-Akhras MA, Al-Omari IA. Optimizing the tilt angle of solar
collectors. Renew Energ 2002;26(4):587e98. http://dx.doi.org/10.1016/
S0960-1481(01)00106-9.
[25] Khalifa AJN, Jabbar RAA. Conventional versus storage domestic solar hot
water systems: a comparative performance study. Energ Convers Manage
2010;51(2):265e70. http://dx.doi.org/10.1016/j.enconman.2009.09.021.
[26] Lamnatou C, Notton G, Chemisana D, Cristofari C. The environmental per-
formance of a building-integrated solar thermal collector, based on multiple
approaches and life-cycle impact assessment methodologies. Build Environ
2015;87:45e58. http://dx.doi.org/10.1016/j.buildenv.2015.01.011.
[27] Hobbi A, Siddiqui K. Optimal design of a forced circulation solar water
heating system for a residential unit in cold climate using TRNSYS. Sol
Energy 2009;83(5):700e14. http://dx.doi.org/10.1016/j.solener.2008.
10.018.
[28] Mazman M, Cabeza LF, Mehling H, Nogues M, Evliya H, HO € Paksoy. Uti-
lization of phase change materials in solar domestic hot water systems.
Renew Energ 2009;34(6):1639e43. http://dx.doi.org/10.1016/
j.renene.2008.10.016.
€ m G, Perers B. Optimization of systems with the com-
[29] Kjellsson E, Hellstro
bination of ground-source heat pump and solar collectors in dwellings. En-
ergy 2010;35(6):2667e73. http://dx.doi.org/10.1016/j.energy.2009.04.011.
[30] Rodríguez-Hidalgo MDC, Rodríguez-Aumente PA, Lecuona A, Legrand M,
Ventas R. Domestic hot water consumption vs. solar thermal energy storage:
the optimum size of the storage tank. Appl Energ 2012;97:897e906. http://
dx.doi.org/10.1016/j.apenergy.2011.12.088.
[31] Szargut J, Stanek W. Thermo-ecological optimization of a solar collector.
Energy 2007;32(4):584e90. http://dx.doi.org/10.1016/j.energy.2006.06.010.
[32] Taherian H, Rezania A, Sadeghi S, Ganji DD. Experimental validation of dy-
namic simulation of the flat plate collector in a closed thermosyphon solar
water heater. Energ Convers Manage 2011;52(1):301e7. http://dx.doi.org/
10.1016/j.enconman.2010.06.063.
[33] Belessiotis V, Mathioulakis E. Analytical approach of thermosyphon solar
domestic hot water system performance. Sol Energy 2002;72(4):307e15.
http://dx.doi.org/10.1016/S0038-092X(02)00011-7.
[34] Jordan U, Vajen K. Influence of the DHW load profile on the fractional
energy savings: a case study of a solar combi-system with TRNSYS sim-
ulations. Sol Energy 2001;69:197e208. http://dx.doi.org/10.1016/S0038-
092X(00)00154-7.
[35] Souliotis M, Kalogirou S, Tripanagnostopoulos Y. Modelling of an ICS solar
water heater using artificial neural networks and TRNSYS. Renew Energ
2009;34(5):1333e9. http://dx.doi.org/10.1016/j.renene.2008.09.007.
[36] Boji
c M, Kalogirou S, Petronijevi

c K. Simulation of a solar domestic water
heating system using a time marching model. Renew Energ 2002;27(3):
441e52. http://dx.doi.org/10.1016/S0960-1481(01)00098-2.
[37] Dagdougui H, Ouammi A, Robba M, Sacile R. Thermal analysis and perfor-
mance optimization of a solar water heater flat plate collector: application to
Te
touan (Morocco). Renew Sust Energ Rev 2011;15(1):630e8. http://
dx.doi.org/10.1016/j.rser.2010.09.010.
[38] Zeghib I, Chaker A. Simulation of a solar domestic water heating system.
Energy Procedia 2011;6:292e301. http://dx.doi.org/10.1016/
j.egypro.2011.05.033.
[39] Shariah A, Dajeh D, Malhi N. Technical note best connection scheme of
collector modules of thermosyphon solar water heater operated at high
temperatures. Renew Energ 1999;17(4):573e86. http://dx.doi.org/10.1016/
S0960-1481(98)00766-6.
[40] Qiu S, Ruth M, Ghosh S. Evacuated tube collectors: a notable driver behind
the solar water heater industry in China. Renew Sust Energ Rev 2015;47:
580e8. http://dx.doi.org/10.1016/j.rser.2015.03.067.
[41] Zambrana-Vasquez D, Aranda-Uso
n A, Zabalza-Bribia
n I, Janez A, Llera-
Sastresa E, Hernandez P, et al. Environmental assessment of domestic solar
hot water systems: a case study in residential and hotel buildings. J Clean
Prod 2015;88:29e42. http://dx.doi.org/10.1016/j.jclepro.2014.06.035.
[42] Mohsenzadeh M, Hosseini R. A photovoltaic/thermal system with a combi-
nation of a booster diffuses reflector and vacuum tube for generation of
electricity and hot water production. Renew Energ 2015;78:245e52. http://
dx.doi.org/10.1016/j.renene.2015.01.010.
425
[43] Koroneos CJ, Nanaki EA. Life cycle environmental impact assessment of a
solar water heater. J Clean Prod 2012;37:154e61. http://dx.doi.org/10.1016/
j.jclepro.2012.07.001.
[44] Duomarco JL. Figure of merit for solar domestic hot water systems. Sol En-
ergy 2015;111:151e6. http://dx.doi.org/10.1016/j.solener.2014.10.023.
[45] Halawa E, Chang KC, Yoshinaga M. Thermal performance evaluation of solar
water heating systems in Australia, Taiwan and Japanea comparative re-
view. Renew Energ 2015;83:1279e86. http://dx.doi.org/10.1016/
j.renene.2015.04.023.
[46] Wang Z, Rosas R, Qian C. Nonlinear observer design for a solar thermal water
heater system. In: Control and Decision Conference (CCDC); Qingdao, Chi-
nese; 2015. p. 4817e22.
[47] Zeghib II, Chaker A. Modeling and simulation of a solar thermal system for
domestic space heating using radiators low temperature. Int J Renew Ener
2015;5(1):266e76.
[48] Andre
s AC, Lo

pez JC. TRNSYS model of a thermosiphon solar domestic water
heater with a horizontal store and mantle heat exchanger. Sol Energy
2002;72(2):89e98. http://dx.doi.org/10.1016/S0038-092X(01)00074-3.
[49] Kalogirou S. Thermal performance, economic and environmental life cycle
analysis of thermosiphon solar water heaters. Sol Energy 2009;83(1):39e48.
http://dx.doi.org/10.1016/j.solener.2008.06.005.
[50] Jing OL, Bashir MJ, Kao JJ. Solar radiation based benefit and cost evaluation
for solar water heater expansion in Malaysia. Renew Sust Energ Rev
2015;48:328e35. http://dx.doi.org/10.1016/j.rser.2015.04.031.
[51] Bessa VM, Prado RT. Reduction of carbon dioxide emissions by solar water
heating systems and passive technologies in social housing. Energ Policy
2015;83:138e50. http://dx.doi.org/10.1016/j.enpol.2015.04.010.
[52] Tanha K, Fung AS, Kumar R. Performance of two domestic solar water heaters
with drain water heat recovery units: simulation and experimental inves-
tigation. Appl Therm Eng 2015;90:444e59. http://dx.doi.org/10.1016/
j.applthermaleng.2015.07.038.
[53] Essabbani T, Moufekkir F, Mezrhab A, Naji H. Numerical computation of
thermal performance of a simulation of a solar domestic hot water system.
Appl Sol Energy 2015;51(1):22e33. http://dx.doi.org/10.3103/
S0003701X15010089.
[54] Kalogirou SA, Papamarcou C. Modelling of a thermosyphon solar water
heating system and simple model validation. Renew Energ 2000;21(3):
471e93. http://dx.doi.org/10.1016/S0960-1481(00)00086-0.
[55] Otanicar TP, Golden JS. Comparative environmental and economic analysis of
conventional and nanofluid solar hot water technologies. Environ Sci Tech-
nol 2009;43(15):6082e7. http://dx.doi.org/10.1021/es900031j.
[56] Koo C, Hong T. A dynamic energy performance curve for evaluating the
historical trends in the energy performance of existing buildings using a
simplified case-based reasoning approach. Energ Build 2015;92:338e50.
http://dx.doi.org/10.1016/j.enbuild.2015.02.004.
[57] Jeong K, Koo C, Hong T. An estimation model for determining the annual
energy cost budget in educational facilities using SARIMA (seasonal autor-
egressive integrated moving average) and ANN (artificial neural network).
Energy 2014;71:71e9. http://dx.doi.org/10.1016/j.energy.2014.04.027.
[58] Koo C, Hong T, Lee M, Park HS. Development of a new energy efficiency
rating system for existing residential buildings. Energ Policy 2014;68:
218e31. http://dx.doi.org/10.1016/j.enpol.2013.12.068.
[59] Park J, Hong T. Analysis of South Korea's economic growth, carbon dioxide
emission, and energy consumption using the Markov switching model.
Renew Sust Energ Rev 2013;18:543e51. http://dx.doi.org/10.1016/
j.rser.2012.11.003.
[60] Hong T, Koo C, Jeong K. A decision support model for reducing electric en-
ergy consumption in elementary school facilities. Appl Energ 2012;95:
253e66. http://dx.doi.org/10.1016/j.apenergy.2012.02.052.
[61] Han S, Hong T, Kim G, Lee S. Technical comparisons of simulation-based
productivity prediction methodologies by means of estimation tools
focusing on conventional earthmovings. J Civ Eng Manag 2011;17(2):
265e77. http://dx.doi.org/10.3846/13923730.2011.574381.
[62] Koo C, Hong T, Hyun C. The development of a construction cost prediction
model with improved prediction capacity using the advanced CBR approach.
Expert Syst Appl 2011;38(7):8597e606. http://dx.doi.org/10.1016/
j.eswa.2011.01.063.
[63] Hong T, Koo C, Park J, Park HS. A GIS (geographic information system)-based
optimization model for estimating the electricity generation of the rooftop
PV (photovoltaic) system. Energy 2014;65:190e9. http://dx.doi.org/10.1016/
j.energy.2013.11.082.
[64] Koo C, Hong T, Lee M, Park HS. Estimation of the monthly average daily solar
radiation using geographic information system and advanced case-based
reasoning. Environ Sci Technol 2013;47(9):4829e39. http://dx.doi.org/
10.1021/es303774a.
[65] Hong T, Koo C, Kwak T. Framework for the implementation of a new
renewable energy system in an educational facility. Appl Energ 2013;103:
539e51. http://dx.doi.org/10.1016/j.apenergy.2012.10.013.
[66] Kim CJ, Hong T, Kim J, Kim D, Seo DY. A process for the implementation of
new renewable energy systems in a building by considering environmental
and economic effect. Sustain 2015;7(9):12870e90. http://dx.doi.org/
10.3390/su70912870.
[67] Google EarthTM; http://www.google.com/intl/ko/earth/.
[68] New renewable energy data center; http://kredc.kier.re.kr/kier/.
[69] Korea Meteorological Administration (KMA); http://www.kma.go.kr/.
426 J. Kim et al. / Energy 102 (2016) 416e426
[70] Public Procurement Service (PPS); http://www.pps.go.kr/kor/index.do.
[71] Koo C, Hong T, Park J. Development of the life-cycle economic and envi-
ronmental assessment model for establishing the optimal implementation
strategy of the rooftop photovoltaic system. 2015. Paper scheduled for
publication in the Technol Econ Dev Eco.
[72] Jo DK, Kang YH. Analysis of maximum solar radiation on inclined surfaces for
the installation of solar thermal systems in Korea using the optimum
installation angle. J Korean Sol Energy Soc 2009;29(2):47e54.
[73] Nijmeh S, Mamlook R. Testing of two models for computing global solar
radiation on tilted surfaces. Renew Energ 2000;20(1):75e81. http://
dx.doi.org/10.1016/S0960-1481(99)00091-9.
[74] Ji C, Hong T, Park HS. Comparative analysis of decision-making methods for
integrating cost and CO2 emissionefocus on building structural designe.
Energ Build 2014;72:186e94. http://dx.doi.org/10.1016/
j.enbuild.2013.12.045.
[75] Hong T, Koo C, Kim H, Park HS. Decision support model for establishing the
optimal energy retrofit strategy for existing multi-family housing complexes.
Energ Policy 2014;66:157e69. http://dx.doi.org/10.1016/j.enpol.2013.10.057.
[76] Koo C, Park S, Hong T, Park HS. An estimation model for the heating and
cooling demand of a residential building with a different envelope design
using the finite element method. Appl Energ 2014;115:205e15. http://
dx.doi.org/10.1016/j.apenergy.2013.11.014.
[77] Hong T, Koo C, Lee S. Benchmarks as a tool for free allocation through
comparison with similar projects: focused on multi-family housing complex.
Appl Energ 2014;114:663e75. http://dx.doi.org/10.1016/j.apenergy.2013.
10.035.
[78] Hong T, Kim J, Lee J, Koo C, Park HS. Assessment of seasonal energy efficiency
strategies of a double skin façade in a monsoon climate region. Energies
2013;6(9):4352e76. http://dx.doi.org/10.3390/en6094352.
[79] Park HS, Kwon B, Shin Y, Kim Y, Hong T, Choi SW. Cost and CO2 emission
optimization of steel reinforced concrete columns in high-rise buildings.
Energies 2013;6(11):5609e24. http://dx.doi.org/10.3390/en6115609.
[80] Hong T, Koo C, Kim H. A decision support model for improving a multi-family
housing complex based on CO2 emission from electricity consumption.
J Environ Manage 2012;112:67e78. http://dx.doi.org/10.1016/
j.jenvman.2012.06.046.
[81] Kim J, Hong T, Koo C. Economic and environmental evaluation model for
selecting the optimum design of green roof systems in elementary schools.
Environ Sci Technol 2012;46(15):8475e83. http://dx.doi.org/10.1021/
es2043855.
[82] Hong T, Ji C, Park H. Integrated model for assessing the cost and CO2 emission
(IMACC) for sustainable structural design in ready-mix concrete. J Environ
Manage 2012;103:1e8. http://dx.doi.org/10.1016/j.jenvman.2012.02.034.
[83] Hong T, Koo C, Park S. A decision support model for improving a multi-family
housing complex based on CO2 emission from gas energy consumption. Build
Environ 2012;52:142e51. http://dx.doi.org/10.1016/j.buildenv.2012.01.001.
[84] Kim B, Lim H, Kim H, Hong T. Determining the value of governmental sub-
sidies for the installation of clean energy systems using real options. J Constr
Eng M 2011;138(3):422e30. http://dx.doi.org/10.1061/(ASCE)CO.1943-
7862.0000443.
[85] Hong T, Kim J, Koo C. LCC and LCCO2 analysis of green roofs in elementary
schools with energy saving measures. Energ Build 2012;45:229e39. http://
dx.doi.org/10.1016/j.enbuild.2011.11.006.
[86] Raftery P, Keane M, Costa A. Calibrating whole building energy models:
detailed case study using hourly measured data. Energ Build 2011;43(12):
3666e79. http://dx.doi.org/10.1016/j.enbuild.2011.09.039.
[87] Hong T, Kim H, Kwak T. Energy-saving techniques for reducing CO2 emis-
sions in elementary schools. J Manage Eng 2011;28:39e50. http://dx.doi.org/
10.1061/(ASCE)ME.1943-5479.0000073.
[88] Hong T, Cho K, Hyun C, Han S. Simulation-based schedule estimation model
for ACS-based core wall construction of high-rise building. J Constr Eng M
2011;137(6):393e402. http://dx.doi.org/10.1061/(ASCE)CO.1943-
7862.0000300.
[89] Park J, Hong T. Maintenance management process for reducing CO2 emission
in shopping mall complexes. Energ Build 2011;43(4):894e904. http://
dx.doi.org/10.1016/j.enbuild.2010.12.010.
[90] Kolhe M, Kolhe S, Joshi JC. Economic viability of stand-alone solar photo-
voltaic system in comparison with diesel-powered system for India. Energ
Econo 2002;24(2):155e65. http://dx.doi.org/10.1016/S0140-9883(01)
00095-0.
[91] Giraudet LG, Guivarch C, Quirion P. Exploring the potential for energy con-
servation in French households through hybrid modeling. Energ Econ
2012;34(2):426e45. http://dx.doi.org/10.1016/j.eneco.2011.07.010.
[92] Koo C, Hong T, Kim S. An integrated multi-objective optimization model for
solving the construction time-cost trade-off problem. J Civ Eng Manag
2015;21(3):323e33. http://dx.doi.org/10.3846/13923730.2013.802733.
[93] Talbi EG. Metaheuristics: from design to implementation. John Wiley & Sons;
2009.
[94] Albright S. VBA for modelers: developing decision support systems with
Microsoft Office Excel. Cengage Learning; 2009.
[95] Leng G. J. RETScreen™ international: a decision support and capacity
building tool for assessing potential renewable energy projects. Ind Environ
2000;23(3):22e3.
[96] Hong T, Kim D, Koo C, Kim J. Framework for establishing the optimal
implementation strategy of a fuel-cell-based combined heat and power
system: focused on multi-family housing complex. Appl Energ 2014;127:
11e24. http://dx.doi.org/10.1016/j.apenergy.2014.04.018.
[97] Lee M, Koo C, Hong T, Park HS. Framework for the mapping of the monthly
average daily solar radiation using an advanced case-based reasoning and a
geostatistical technique. Environ Sci Technol 2014;48(8):4604e12. http://
dx.doi.org/10.1021/es405293u.
[98] Hong T, Koo C, Kwak T, Park HS. An economic and environmental assessment
for selecting the optimum new renewable energy system for educational
facility. Renew Sust Energ Rev 2014;29:286e300. http://dx.doi.org/10.1016/
j.rser.2013.08.061.
[99] Koo C, Hong T, Park HS, Yun G. Framework for the analysis of the potential of
the rooftop photovoltaic system to achieve the net zero energy solar
buildings. Prog Photovoltaics 2014;22(4):462e78. http://dx.doi.org/10.1002/
pip.2448.
[100] Statistics Korea; http://kosis.kr.
[101] Son J, Hong T, Lee S. A mixed (continuousþ discrete) time-cost trade-off
model considering four different relationships with lag time. KSCE J Civ Eng
2013;17(2):281e91. http://dx.doi.org/10.1007/s12205-013-1506-3.