Building and Environment 203 (2021) 108061

Contents lists available at ScienceDirect

Building and Environment

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

Multi-objective optimization of the multi-story residential building with

passive design strategy in South Korea
Yujun Jung a, Yeonsook Heo b, Hoseong Lee a, *
a
Department of Mechanical Engineering, Korea University, 409 Innovation Hall Bldg., Anam-Dong, Sungbuk-Gu, Seoul, Republic of Korea
b
School of Civil, Architectural, and Environmental Engineering, Korea University, 335 Engineering Bldg., Anam-Dong, Sungbuk-Gu, Seoul, Republic of Korea

A R T I C L E I N F O A B S T R A C T

Keywords: In this study, the multi-objective optimization of passive design strategy was conducted for multi-story resi­
Passive design strategy dential buildings in South Korea. The performance of passive design was analyzed in terms of energy, envi­
Sensitivity analysis ronmental, and economic metrics, considering the life cycle performance. Various passive design factors were
Multi-objective optimization
considered to better reflect the actual phenomenon, and a novel simulation modeling method which includes
Uncertainty analysis
Building simplification
building model simplification and meta model was developed for an accurate and fast calculation of optimiza­
Life cycle performance tion. The sensitivity analysis was conducted to decide critical design factors in optimization and showed that the
most influential design factors are airtightness, occupants, and window-to-wall ratio. Throughout optimization,
the energy, environmental impact, and economic feasibility in multi-story residential building can be improved
by 52.7%, 39.5%, and 36.9%, respectively. The uncertainty of optimal solution was analyzed though the un­
certainty analysis. If the utility cost increases and energy system efficiency improves in the future, the optimal
passive design solution will not change significantly in energy and economic feasibility, however it may become
more critical in the environmental impact.

1. Introduction engineers have considered exploiting passive design factors that have
been reported in many studies to enhance building energy performance
Global energy consumption has been rapidly increasing due to [11–14]. However, the simple application of passive design strategies is
population and economic growth. The Energy Information Administra­ no longer sufficient to achieve a sustainable design target. It is essential
tion predicts that global energy consumption will grow by nearly 50% to consider the interactions between all design variables and determine
between 2018 and 2050 [1]. Greenhouse gas (GHG) emissions are also their limitations on building performance to achieve a successful design
increasing as a result of increasing energy consumption. This inevitable process [15–20].
increase in energy consumption and resulting environmental degrada­ In this respect, recent studies agree on three main issues of passive
tion also occur in the building sector. Building-related consumption design strategy: (1) identifying the factors that have the most significant
accounts for over 30% of global energy consumption and over 30% of long-term effect, (2) deriving their optimal design conditions, and (3)
global GHG emissions [2]. Recently, building energy has emerged as a predicting the variation in the optimal solution of passive building due
global focus for sustainability, which requires the improvement of to uncertainty. To address these critical issues, many studies have been
building energy performance. Several initiatives have been developed conducted based on sensitivity analysis, multi-objective optimization,
and implemented to further this goal by evaluating the energy savings and uncertainty analysis. Table 1 shows the passive design studies that
potential of the building sector. Various studies, methods, and political dealt with the multi-objective optimization in terms of energy, envi­
regulations have been suggested to improve building energy perfor­ ronmental, and economic metrics. Sensitivity analysis plays an impor­
mance [3]. The first step in the design of an energy-efficient building is tant role in improving the understanding of the most influential design
to apply passive design strategies before considering mechanical facil­ factors by determining how the design variables or their interactions
ities driven by additional energy [4–7]. Therefore, understanding the affect performance. Most studies on sensitivity analysis focused on
potential benefits of passive design strategies and limits of their appli­ identifying the influential passive design factor [21] by analyzing
cation has become a focus of numerous studies [8–10]. Designers and various passive design strategies in their specific climate conditions and

* Corresponding author.
E-mail address: [email protected] (H. Lee).

https://doi.org/10.1016/j.buildenv.2021.108061
Received 5 April 2021; Received in revised form 19 May 2021; Accepted 15 June 2021
Available online 19 June 2021
0360-1323/© 2021 Elsevier Ltd. All rights reserved.
 Y. Jung et al.
Table 1
Literature review for passive design strategy.
Author Location Typology Method Passive design factor Objective function

Building Thermal Building Glazing WWR Airtightness Solar Internal Energy Environment Economics
insulation inertia orientation property absorptance heat gain

Sun, Yongjun [6] Hong Kong Thermal zone SA ✓ ✓ ✓ ✓ ○ ○

Huang et al. [16] Taiwan Residential SA ✓ ✓ ✓ ✓ ○

Gagnon et al. [20] Canada Office SA ✓ ✓ ✓ ✓ ✓ ✓ ○

Prieto et al. [3] Netherlands Commercial SA ✓ ✓ ✓ ○

Magnier and Canada Residential OPT ✓ ✓ ✓ ○

Haghighat [42]
Asadi et al. [35] Portugal Residential OPT ✓ ✓ ○ ○

Hamdy et al. [32] Netherlands Residential OPT ✓ ○ LCC

Wang et al. [45] Canada Office OPT ✓ ✓ ✓ ✓ ○ ○ ○

Giouri et al. [40] Netherlands Office OPT ✓ ✓ ✓ ✓ ✓ ○

Krarti and USA Office OPT ✓ ✓ ○ LCC

Deneuville [33]
Zhai et al. [41] China Office OPT ✓ ✓ ✓ ○

Rabani et al. [37] Norway Office OPT ✓ ✓ ✓ ○ LCC

Lapisa et al. [38] France Commercial OPT ✓ ✓ ✓ ○ ○

Shadram et al. Subarctic Residential OPT ✓ ✓ ○

[36]
Salata et al. [47] Italy Residential OPT ✓ ✓ ✓ ○ ○ ○
2

Rosso et al. [53] Italy Residential OPT ✓ ✓ ✓ ○ ○ ○

Wang et al. [44] China Railway OPT ✓ ✓ ✓ ✓ ○ ○

station
Ascione et al. [34] Italy Residential OPT ✓ ✓ ✓ ○ ○

Ascione et al. [39] Italy Residential OPT ✓ ✓ ✓ ○ ○

Zhai et al. [41] China Office OPT ✓ ✓ ✓ ○

Rivalin et al. [23] France Office UA ✓ ✓ ✓ ✓ ○

Rodríguez et al. Spain Residential UA ✓ ✓ ✓ ○

[22]
O’Neill and Niu USA Residential UA ✓ ○

[50]
Wang et al. [27] China Residential SA, OPT ✓ ✓ ✓ ✓ ✓ ✓ ○ ○

Chen and Yang Hong Kong Residential SA, OPT ✓ ✓ ○

[28]
Harkouss et al. Lebanon Residential SA, OPT ✓ ✓ ✓ ○ LCC
[54]

Building and Environment 203 (2021) 108061

Gou et al. [29] China Residential SA, OPT ✓ ✓ ✓ ✓ ✓ ✓ ✓ ○

Chen and Yang China Residential SA, OPT ✓ ✓ ✓ ✓ ✓ ✓ ○

[30]
Gabrielli and Italy Various OPT, UA ✓ ✓ ○ LCC
Ruggeri [51] building
Ferrara et al. [52] Italy Residential OPT, UA ✓ ✓ ○ LCC
Current study S. Korea Residential SA, OPT, ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ○ LCA LCC
UA
Y. Jung et al. Building and Environment 203 (2021) 108061

building styles. Building insulation and glazing were the most optimization is conducted to elucidate the best performance of the
commonly considered factors, and other design factors such as building passive design strategy and its design conditions. The uncertainty
orientation [3], window-to-wall ratio (WWR) [6], shading [16], thermal analysis is conducted to predict uncertainty in the passive optimal so­
inertia [20], infiltration [22], and internal heat gain [23] were selec­ lution which can be occurred by a future uncertainty, such as changes in
tively analyzed considering their regional conditions. The sensitivity utility cost or energy system performance. These analyses yield optimal
analysis of passive design strategies is primarily conducted in terms of passive design building solutions and provide integrated findings that
buildings’ energetic performances, and other objective functions such as consider the relation between the input and output of the passive design
environmental impact and economics are not analyzed. Sensitivity in terms of energy, environment, and economics.
analysis is also conducted to reduce the number of input variables
required for optimization [24–26]. During the optimization process, 2. Research approach
which requires long computational times owing to many input variables
and cases, sensitivity analysis is used to reduce computational time by Fig. 1 illustrates the framework of this study, which was divided into
reducing the number of input variables and screening only important five stages: building model simplification, passive design strategy
factors from a large number of initial input variables [27–31]. The determination, sensitivity analysis, multi-objective optimization, and
multi-objective optimization process is conducted to successfully deploy uncertainty analysis. In step 1, the three-dimensional building model
the best performance achievable from the passive design strategies at the was simplified to reduce the computational time required for sensitivity
first opportunity [32–36]. In real-world building design applications, analysis and optimization. In step 2, passive design strategies used as
designers often have to deal with conflicting design criteria simulta­ input variables of the sensitivity analysis were determined through an
neously, such as minimum energy consumption and minimum con­ extensive literature review. The sampling range of the selected 22 input
struction cost [32]. Optimization has been utilized as an efficient design variables was designated based on extant standards and passive house
approach to satisfy these conflicting design criteria in designing standards. In step 3, the sensitivity analysis was conducted to reduce the
high-performance buildings. Optimization is generally applied to number of variables considered in the optimization process. In step 4,
building design aspects such as the building envelope, systems, and optimization was conducted based on 12 dominant variables selected in
control and management; however, most optimization strategies focus the sensitivity analysis. The optimization results are presented as the
on improving the building envelope [37]. Building envelope insulation Pareto optimal solution in terms of energy, environmental, and eco­
and glazing properties are frequently considered as optimization input nomic metrics. The uncertainty analysis aims to evaluate the variability
variables, and other design variables such as solar absorptance [38,39], in the optimal solution in terms of future uncertainty, such as a utility
building orientation [40,41], airtightness [42,43], and WWR [41,44] cost increase and energy system efficiency improvement.
are often selected and analyzed according to the intention of the study. The building simulation using 3D models requires huge computation
Moreover, building optimization performance is mainly analyzed from time owing to its modeling complexity and numerous design variables.
an energy perspective, and an extensive analysis considering environ­ For this reason, the optimization process which requires many simula­
mental impact, economics, and thermal comfort has been conducted tions faces an obstacle of long computational time. Therefore, a novel
only in a few studies [45–47]. Sensitivity analysis and multi-objective simulation modeling method, which includes a building model simpli­
optimization are advanced approaches that determine the key factors fication and meta model development, was developed to reduce
in passive design and derive optimal design conditions; however, the computational time while maintaining the modeling accuracy. For the
obtained results may vary depending on the uncertainty of the input simplification, a 19-story high-rise residential building, approximately
variables used because the input variables are not deterministic and 83 m2 per household, was selected as a building model, considering the
have stochastic characteristics [48]. Therefore, it is important to identify market share [55]. The building was 3D modeled in Sketchup according
the uncertainty of passive building strategies due to uncertainty in input to the standard building floor [55]. The building orientation is such that
variables. Uncertainty analysis focus on quantifying uncertainty due to the façade of the building faces south. The conditioned zones were
the inherent simplifications and lack of detailed input information and modeled with 2.5 m heights. The floor is consisted of bedrooms, rest­
predicting how uncertain inputs influence the output of the model [49]. rooms, living room, kitchen, balcony, and evacuation shelter. Then, its
In many studies, various sources of uncertainty have been examined, properties such as insulation layer, window properties, and solar
and most of them focused on aleatory uncertainty such as occupancy absorptance were assigned in the TRNbuild [56]. The properties used to
[50], weather [22], and various economic parameters [51]. Evaluating the building construction are shown in Table 2. The building structure is
the results of uncertainty analysis is typically focused on energy con­ composed of concrete, extruded polystyrene (XPS), plasterboard,
sumption and economics, and environmental impacts are rarely cement mortar, and concrete slab. The windows are modeled without
addressed directly [52]. any specific shading, which has a U-value of 1.69 (W m− 2 K− 1) and
From the literature review, it is confirmed that the building perfor­ G-value of 0.66 (%/100). In addition, the solar absorptance of the
mance with passive designs has been analyzed through various building were set according to the TRNSYS internal standard docu­
advanced techniques. However, most studies have been suffered a large mentation [57].
amount of time spent on optimization process with a limited range of The building schedule was modeled by considering the occupants,
passive strategy. Furthermore, simulation process becomes more electric device, and lighting patterns [58]. The occupants in one
complicated with the multi-story residential building, which is the most household were assumed to three-people, and the occupant schedules
common type of residence in South Korea. In this context, a novel are set for weekday and weekend. The lighting time was assumed to be
simulation method is proposed, which mainly includes building model dark until for hours after sunrise and bright until 3 h before sunset, and
simplification and meta model development for accurate and fast the operation time of electric device was separated into intermittent and
calculation of optimization. The building model simplification can constant heat. For the intermittent devices, a TV, electric rice cooker,
reduce the TRNSYS simulation time, and the simplified building model desktop computer, vacuum cleaner and washing machine are consid­
is used for developing the meta model, which is used for the optimiza­ ered. The refrigerators are considered as constant load devices. The in­
tion to reduce the computational time. The newly developed method ternal heat generation from occupants, devices, and lighting were
enables the life cycle-based multi-criteria evaluation. The performance assumed to be uniformly distributed over the entire area. The heat from
of the passive strategy is evaluated in terms of energetic, environmental, the occupants was set to 75 W⋅person− 1 and 95 W⋅person− 1 in sensible
and economic metrics considering life cycle performance. Sensitivity and latent heat, respectively, assuming sitting and eating [59]. The heat
analysis focuses on determining the most influential passive design from the lighting was set according to type of zone, and the intermittent
factor and reducing the number of optimization input parameters. The heat and constant heat from device were set to 50 W and 400 W,

3
Y. Jung et al. Building and Environment 203 (2021) 108061
Fig. 1. Framework of novel multi-criteria evaluation.
4
Y. Jung et al. Building and Environment 203 (2021) 108061

Table 2 developed based on design of experiments (DOE) for the input variables.
Construction set up in model simplification. DOE is a procedure used to plan and define the conditions for per­
Categorization Layer U-value (W⋅m− 2⋅K− 1) Solar absorptance forming controlled experimental trials. It is often used in simulations as
well as in experiments, and its primary aim is to generate sample points
Front Back
to fill the space of the input variable domain [62]. Space-filling of DOE is
External wall Concrete 0.25 0.4 0.5 conducted through the Latin hypercube sampling (LHS) method. LHS is
XPS
plasterboard
the most popular DOE technique in the engineering field with high
External roof Cement mortar 0.124 0.1 0.5 generality, in which the space-filling is improved as the number of
Concrete slab variables increases. It exhibits high accuracy owing to low clustering
XPS and a robust DOE [62]. The meta model was trained using the DOE
plasterboard
generated to represent the passive building model and performed in
Ground floor Concrete 0.295 0.8 0.5
XPS MATLAB using the Kriging method, which is suitable for analyzing
Cement mortar nonlinear functions, also known as the Gaussian process [63].
Adjacent wall plasterboard 0.358 0.4 0.5 The passive design strategies used for optimization were carefully
Concrete selected based on the literature review. Table 3 shows the selected
plasterboard
Adjacent ceiling Concrete slab 0.896 0.1 0.5
passive design factors and their sampling ranges. The boundary of the
XPS sampling range was determined by the extant building and passive
Cement mortar house standard recently announced by the government, including the
standards applicable to the reference building currently under con­
struction. It should be noted that airtightness (x1) includes both venti­
respectively [60].
lation and infiltration [64]. Building orientation is considered by
Fig. 2 shows the geometrical simplification process of the 3D build­
dividing the input variable into each façade of the building. The enve­
ing model. The building model was geometrically simplified by dividing
lope insulation (x2–x6) is adjusted by the U-value of the wall and roof
it into two sections: (1) floor substitutability and (2) zone complexity.
insulation [55,65]. The surface solar absorptance (x7–x11) is determined
Floor substitutability refers to the simplification method based on the
by considering the general color and roughness of building exterior
number of floors. If the number of floors is reduced in the modeling
paint [66], and the window properties (x12–x15) are adjusted by the
process, modeling convenience and computational times can be signif­
U-value and SHGC of the glazing [55,65]. The thermal inertia (x16–x19),
icantly improved compared to modeling a full-sized building.
the thermal mass inside buildings, is addressed by controlling the in­
Zone complexity refers to simplification according to internal zone
ternal heat capacity inside the envelope insulation (0.03 for internal
modeling methods. The zone complexity was evaluated by integrating
insulation and 30.07 for external insulation) [67,68]. The sampling
the internal zone, as shown in Fig. 3. In the full-sized model, each zone
range of WWR (x20–x21) was assigned from 10% to 100%, assuming a
was divided into four types: direct air conditioning zone, indirect air
minimum WWR of 10% and full glass building. The occupant scenario
conditioning zone, unconditioned zone, and facility space. The direct air
(x22), which refers to the internal heat gain, electricity and water heating
conditioning zone is a space where the air conditioner was directly
demand of the building, was allocated from 1 to 4 by considering a
installed, including the space where people live and work, such as living
general household in Korea [69].
rooms, bedrooms, and dressing rooms. The indirect air conditioning
The sensitivity analysis was also conducted to reduce the computa­
zone is a space that is indirectly air-conditioned by the direct air con­
tional time of optimization. The sensitivity analysis focuses on reducing
ditioning zone. The temperature in the indirect zone is changed by the
the number of optimization input parameters. A sensitivity analysis was
adjacent direct air conditioning zone, and two restrooms are included
conducted by creating 220 design of experiments (DoE) for the selected
here. The unconditioned zone is a space without air conditioning, such
22 input variables. The LHS method generates 220 samples from the 22
as a balcony, evacuation shelter, staircase, corridor, and elevator. The
input variables according to the suitable criteria that suggest a sample
facility space is a space where the necessary facilities for residential
size of 1.5–10 times the number of input variables [70]. These 220
convenience are installed. It includes the space of the air duct (AD), pipe
samples were simulated using TRNSYS–Excel VBA automation. The
duct (PD), and smoke tower (ST). Based on the zone type, the full-sized
calculated results were used to train an artificial neural network (ANN)
building is simplified into four cases: case 1, no facility zone; case 2,
model. Of 220 samples, 90% (198 samples) were used to generate the
direct zone integrated model; case 3, indirect zone integrated model;
ANN model, and 10% were used to validate the ANN model. The ac­
and case 4, single zone. The doors in the unconditioned zone are
curacy of the ANN model is affected by the number of hidden neurons. It
assumed to be closed (C), and other doors are set to open (O). The
is therefore important to determine the appropriate number of hidden
coupling air flow is assumed to be 0.1 ft‧s− 1 considering the typical
neurons, which can be obtained by the trial-and-error method, evalu­
building condition [61].
ating the R-squared and root-mean-square error (RSME), as shown in
In addition to the building model simplification, the meta model was
Eqs. (1) and (2). Fig. 4 shows the validation results of the ANN model

Fig. 2. Geometric simplification process of 3D building model.

5
Y. Jung et al. Building and Environment 203 (2021) 108061

Fig. 3. Zone complexity simplification.

Table 3
Sampling range of input variables.
Categories Description for variables Variable names Probability density functions Reference values Sampling ranges Ref.

Airtightness Air mass flow [ACH] x1 Continuous 2 {0.5, 2} [64]

Envelope insulation Roof U-value [W‧m− 2‧K− 1] x2 Continuous 0.23 {0.15, 0.41} [55,65]
North wall U-value [W‧m− 2‧K− 1] x3 Continuous 0.36 {0.15, 0.76} [55,65]
South wall U-value [W‧m− 2‧K− 1] x4 Continuous 0.36 {0.15, 0.76} [55,65]
East wall U-value [W‧m− 2‧K− 1] x5 Continuous 0.36 {0.15, 0.76} [55,65]
West wall U-value [W‧m− 2‧K− 1] x6 Continuous 0.36 {0.15, 0.76} [55,65]
Surface solar absorptance Roof solar absorptance [− ] x7 Continuous 0.6 {0.2, 0.8} [66]
North wall solar absorptance [− ] x8 Continuous 0.6 {0.2, 0.8} [66]
South wall solar absorptance [− ] x9 Continuous 0.6 {0.2, 0.8} [66]
East wall solar absorptance [− ] x10 Continuous 0.6 {0.2, 0.8} [66]
West wall solar absorptance [− ] x11 Continuous 0.6 {0.2, 0.8} [66]
Window U-value North window U-value [W‧m− 2‧K− 1] x12 Continuous 2.1 {0.9, 4.2} [55,65]
South window U-value [W‧m− 2‧K− 1] x13 Continuous 2.1 {0.9, 4.2} [55,65]
Window SHGC North window SHGC [− ] x14 Continuous 0.5 {0.466, 0.87} [55,65]
South window SHGC [− ] x15 Continuous 0.5 {0.466, 0.87} [55,65]
Thermal inertia North wall capacity ratio [− ] x16 Continuous 0.03 {0.033, 30.08} [67,68]
South wall capacity ratio [− ] x17 Continuous 0.03 {0.033, 30.08} [67,68]
East wall capacity ratio [− ] x18 Continuous 0.03 {0.033, 30.08} [67,68]
West wall capacity ratio [− ] x19 Continuous 0.03 {0.033, 30.08} [67,68]
Window-to-wall ratio (WWR) North WWR [%] x20 Continuous 40 {10, 100} –
South WWR [%] x21 Continuous 40 {10, 100} –
Occupant Number of occupants [− ] x22 Discrete 3 {1, 2, 3, 4} [69]

according to the number of hidden neurons. The ANN model showed the nonlinear models and yields high accuracy. The two main sensitivity
highest accuracy with an R2 of 0.9998% and RSME 5531 MJ when the measures used in this approach are the first-order index and total order
number of hidden neurons was 20. Therefore, the SA was conducted index. In Eq. (3), the first-order Sobol index, Sn, calculates the impact of
based on the ANN model, which consists of one input layer with 22 input the input parameter xn by estimating the partial variance of Y explained
variables and one hidden layer with 20 neurons. by this parameter as shown in Eq. (4). It estimates how much the vari­
∑ ance of the response is reduced, on average, when the parameter xn is
(y − f )2 ( )
fixed; it measures the contribution of the parameter xn to the total
R2 = 1 − ∑ , 0 ≤ R2 ≤ 1 (1)
(y − y)2 variance of the response. The total order effect, STn, considers the total
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ contributions to the output variance, including the interaction between
RMSE =
1∑
(y − f )2 (2) variables due to the higher-order effects. The total effect index repre­
n sents the total contribution, including the interaction of a parameter xn
to the response variance. It was obtained by summing all first-order and
where the original value (y) in [MJ], the fitted value (f) in [MJ], the higher-order effects involving the parameter xn. Therefore, the differ­
original mean value (y) in [MJ], and the number of value (n) in [− ]. ence between the first and the total orders demonstrates the effects of
The sensitivity measures of the input variables to the output were interaction in the input variables [24]. In this study, SA was conducted
conducted using a Sobol’s method. Sobol’s method is an algorithm for using the total effect index to exclude non-dominant factors and deter­
global SA, which is a variance-based method suitable for complex mine influential factors.

6
Y. Jung et al. Building and Environment 203 (2021) 108061

Fig. 4. Trial-and-error method for determining the number of hidden neurons.

STn = Sn + Sn,n′ + ⋯ + Sn,n′ …n˝ (3) passive building design in terms of distance to the true Pareto front and
the spread of optimal points [32,54]. The parameters and settings for
[ ]
VXn EX∼n (Y|Xn ) optimization are listed in Table 4 [63]. The optimization was terminated
Sn = (4) when the geometric average relative variation in the spread of Pareto
V(Y) frontier is less than 1.0E–04. If this criterion is not satisfied, the opti­
mization will continue until 1200 times. The optimization results are
where the total effect index (STn ) in [− ], the first order index (Sn ) in [− ],
represented as a Pareto approach. The Pareto method can provide a
the variance (V) in [− ], and the expected value (E) in [− ].
non-dominated optimization solution that allows for the selection of a
The multi-objective optimization is conducted based on the meta-
solution based on preferences and is superior to the conventional
model with non-dominated sorting genetic algorithm (NSGA– II) sug­
approach by assigning different weight factors to convert a
gested by Deb et al. [71]. For meta-modeling, a 4000 DOE was gener­
single-objective problem using linear scalarization [27]. Multi-criteria
ated, and the meta model was trained using 3600 DOE (90% of 4000
decision making (MCDM) is used to designate a pareto-optimal solu­
DoE) to represent the passive building model. The validation results of
tion at the Pareto front. The technique for order of preference by simi­
meta model are shown in Fig. 5. The validation was conducted for 400
larity to ideal solution (TOPSIS) was selected as the MCDM analysis,
DOE (10% of 4000 DoE), and the energy, environment, and economic
which provides an ideal solution by considering the weight of solutions
models were validated within the maximum error of 1.1%, 0.54%, and
and selecting Pareto front points with a minimum distance from the
1.53%, respectively. The NSGA–II is considered suitable for optimizing

Fig. 5. Validation results of meta model for optimization.

7
Y. Jung et al. Building and Environment 203 (2021) 108061

− 1
Table 4 in simplified model (Dsimp
h,peak ) in [MJ⋅h ], the cooling peak demand in
Parameters and search space for optimization.
full-sized model (Dc,peak ) in [MJ⋅h− 1], and the cooling peak demand in
full
Parameter Value
− 1
simplified model (Dsimp
c,peak ) in [MJ⋅h ].
Number of variables 12
Population size 200 The performance of the passive residential building is analyzed in
Selection function Tournament terms of the life cycle for three objectives: building energy demand
Tournament size 2 (BED), environmental impact, and economic feasibility. Fig. 6 shows the
Reproduction fraction 0.8
life cycle stages of the buildings suggested by the European Committee
Mutation function Uniform
Mutation probability (%) 0.01 for Standardization [73]. The life cycle of a building is divided into five
Crossover function Intermediate stages: product, construction, use, end-of-life, and recycling. The
Crossover probability (%) 1 building energy demand is calculated by considering the use stage, and
Migration direction Both the environmental and economic factors are evaluated by considering all
Migration fraction 0.2
Migration interval 20
stages of production, construction, use, and end-of-life processes,
Maximum number of generations 1200 excluding recycling.
Stall generations 100 The energy performance of the building due to the application of the
Function tolerance 1.E− 04 passive design is evaluated as the energy demand of the building. The
energy demand is calculated by considering the B6 module: operational
energy use in the building use stage. The building energy demand in­
origin [72].
cludes heating, cooling, electricity, and water heating demands. The
energy required to heat and cool the building is supplied by the air
3. Evaluation criteria
source heat pump, and the water heating and electricity demand are
supplied using the gas-fired boiler and grid power, respectively [55].
The availability of the building simplification is analyzed in terms of
The demands of electricity include lights, electric devices [74].
the model accuracy and computational time. The model accuracy is
The environmental impact was analyzed based on life cycle assess­
compared by error of the heating, cooling, peak heating, and peak
ment (LCA) [75]. The environmental impact of passive buildings is
cooling demand of the full-sized model and simplified model, as seen in
evaluated in terms of the global warming potential (GWP), acidification
Eqs. (5–8). The computational time (CT) is automatically measured in
potential (AP), ozone depletion potential (ODP), photochemical ozone
the TRNSYS.
creation potential (POCP), and eutrophication potential (EP) [12,
Dfull
h − Dh
simp
76–78]. The GWP represents the potential of heat absorbed by any
Eh = × 100 (%) (5)
full
Dh greenhouse gas in the atmosphere, which is an indicator of how much
other greenhouse gases contribute to global warming based on the
Dfull
c − Dc
simp impact of carbon dioxide on global warming. The high GWP can lead to
Ec = × 100 (%) (6) droughts, losses of polar ice caps, sea-level rise, and changes in agri­
Dfull
c
cultural production. The ODP indicates the potential for reducing the
Dfull simp
h,peak − Dh,peak
ozone layer in the stratosphere, which refers to the relative effects of
Eh,peak = × 100 (%) (7) substances that affect ozone depletion when the ozone layer depletion
Dfull
effect of CFC–11 is equal to 1. The destruction of the ozone layer in­
h,peak

creases the number of harmful UV rays that have significant impacts on

Dfull simp
c,peak − Dc,peak
Ec,peak = × 100 (%) (8) human health, crops, and ecosystems. AP refers to the potential of
Dfull
c,peak acidifying pollutants to form H+ ions, which is harmful when humans
are subject to long-term air pollutant exposure from metal corrosion and
where the heating demand error (Eh ) in [− ], the cooling demand error combustion processes [78]. AP can occur from sulfur dioxide (SO2),
(Ec ) in [− ], the heating peak demand error (Eh,peak ) in [− ], the heating nitrogen oxides (NOx), nitrogen monoxide (NO), nitrogen dioxide (N2O),
peak demand error (Ec,peak ) in [− ], the heating demand in full-sized and other substances, which are usually characterized by SO2-equiva­
full
building (Dh ) in [MJ], the heating demand in simplified model lence [79]. POCP, known as ground-level smog, refers to the potential
(Dsimp
full for volatile organic compounds to participate in photochemical re­
h ) in [MJ], the cooling demand in full-sized model (Dc ) in [MJ],
actions in the atmosphere, leading to ozone pollution. It is formed in the
the cooling demand in simplified model (Dsimp
c ), the heating peak de­
troposphere with volatile organic compounds in the presence of high
mand in full-sized model (Dh,peak ) in [MJ⋅h− 1], the heating peak demand
full
temperature and sunlight, which has a negative impact on human health

Fig. 6. Life cycle stages of the buildings.

8
Y. Jung et al. Building and Environment 203 (2021) 108061

and the environment. EP refers to a phenomenon in which nutrients such Table 6

as phosphorus (P) are oversupplied to water owing to the inflow of Construction waste of the building in end-of-life stage [81].
chemical ingredients or sewage. The EP is mainly released from the Material Unit Recycle Ratio Incineration Ratio Landfill Ratio
production process of concrete and removes dissolved oxygen from (%) (%) (%)
water, causing the rapid growth of plankton and killing organisms [76]. Concrete kg 100.0 0.0 0.0
The LCA in buildings was conducted for (1) production and construction XPS kg 46.7 53.3 0.0
stage (A1–A4), (2) use stage (B2, B4, B6, B7), and (3) disposal stage (C2, Plaster board kg 62.7 0.2 37.1
C4). The LCA calculation process is expressed in Eqs. 9–11. LCA analysis Cement_mor kg 100.0 0.0 0.0
concrete slab kg 100.0 0.0 0.0
is mainly based on the Korea Life Cycle Inventory (LCI) database, which Window kg 79.0 0.0 21.0
compiles the national fundamental data required to calculate environ­
mental impact [80]. Regarding electricity consumption, the coefficients
of AP, ODP, POCP, and EP were assumed to be post-treated in the power environmental impact (of category x) of transportation b (EIbV ) in
generation process. The service life of the building is set to 50 years, and [kgCO2eq⋅m− 1], the environmental impact of category x resulting from
replacement cycle and the maintenance cycle are assumed to 20 and 10 operation phase (EIxO ) in [kgCO2eq], the number of replacements during
years, respectively [81]. The distance from supplier to construction are use phase based on materials’ durability (Nr ) in [− ], the consumption of
set by considering the location of industrial factories near the Seoul. The utility b during use phase per year (Qb ) in [kJ], the environmental
LCA evaluation criteria used are listed in Tables 5 and 6. impact (of category x) of utility b (EIb ) in [kgCO2eq⋅kJ− 1], the period of

) ( ) ( )
∑
i j
∑
1 Production and construction stage : EIxMC = QM
a × EI M
a + D V
b × n V
× EI V
b (9)
a=1 b=1

analysis (n) in [yr], the environmental impact of category x resulting

from end-of-life phase (EIxEOL ) in [kgCO2eq], the distance from the de­
)
∑
l
( )
∑
o
( ) molition site to waste facilities (DVa ) in [m], the number of travels be­
2 Use stage : EIxO = M M
Qa × Nr × EIa + Qb × EIb × n (10) tween the demolition site and waste treatment facilities (Nt ) in [− ], the
a=1 b=1 environmental impact (of category x) of transportation a (EIaV ) in
[kgCO2eq⋅m− 1], the quantity of material to be treated and to be disposed
(QM M
b ) in [kg], the percentage of materials to be treated (Q ) in [− ], and

) ( )
p
∑ ∑
r
([ M ( )] )
3 End − of − life stage : EIxEOL = V V
Da × Nt × EIa + Qb × 1 − QM × EIcD (11)
a=1 c=1

the environmental impact (of category x) owing to the disposal of ma­

where the environmental impact of category x resulting from terial c (EIcD ) in [kgCO2eq⋅kg− 1].
manufacturing and construction phase (EIxMC ) in [kgCO2eq], the quantity The economic performance of the passive building design is analyzed
of material a (QM a ) in [kg], the environmental impact (of category x) of
based on life cycle cost (LCC), an evaluation method that determines the
material a (EIaM ) in [kgCO2eq⋅kg− 1], the distance from supplier to con­ feasibility of a project by considering expenditure and profits over the
struction (DVb ) in [m], the number of travels (nV ) in [− ], the entire project period. The LCC analysis was conducted for the initial
investment (A1–A5), maintenance and operation (B2, B4, B6, B7), and

Table 5
Life cycle assessment criteria: production, construction, transportation, and use stages.
Stage Material Unit GWP AP ODP POCP EP Ref.

[kg CO2eq/unit] [kg SO2eq/unit] [kg CFC-11eq/unit] [kg Ethyleneeq/unit] [kg PO3−
4 eq/unit]

Production and construction Concrete kg 1.23E-01 1.57E-04 4.71E-09 1.31E-05 2.27E-05 [82]
XPS kg 2.06E+00 4.05E-02 2.89E-08 6.39E-03 2.75E-03 [82]
Plaster board kg 1.35E-01 1.87E-04 7.61E-13 1.42E-08 6.03E-08 [80]
Cement_mor kg 1.05E+00 3.03E-04 3.13E-11 5.36E-10 1.77E-07 [80]
Concrete slab kg 1.23E-01 1.57E-04 4.71E-09 1.31E-05 2.27E-05 [82]
Window kg 7.51E-01 3.62E-03 5.06E-10 2.37E-07 1.13E-06 [80]
Paint kg 3.09E-01 6.37E-04 8.73E-10 2.62E-07 1.96E-06 [80]
Rebar kg 3.40E-01 4.42E-04 4.02E-13 1.04E-08 4.22E-08 [80]
Transportation Truck kg∙km 2.44E-04 1.48E-07 4.15E-15 9.29E-11 3.53E-10 [80]
Use Electricity kWh 4.87E-01 – – – 3.26E-09 [80]
Gas kg 5.36E-01 5.75E-03 7.15E-12 5.54E-04 7.45E-07 [80]
Water kg 1.96E-04 1.38E-07 4.50E-15 1.76E-12 7.14E-12 [80]

9
Y. Jung et al. Building and Environment 203 (2021) 108061

Table 7
Life cycle cost evaluation criteria.
Stage Variable Value

Material Labor

Product and construction Temporary enclosure ($‧m− 1) – 26.0

Protective film ($‧m− 2) 0.2 1.8
System scaffolding ($‧m− 2) 0.1 7.3
Steel pipe support ($‧m− 2) 4.5 5.8
Contractor’s office ($‧m− 2) 22.9 51.6
Container-type temporary building ($‧m− 2) 517.6 177.2
Tower crane rent ($‧nr− 1) 6574.3 3540.0
On-site clean up ($‧m− 2) – 7.6
Plywood formwork ($‧month− 1) 9.5 16.1
Rebar processing and assembly ($‧m− 2) 14.8 281.7
Concrete curing ($‧ton− 1) – 0.3
Plain concrete pouring/pump car ($‧m− 3) 3.4 6.6
Reinforced concrete pouring/pump car ($‧m− 3) 3.3 7.2
Trowel finish ($‧m− 3) – 3.8
Concrete face treatment ($‧m− 2) 0.1 2.2
Mortar depositing ($‧m− 2) – 201.4
Cement mortar mixing ($‧m− 3) 1.5 47.8
Cement mortar plastering (wall) ($‧m− 2) – 4.8
Cement mortar plastering (floor) ($‧m− 2) – 4.9
Cement mortar plastering (ceil) ($‧m− 2) – 18.3
Cement paste plastering ($‧m− 2) 0.2 5.5
Lightweight floor panel ($‧m− 2) 3.7 7.8
Single-glazed window ($‧m− 2) 28.7 11.1
Water-based paint work ($‧m− 2) 0.7 4.0
Acid-resistant epoxy paint ($‧m− 2) 3.4 4.7
Plaster board ($‧m− 2) 1.6 11.0
Extrude polystyrene ($‧m− 2) 9.2 6.2
Use Water-based paint work ($‧m− 2) 0.7 4.0
Plaster board ($‧m− 2) 1.6 11.0
Electricity rate ($⋅kWh− 1) Summer: Base: 0.759 ($), operating: 0.078 (λ ≤ 300 kWh)
- Residential low voltage Base: 1.334 ($), Operating: 0.157 (300 kWh < λ ≤ 450 kWh)
Base: 6.086 ($), operating: 0.234 (450 kWh < λ)
Other seasons: Base: 0.759 ($), operating: 0.078 (λ ≤ 200 kWh)
Base: 1.334 ($), operating: 0.157 (200 kWh< λ ≤ 400 kWh)
Base: 6.086 ($), operating: 0.234 (400 kWh < λ)
Natural gas rate ($⋅MJ− 1) Base: 0.83 ($)
Operating: 0.01
surtax: 1.1%
Water consumption rate ($‧m− 3) Base: 0.90 ($)
Water supply: 0.30
Sewer: 0.33
Water use charge: 0.14
End-of-life Plain concrete ($‧m− 3) 100.0
Reinforced concrete ($‧m− 3) 208.4
waterproofing layer ($‧m− 2) 11.7
Radiant heating floor ($‧m− 2) 4.6
Ceiling ($‧m− 2) 2.4
Door and framework ($‧place− 1) 1.5
Tile on the wall ($‧m− 2) 5.0
Tile on the floor ($‧m− 2) 5.8
Roof truss ($‧m− 2) 3.0
Assumption Discount rate (%) 5
Service life of building (yr) 50
Maintenance cycle (yr) 10
Replacement cycle (yr) 20
Construction period (month) 24

*1$ = 1199.5 Won, summer: July to August; other seasons: January to June and September to December.

disposal (C1, C2, C4) stages. The cost for initial investment was calcu­ government-announced standard market unit prices [83], and the con­
lated using the costs of building materials, procurement, transport, and struction period was assumed to be 24 months. Regarding operating
construction, as shown in Eq. (12). The costs for the maintenance and costs, utility costs such as electricity [84], gas [85], and water usage
operation stages were calculated using Eq. (13). The maintenance cost costs [86] were applied. The cost for building demolition was estimated
includes the painting and plastering interior due to the aging of build­ using the standard unit prices set by the Korean demolition company
ings, and the operation stage includes water consumption and energy [87]. Economic feasibility was calculated considering cash flow using
consumption for heating, cooling, electricity, and water heating in the Eq. (15). The cash flow was considered using the capital recovery factor
use stage of buildings. The cost of building disposal considers the (CRF) which is the ratio of a constant annuity to the present value of
disposal cost of the building and the disposal cost of waste matter, as receiving that annuity for a given period.
shown in Eq. (14). Table 7 shows the economic criteria used in the LCC.
The initial investment and maintenance costs were applied to

10
Y. Jung et al. Building and Environment 203 (2021) 108061

Table 8
Zone simplification: model accuracy and computational time.
Classification Model Eh (%) Ec (%) Eh,peak (%) Ec,peak (%) CT (s)

Floor substitutability Single story − 24.76 68.95 − 13.38 39.45 8.81

3-story − 1.75 12.43 − 0.93 5.37 14.69
5-story − 0.18 1.96 − 0.12 1.08 21.74
7-story − 0.03 0.34 − 0.03 0.34 27.20
9-story − 0.01 0.07 − 0.01 0.11 33.51
19-story – – – – 86.89
Zone simplification Case 1 − 0.77 0.83 − 0.42 0.17 12.9
Case 2 − 1.10 12.08 0.73 15.49 9.8
Case 3 − 2.32 13.08 − 0.11 16.48 9.4
Case 4 − 32.24 − 3.91 − 29.88 − 4.24 9.0
Full-sized model – – – – 15.10

∑
i
( ) ∑j
( M ) ∑k
( Con ) consumption of utility c throughout the construction (QCon c ) in [kJ], the
CMC = QM M
a × ACa + Qb × CbC2B + Qc × CcCon + L − 1
cost of utility c (CCon
c ) in [$⋅kJ ], the percentage of wasted materials (L)
a=1 b=1 c=1
j
/ / in [− ], the capital recovery factor (CRF) in [− ], the costs resulting from
∑ ( )
× M M
Qa × ACa LMC
CRF MC (12) operation phase (CO ) in [$], the quantity of material a to be replaced in
b=1 year n during the use phase of the building (QM a, n ) in [kg], the discount
( rate (d) in [− ], the consumption of utility b in year n during operation
∑
l m (
∑ ))
O
C = QM M 1 phase (QOP OP
b, n ) in [kJ], the cost of utility b (Cb ) in [$/kJ], the costs
a, n × ACa × n
(1 + d)
a=1
(
n=1 resulting from end-of-life phase (CEOL ) in [$], the distance from the
( )) / /
∑ ∑
o m
1 demolition site to waste facilities (DVa ) in [m], the number of travels
+ QOP × C OP
× LO CRF O (13)
b=1 n=1
b, n b
(1 + d)n between the demolition site and waste treatment facilities (Nt ) in [− ],
the cost of transportation a (CVa ) in [$⋅m− 1], the last year of the period of
s (
∑ 1
) t (
∑ [ M ] 1
) analysis (ml) in [− ], the percentage of materials to be treated (e.g.
CEOL = DVa × Nt ×CaV × + Qb ×Q ×CbWT × reused, recycled) (Q) in [− ], the costs owing to the waste treatment of
(1 + d)ml (1 + d)ml − 1
material b (CWTb ) in [$⋅kg ], and the costs owing to the disposal of
a=1 b=1

∑u ([ )/
]
+ QM D
b × (1− Q) ×Cb ×
1
CRF EOL material c (CDb ) in [$⋅kg− 1].
c=1 (1 + d)ml
(14) 4. Results

d(1 + d)n 4.1. Building model simplification

CRF = (15)
(1 + d)n − 1
The zone simplification results are shown in Table 8. In the case of
where the costs resulting from manufacturing and construction phase floor substitutability, the full-sized model was compared with the
− 1
(CMC ) in [$], the acquisition cost of material a (ACM
a ) in [$⋅kg ], the cost simplified model of the 3rd, 5th, 7th, and 9th stories. The three-story
− 1
to build/assemble construction elements (CC2B
b ) in [$⋅kg ], the building is modeled in three layers: ground level floor, middle floor,

Fig. 7. Results of sensitivity analysis for passive design strategy.

11
Y. Jung et al. Building and Environment 203 (2021) 108061

Table 9 and roof floor, and the result is derived by multiplying the value of
Boundary conditions for design variable optimization. middle floor and adding to the value of ground and roof floors. The 5th,
Variables Unit Reference Lower Upper 7th, and 9th story buildings consider the building temperature gradient
limit limit more closely. For example, in the case of seventh-story building, the
x1 (Airtightness) [ACH] 2 0.5 2 result considers not only the value of ground floor, middle floor, and roof
x2 (Roof U-value) [W‧m− 2‧ 0.23 0.15 0.41 floor but also the value of upper floor of the ground floor and the lower
K− 1] floor of the roof floor. The accuracy of the model was evaluated in terms
x3 (North wall U-value) [W‧m− 2‧ 0.36 0.15 0.76 of the heating, cooling, heating peak, and cooling peak demands. The
K− 1]
x4 (South wall U-value) [W‧m− 2‧ 0.36 0.15 0.76
average accuracy of the simplified model decreased as the number of
K− 1] floors increased; however, the computational time was conversely
x5 (East wall U-value) [W‧m− 2‧ 0.36 0.15 0.76 improved. As a result, the seven-story simplified model, which reduces
K− 1] the computational time by 68.7% and ensure model accuracy within
x6 (West wall U-value) [W‧m− 2‧ 0.36 0.15 0.76
0.16%, was considered as the appropriate level of simplification for the
K− 1]
x12 (North window U- [W‧m− 2‧ 2.1 0.9 4.2 full-sized building. In the case of zone simplification, the case 1 shows
value) K− 1] the best accuracy on average when comparing with the full-sized model.
x13 (South window U- [W‧m− 2‧ 2.1 0.9 4.2 Case 1 had reduced computational time by 14.6% while the model ac­
value) K− 1] curacy was maintained within 0.83%. Case 1 shows similar heating and
x14 (North window [− ] 0.5 0.466 0.87
SHGC)
cooling demands as the full-sized model because the facility space does
x15 (South window [− ] 0.5 0.466 0.87 not significantly affect building demand prediction. In other cases, the
SHGC) model accuracy decreases gradually as the building model is simplified.
x20 (North WWR) [%] 40 10 100 Case 2 and 3 shows a different trend in model accuracy, particularly
x21 (South WWR) [%] 40 10 100
regarding cooling demand. This difference can be explained by the
coupling air flow due to the integration of the direct air conditioning

Fig. 8. Multi-objective optimization for three objectives.

12
Y. Jung et al. Building and Environment 203 (2021) 108061

Fig. 9. Monthly heating and cooling demand in reference building and optimal solution.

zone. The full-sized model that does not integrate the direct air condi­ 4.2. Multi-objective optimization
tioning zone is considered to have a slow coupling air velocity. In
contrast, case 2 and 3, which integrates the direct air conditioning zone, Multi-objective optimization was performed to determine the
has a relatively fast coupling air velocity. This different coupling air flow optimal passive design conditions and develop a passive design strategy
results in different building demands. Case 4 has significantly higher that yields excellent performance in energy, environmental, and eco­
heating and cooling demands compared with those of other cases. This is nomic metrics. Non-influential variables were excluded through the
because the total air conditioning space significantly increased as the sensitivity analysis to reduce the optimization computational time.
building was integrated into a single zone. Therefore, the most appro­ Fig. 7 shows the results of the sensitivity analysis based on equal weights
priate level of zone complexity that can clearly represent the full-sized of BED, LCA, and LCC. The performance was highly sensitive to
model is considered as case 1: no facility space. As a result, a combi­ airtightness, followed by the number of occupants and WWR. Consid­
nation of the 7-story building and case 1: no facility space, which ering the sensitivity measures, the variables used for the optimization
reduced computational time by 73.3% while maintaining model accu­ are as follows: x1 (airtightness), x2 (roof U-value), x3 (north wall U-
racy within 1.17%, was used for optimization process. value), x4 (South wall U-value), x5 (East wall U-value), x6 (West wall U-
value), x12 (North window U-value), x13 (South window U-value), x14
(north window SHGC), x15 (South window SHGC), x20 (North WWR),
x21 (South WWR). The excluded variables, such as solar absorptance

Fig. 10. Environmental pollutant emissions in reference building and optimal solution.

13
Y. Jung et al. Building and Environment 203 (2021) 108061

and wall capacity ratio, were assumed to be identical to those of the heating demand and increase in cooling demand are mainly due to
reference building, and the number of occupants was assumed to be enhanced airtightness, which prevents heat exchange between indoor
three, considering the typical Korean household [69]. The optimization and outdoor environments. Therefore, cold ambient air in winter cannot
objectives and boundary conditions of the design variables are shown in enter the indoor environment, which reduces the heating demand, and
Eq. (16), and Table 9. the internal heat generated in summer does not release into the external
Minimize: BED, LCA, and LCC environment, increasing the cooling demand. As a result, in the Pareto
optimal solution, the heating demand is drastically reduced and the
Subject to: equal weight for BED, LCA, and LCC (16)
cooling demand is slightly increased, thus decreasing the annual energy
The results of the optimization are shown in Fig. 8. The red dot refers demand for heating and cooling by 76%.
to the Pareto optimal solution, and the white dots refer to the alterna­ Fig. 10 shows the environmental pollutant emissions in the reference
tives. The Pareto optimal solution was derived through the TOPSIS, building and the optimal solution. In the case of the reference building,
which is one method of MCDM. The closer the point is to zero on each environmental pollutants mainly originate from the operational and
axis, the better the solution for that criterion. The optimal solution product stages. GWP, which is emitted more than other types of envi­
shows an annual BED of 1100 GJ, an LCA of 5499 tonCO2e, and an LCC ronmental pollutants, mainly occurs in the operational stage and ac­
of 1609 thousand dollars. This is the optimal solution when BED, LCA, counts for 75% of total emissions. AP and POCP emissions are also
and LCC have equal weight, and the optimal solution can vary mainly emitted from the operational stage, which accounts for the 63%
depending on the weight ratio. If the building energy is further and 69% of total emissions, respectively. ODP and EP emissions are
weighted, BED reduces to 1083 GJ, and the LCA and LCC are 5451 mainly occurred from the product stage, and they account for 99% and
tonCO2e and 1648 thousand dollars, respectively. If the environmental 85% of total emissions, respectively. The emissions of environmental
impact of building is further weighted, the LCA reduces to 5415 ton­ pollutants in the Pareto optimal solution are also mainly released from
CO2e, and the BED and LCC are 1100 GJ and 1759 thousand dollars, the product and operational stages. However, as the passive design is
respectively. In addition, if weighting is given more to economic feasi­ applied, the materials added for building structure cause the increase in
bility, the LCC decreases to 1582 thousand dollars, and the BED and LCA the emissions from the product compared to the that of the reference
increase to 1160 GJ and 5648 tonCO2e, respectively. building. When comparing the reference building to the optimal solu­
The performance of the Pareto optimal solution was compared to tion, the GWP emissions from the product stage is increased by 0.9%
that of the reference building. The reference building with the extant from 1,992,605 kgCO2eq to 2,009,791 kgCO2eq. Similarly, the emissions
building standard has a BED of 2327 GJ, an LCA of 9082 tonCO2e, and an from product stage in AP, ODP, POCP, and EP are increased by 13%,
LCC of 2549 thousand dollars. Compared to the reference building, the 3.5%, 28%, and 15%, respectively. In contrast, the emissions from the
BED, LCA, and LCC were decreased by 52.7%, 39.5%, and 36.9%, operational stage decrease because the energy consumption in building
respectively. Therefore, it is confirmed that the optimal solution can heating and cooling is decreased. The GWP emissions from the opera­
significantly reduce the BED, LCA, and LCC outcomes. The performances tional stage decrease by 53% from 6,885,243 kgCO2eq of the reference
of the Pareto optimal solution and reference building were further building to 3,267,720 kgCO2eq of the optimal solution. Similarly, other
analyzed as follows. Fig. 9 shows the monthly heating and cooling de­ emissions are also reduced by 85% on average. Therefore, when passive
mand of the optimal solution and reference building. In the optimal design building strategies are implemented, environmental pollutant
solution, the demand for heating is drastically decreased. The annual emissions from the building are increased, and emissions from energy
heating demand was decreased by 85% from 1,485,430 MJ to 219,636 consumption are significantly decreased. As a result, the total emissions
MJ, and the heating period was shortened from 10 months (Januar­ of GWP, AP, and POCP are decreased by 39%, 49%, and 50%, respec­
y–June, September–December) to 8 months (January–May, Octo­ tively, and the ODP and EP emissions are increased by 4% and 18%,
ber–December). Conversely, the annual cooling demand is slightly respectively.
increased by 28% from 128,194 MJ to 164,140 MJ. The decrease in The LCC of the reference building and the optimal solution are

Fig. 11. Total expenditure of buildings in life cycle stage.

14
Y. Jung et al. Building and Environment 203 (2021) 108061

Fig. 12. Variation in optimal solution considering future uncertainty.

Table 10
Design conditions of reference building, optimal solution, and uncertainty solution.
Design variables Sampling ranges Reference building Optimal solution Uncertainty solution

x1 (Air tightness) {0.5, 2} 2.000 0.500 0.505

x2 (Roof U-value) {0.15, 0.41} 0.230 0.195 0.220
x3 (North wall U-value) {0.15, 0.76} 0.360 0.210 0.229
x4 (South wall U-value) {0.15, 0.76} 0.360 0.353 0.319
x5 (East wall U-value) {0.15, 0.76} 0.360 0.211 0.260
x6 (West wall U-value) {0.15, 0.76} 0.360 0.206 0.279
x12 (North window U-value) {0.9, 4.2} 2.100 4.079 4.072
x13 (South window U-value) {0.9, 4.2} 2.100 4.171 4.113
x14 (North window SHGC) {0.466, 0.87} 0.503 0.490 0.494
x15 (South window SHGC) {0.466, 0.87} 0.503 0.476 0.478
x20 (North WWR) {10, 100} 40.06 10.14 10.38
x21 (South WWR) {10, 100} 40.06 12.34 12.51

compared in Fig. 11. The LCC is represented by the stages of the con­ The costs for maintenance and replacement in the optimal solution in­
struction (A1–A5), maintenance (B2 and B4), operation (B6 and B7), and crease by 6% from $75,076 to $79,727, compared to that of the refer­
disposal (C1, C2, and C4), which is calculated by considering the dis­ ence building. The cost of maintenance and replacement refers to the
count rate. The initial construction cost accounts for the largest portion cost of painting the external wall and plastering the internal wall. These
of both the reference building and the optimal solution. The initial costs slightly increase as the area of the external wall increases. The costs
construction cost in the optimal solution decreases by 18% from for operational energy and water use are less than the cost for the initial
$1,357,896 to $1,118,698 compared to that of the reference building. construction owing to the discount rate. The costs for operational energy
The windows used in the passive design account for a significant portion and water use were significantly reduced when the reference building
of the initial construction cost. When the reference building is opti­ was optimized as the passive building. Water use, which relies on the
mized, the WWR decreases from 40% to 10%, reducing the area of the occupant’s behavior, does not differ between the reference building and
external window and reducing the initial cost of the window signifi­ optimal solution; however, the cost of operational energy varies signif­
cantly. Therefore, the initial construction cost decreases significantly. icantly. The costs of energy and water consumed were reduced by 59%

15
Y. Jung et al. Building and Environment 203 (2021) 108061

from $1,087,228 to $440,550. This is mainly due to the energy perfor­ adjusted according to the optimal design conditions.
mance enhancement of the building by the application of passive de­
signs. The cost of building disposal is increased by 3% from $28,349 to 5. Conclusions
$29,099 for the optimized building compared to the reference building.
A large portion of the disposal cost mainly occurs during the demolition In this paper, the multi-objective optimization of passive design
process of building structures such as walls, ceilings, and floors. As the strategy was conducted for the multi-story residential building in South
WWR of the building decreases in the optimal solution, the area of the Korea. The novel simulation modeling method which includes building
external wall increases, and the cost of building construction also in­ model simplification and meta model was developed, and the perfor­
creases. As a result, it is confirmed that the total expenditure is signifi­ mance of the passive design was analyzed in terms of energy, environ­
cantly decreased during the building life by 36.9%, when the reference mental, and economic metrics, considering life cycle performance. The
building is optimized as the passive building. results are summarized as follows:

4.3. Uncertainty analysis ● A novel simulation modeling method reduced the computational
time required for optimization. The floor simplification and the zone
Optimization becomes more valuable when the optimal decision can simplification of the building reduced the computational time by
consider uncertainty. Although many studies have examined various 73.3% while maintaining the model accuracy within 1.17%.
sources of uncertainty, such as occupants, weather, and economic pa­ ● The overall performance of the passive design applied building had
rameters, the uncertainties in the optimal solution due to future un­ the highest sensitivity to airtightness, followed by the number of
certainties such as utility costs and heating, ventilation, and air- occupants and window to wall ratio.
conditioning (HVAC) system performance have not yet been analyzed. ● The building energy demand, life cycle assessment, and life cycle cost
The uncertainty analysis therefore analyzes potential variations in the in the optimal solution were decreased by 52.7%, 39.5%, and 36.9%,
optimal solution considering future uncertainties: (1) future increases in respectively, compared to that of the reference building.
utility costs and (2) efficiency improvements in energy system (air ● The annual heating demand was decreased by 85%, but the annual
source heat pump for heating and cooling). cooling demand was increased by 28%. This is because the enhanced
The utility costs, such as electricity and natural gas, were assumed to airtightness prevents heat exchange between the indoor and outdoor
increase by 10%. The COP for heat and cool the building were assumed environments.
to increase from 2.5 to 3 and 4 from 4.5, respectively. Fig. 12 shows the ● Most environmental pollutants were occurred primarily during the
variation in the optimal solution. The energy demand of the building operational and product stages. When comparing the reference
does not change theoretically with an increase in the utility cost and building to the optimal solution, the emissions of global warming
improvement of the COP. The energy demand fluctuates slightly by potential, acidification potential, and photochemical ozone creation
1.1%, from 1100 GJ to 1112 GJ. The LCA was significantly improved by potential were decreased by 39%, 49%, and 50%, respectively, and
2.0% from 5499 tonCO2e to 5390 tonCO2e due to the improvement of the the ozone depletion potential and eutrophication potential emissions
COP. The energy systems with improved efficiency save the energy were increased by 4% and 18%, respectively.
consumption, which causes the reduction in the CO2 emissions. The ● The total expenditure was significantly decreased by 35% when the
increase in the utility cost does not affect CO2 emissions. The LCC is reference building was optimized as a passive building. Costs for the
affected by both the improvement of the COP and the increase in the initial construction and operation that account for most of the
utility cost. The improvement in efficiency decreases the total expen­ expenditure were decreased by 18% and 59%, respectively.
diture, and the increase in the utility cost increases the total expenditure. ● Variation in the optimal solution due to future uncertainty was
The LCC in the future, therefore, slightly increases by 0.7%, from 1609 evaluated considering the (1) future increase in utility costs and (2)
thousand dollars to 1621 thousand dollars without any significant improvements in energy system efficiency. The energy and economic
change. As a result, the optimal solution for passive design will not performance of optimal solution will not change significantly in the
significantly change the building demand and economic feasibility in the future; however, the environmental performance can be significantly
future; however, environmental emissions are expected to be signifi­ changed.
cantly reduced.
Table 10 shows the specific design conditions of the reference According to the results outlined above, attention should be paid to
building, Pareto optimal solution, and uncertainty solution. Compared airtightness, occupants, and window to wall ratio, which are the most
to the design condition in the reference building, the passive design influential factors for building performance. The optimization results
variables in the optimal and uncertainty solutions, such as airtightness showed that the building energy demand, life cycle assessment, and life
(x1), North window SHGC (x14), South window SHGC (x15), North cycle cost in the passive design can be reduced by up to 52.7%, 39.5%,
WWR (x20), and South WWR (x21), converge to a low bound value of and 36.9%, respectively. If system performance and utility cost increase
sampling ranges. North window U-value (x12) and South window U- in the future, this optimal solution will not significantly change in
value (x13) converge to the high bound value of the sampling ranges. building demand and economics, however the environmental emissions
This suggests that the optimal solution of passive design can be further can be significantly reduced.
improved if the building energy regulations are strengthened. By
contrast, Roof U-value (x2), North wall U-value (x3), South wall U-value
(x4), East wall U-value (x5), and West wall U-value (x6) have relatively Declaration of competing interest
moderate values in the sampling ranges. This indicates that there is an
optimal condition of North window U-value (x12) and South window U- The authors declare that they have no known competing financial
value (x13) that satisfies both the costs of initial investment and oper­ interests or personal relationships that could have appeared to influence
ational energy. For example, if the insulation condition is close to high the work reported in this paper.
boundary values, they are not suitable for the initial cost but suitable for
operating energy costs. Conversely, if it is close to low boundary values, Acknowledgements
they are suitable for operating energy cost but not suitable for initial
investment cost. Therefore, it is expected that the passive design vari­ This work was supported by the National Research Foundation of
ables, such as Roof U-value (x2), North wall U-value (x3), South wall U- Korea (NRF) grant funded by the Korea government (MSIT) (No.
value (x4), East wall U-value (x5), and West wall U-value (x6), can be 2020R1A5A1018153).

16
Y. Jung et al. Building and Environment 203 (2021) 108061

References [26] M.H. Kristensen, S. Petersen, Choosing the appropriate sensitivity analysis method
for building energy model-based investigations, Energy Build. 130 (2016)
166–176, https://doi.org/10.1016/j.enbuild.2016.08.038.
[1] C. Park, C. Kim, S. Lee, G. Lim, S. Lee, Y. Choi, Effect of control strategy on
[27] R. Wang, S. Lu, W. Feng, A three-stage optimization methodology for envelope
performance and emissions of natural gas engine for cogeneration system, Energy
design of passive house considering energy demand, thermal comfort and cost,
82 (2015) 353–360, https://doi.org/10.1016/j.energy.2015.01.045.
Energy 192 (2020) 116723, https://doi.org/10.1016/j.energy.2019.116723.
[2] APERC, Towards Zero-Emission Efficient and Resilient Buildings GLOBAL STATUS
[28] X. Chen, H. Yang, A multi-stage optimization of passively designed high-rise
REPORT 2016, 2001, pp. 1–32.
residential buildings in multiple building operation scenarios, Appl. Energy 206
[3] A. Prieto, U. Knaack, T. Auer, T. Klein, Passive cooling & climate responsive façade
(2017) 541–557, https://doi.org/10.1016/j.apenergy.2017.08.204.
design exploring the limits of passive cooling strategies to improve the
[29] S. Gou, V.M. Nik, J.L. Scartezzini, Q. Zhao, Z. Li, Passive design optimization of
performance of commercial buildings in warm climates, Energy Build. 175 (2018)
newly-built residential buildings in Shanghai for improving indoor thermal
30–47, https://doi.org/10.1016/j.enbuild.2018.06.016.
comfort while reducing building energy demand, Energy Build. 169 (2018)
[4] I. Oropeza-Perez, P.A. Østergaard, Active and passive cooling methods for
484–506, https://doi.org/10.1016/j.enbuild.2017.09.095.
dwellings: a review, Renew. Sustain. Energy Rev. 82 (2018) 531–544, https://doi.
[30] X. Chen, H. Yang, Integrated energy performance optimization of a passively
org/10.1016/j.rser.2017.09.059.
designed high-rise residential building in different climatic zones of China, Appl.
[5] A. Kilaire, M. Stacey, Design of a prefabricated passive and active double skin
Energy 215 (2018) 145–158, https://doi.org/10.1016/j.apenergy.2018.01.099.
façade system for UK offices, J. Build. Eng. 12 (2017) 161–170, https://doi.org/
[31] B. Lee, Y. Jang, J. Choi, Multi-stage optimization and meta-model analysis with
10.1016/j.jobe.2017.06.001.
sequential parameter range adjustment for the low-energy house in Korea, Energy
[6] Y. Sun, Sensitivity analysis of macro-parameters in the system design of net zero
Build. 214 (2020) 109873, https://doi.org/10.1016/j.enbuild.2020.109873.
energy building, Energy Build. 86 (2015) 464–477, https://doi.org/10.1016/j.
[32] M. Hamdy, A.T. Nguyen, J.L.M. Hensen, A performance comparison of multi-
enbuild.2014.10.031.
objective optimization algorithms for solving nearly-zero-energy-building design
[7] X. Xue, S. Wang, Y. Sun, F. Xiao, An interactive building power demand
problems, Energy Build. 121 (2016) 57–71, https://doi.org/10.1016/j.
management strategy for facilitating smart grid optimization, Appl. Energy 116
enbuild.2016.03.035.
(2014) 297–310, https://doi.org/10.1016/j.apenergy.2013.11.064.
[33] M. Krarti, A. Deneuville, Comparative evaluation of optimal energy efficiency
[8] W.J. Hee, M.A. Alghoul, B. Bakhtyar, O. Elayeb, M.A. Shameri, M.S. Alrubaih,
designs for French and US office buildings, Energy Build. 93 (2015) 332–344,
K. Sopian, The role of window glazing on daylighting and energy saving in
https://doi.org/10.1016/j.enbuild.2015.01.046.
buildings, Renew. Sustain. Energy Rev. 42 (2015) 323–343, https://doi.org/
[34] F. Ascione, N. Bianco, G. Maria Mauro, D.F. Napolitano, Building envelope design:
10.1016/j.rser.2014.09.020.
multi-objective optimization to minimize energy consumption, global cost and
[9] Z. Liu, Y. Liu, B.J. He, W. Xu, G. Jin, X. Zhang, Application and suitability analysis
thermal discomfort. Application to different Italian climatic zones, Energy 174
of the key technologies in nearly zero energy buildings in China, Renew. Sustain.
(2019) 359–374, https://doi.org/10.1016/j.energy.2019.02.182.
Energy Rev. 101 (2019) 329–345, https://doi.org/10.1016/j.rser.2018.11.023.
[35] E. Asadi, M.G. da Silva, C.H. Antunes, L. Dias, A multi-objective optimization
[10] H. Omrany, A. GhaffarianHoseini, A. GhaffarianHoseini, K. Raahemifar, J. Tookey,
model for building retrofit strategies using TRNSYS simulations, GenOpt and
Application of passive wall systems for improving the energy effciency in buildings:
MATLAB, Build. Environ. 56 (2012) 370–378, https://doi.org/10.1016/j.
a comprehensive review, Renew. Sustain. Energy Rev. 62 (2016) 1252–1269,
buildenv.2012.04.005.
https://doi.org/10.1016/j.rser.2016.04.010.
[36] F. Shadram, S. Bhattacharjee, S. Lidelöw, J. Mukkavaara, T. Olofsson, Exploring
[11] H. Goudarzi, A. Mostafaeipour, Energy saving evaluation of passive systems for
the trade-off in life cycle energy of building retrofit through optimization, Appl.
residential buildings in hot and dry regions, Renew. Sustain. Energy Rev. 68 (2017)
Energy 269 (2020) 115083, https://doi.org/10.1016/j.apenergy.2020.115083.
432–446, https://doi.org/10.1016/j.rser.2016.10.002.
[37] M. Rabani, H. Bayera Madessa, O. Mohseni, N. Nord, Minimizing delivered energy
[12] M. Žigart, R. Kovačič Lukman, M. Premrov, V. Žegarac Leskovar, Environmental
and life cycle cost using Graphical script: an office building retrofitting case, Appl.
impact assessment of building envelope components for low-rise buildings, Energy
Energy 268 (2020) 114929, https://doi.org/10.1016/j.apenergy.2020.114929.
163 (2018) 501–512, https://doi.org/10.1016/j.energy.2018.08.149.
[38] R. Lapisa, E. Bozonnet, P. Salagnac, M.O. Abadie, Optimized design of low-rise
[13] K. Lai, W. Wang, H. Giles, Solar shading performance of window with constant and
commercial buildings under various climates – energy performance and passive
dynamic shading function in different climate zones, Sol. Energy 147 (2017)
cooling strategies, Build. Environ. 132 (2018) 83–95, https://doi.org/10.1016/j.
113–125, https://doi.org/10.1016/j.solener.2016.10.015.
buildenv.2018.01.029.
[14] S. Grynning, A. Gustavsen, B. Time, B.P. Jelle, Windows in the buildings of
[39] F. Ascione, N. Bianco, G.M. Mauro, D.F. Napolitano, Retrofit of villas on
tomorrow: energy losers or energy gainers? Energy Build. 61 (2013) 185–192,
Mediterranean coastlines: pareto optimization with a view to energy-efficiency and
https://doi.org/10.1016/j.enbuild.2013.02.029.
cost-effectiveness, Appl. Energy 254 (2019), https://doi.org/10.1016/j.
[15] C. Ji, T. Hong, J. Jeong, J. Kim, M. Lee, K. Jeong, Establishing environmental
apenergy.2019.113705.
benchmarks to determine the environmental performance of elementary school
[40] E.D. Giouri, M. Tenpierik, M. Turrin, Zero energy potential of a high-rise office
buildings using LCA, Energy Build. 127 (2016) 818–829, https://doi.org/10.1016/
building in a Mediterranean climate: using multi-objective optimization to
j.enbuild.2016.06.042.
understand the impact of design decisions towards zero-energy high-rise buildings,
[16] K.T. Huang, R.L. Hwang, Future trends of residential building cooling energy and
Energy Build. 209 (2020), https://doi.org/10.1016/j.enbuild.2019.109666.
passive adaptation measures to counteract climate change: the case of Taiwan,
[41] Y. Zhai, Y. Wang, Y. Huang, X. Meng, A multi-objective optimization methodology
Appl. Energy 184 (2016) 1230–1240, https://doi.org/10.1016/j.
for window design considering energy consumption, thermal environment and
apenergy.2015.11.008.
visual performance, Renew. Energy 134 (2019) 1190–1199, https://doi.org/
[17] M. Kaboré, E. Bozonnet, P. Salagnac, M. Abadie, Indexes for passive building
10.1016/j.renene.2018.09.024.
design in urban context – indoor and outdoor cooling potentials, Energy Build. 173
[42] L. Magnier, F. Haghighat, Multiobjective optimization of building design using
(2018) 315–325, https://doi.org/10.1016/j.enbuild.2018.05.043.
TRNSYS simulations, genetic algorithm, and Artificial Neural Network, Build.
[18] X. Chen, H. Yang, Y. Wang, Parametric study of passive design strategies for high-
Environ. 45 (2010) 739–746, https://doi.org/10.1016/j.buildenv.2009.08.016.
rise residential buildings in hot and humid climates: miscellaneous impact factors,
[43] J. Xu, J.H. Kim, H. Hong, J. Koo, A systematic approach for energy efficient
Renew. Sustain. Energy Rev. 69 (2017) 442–460, https://doi.org/10.1016/j.
building design factors optimization, Energy Build. 89 (2015) 87–96, https://doi.
rser.2016.11.055.
org/10.1016/j.enbuild.2014.12.022.
[19] J. Jazaeri, R.L. Gordon, T. Alpcan, Influence of building envelopes, climates, and
[44] X. Wang, X. Mai, B. Lei, H. Bi, B. Zhao, G. Mao, Collaborative optimization between
occupancy patterns on residential HVAC demand, J. Build. Eng. 22 (2019) 33–47,
passive design measures and active heating systems for building heating in
https://doi.org/10.1016/j.jobe.2018.11.011.
Qinghai-Tibet plateau of China, Renew. Energy 147 (2020) 683–694, https://doi.
[20] R. Gagnon, L. Gosselin, S. Decker, Sensitivity analysis of energy performance and
org/10.1016/j.renene.2019.09.031.
thermal comfort throughout building design process, Energy Build. 164 (2018)
[45] W. Wang, R. Zmeureanu, H. Rivard, Applying multi-objective genetic algorithms in
278–294, https://doi.org/10.1016/j.enbuild.2017.12.066.
green building design optimization, Build. Environ. 40 (2005) 1512–1525, https://
[21] Y. Wang, J. Kuckelkorn, F.Y. Zhao, H. Spliethoff, W. Lang, A state of art of review
doi.org/10.1016/j.buildenv.2004.11.017.
on interactions between energy performance and indoor environment quality in
[46] T. Hong, J. Kim, M. Lee, A multi-objective optimization model for determining the
Passive House buildings, Renew. Sustain. Energy Rev. 72 (2017) 1303–1319,
building design and occupant behaviors based on energy, economic, and
https://doi.org/10.1016/j.rser.2016.10.039.
environmental performance, Energy 174 (2019) 823–834, https://doi.org/
[22] G. Calleja Rodríguez, A. Carrillo Andrés, F. Domínguez Muñoz, J.M. Cejudo López,
10.1016/j.energy.2019.02.035.
Y. Zhang, Uncertainties and sensitivity analysis in building energy simulation using
[47] F. Salata, V. Ciancio, J. Dell’Olmo, I. Golasi, O. Palusci, M. Coppi, Effects of local
macroparameters, Energy Build. 67 (2013) 79–87, https://doi.org/10.1016/j.
conditions on the multi-variable and multi-objective energy optimization of
enbuild.2013.08.009.
residential buildings using genetic algorithms, Appl. Energy 260 (2020) 114289,
[23] L. Rivalin, P. Stabat, D. Marchio, M. Caciolo, F. Hopquin, A comparison of methods
https://doi.org/10.1016/j.apenergy.2019.114289.
for uncertainty and sensitivity analysis applied to the energy performance of new
[48] W. Tian, Y. Heo, P. de Wilde, Z. Li, D. Yan, C.S. Park, X. Feng, G. Augenbroe,
commercial buildings, Energy Build. 166 (2018) 489–504, https://doi.org/
A review of uncertainty analysis in building energy assessment, Renew. Sustain.
10.1016/j.enbuild.2018.02.021.
Energy Rev. 93 (2018) 285–301, https://doi.org/10.1016/j.rser.2018.05.029.
[24] K. Menberg, Y. Heo, R. Choudhary, Sensitivity analysis methods for building
[49] A. Mastrucci, P. Pérez-López, E. Benetto, U. Leopold, I. Blanc, Global sensitivity
energy models: comparing computational costs and extractable information,
analysis as a support for the generation of simplified building stock energy models,
Energy Build. 133 (2016) 433–445, https://doi.org/10.1016/j.
Energy Build. 149 (2017) 368–383, https://doi.org/10.1016/j.
enbuild.2016.10.005.
enbuild.2017.05.022.
[25] T. Wei, A review of sensitivity analysis methods in building energy analysis,
[50] Z. O’Neill, F. Niu, Uncertainty and sensitivity analysis of spatio-temporal occupant
Renew. Sustain. Energy Rev. 20 (2013) 411–419, https://doi.org/10.1016/j.
behaviors on residential building energy usage utilizing Karhunen-Loève
rser.2012.12.014.

17
Y. Jung et al. Building and Environment 203 (2021) 108061

expansion, Build. Environ. 115 (2017) 157–172, https://doi.org/10.1016/j. [68] Passive Design Guideline, Presidential Commision on Architecture Policy, 2012 (n.
buildenv.2017.01.025. d.).
[51] L. Gabrielli, A.G. Ruggeri, Developing a model for energy retrofit in large building [69] Statistics Korea Life Time Survey Results vol. 2014, Statistics Korea, 2014 (n.d.).
portfolios: energy assessment, optimization and uncertainty, Energy Build. 202 [70] SimLab Version 2, 2 Simulation Environment for Uncertainty and Sensitivity
(2019) 109356, https://doi.org/10.1016/j.enbuild.2019.109356. Analysis, Developed by the Joint Research Center of the European Commission,
[52] M. Ferrara, A. Rolfo, F. Prunotto, E. Fabrizio, EDeSSOpt – energy demand and 2004 (n.d.).
supply simultaneous optimization for cost-optimized design : application to a [71] K. Deb, A. Pratap, S. Agarwal, T. Meyarivan, A fast and elitist multiobjective
multi-family building, Appl. Energy 236 (2019) 1231–1248, https://doi.org/ genetic algorithm: NSGA-II, IEEE Trans. Evol. Comput. 6 (2002) 182–197, https://
10.1016/j.apenergy.2018.12.043. doi.org/10.1109/4235.996017.
[53] F. Rosso, V. Ciancio, J. Dell’Olmo, F. Salata, Multi-objective optimization of [72] J.J. Wang, Y.Y. Jing, C.F. Zhang, J.H. Zhao, Review on multi-criteria decision
building retrofit in the Mediterranean climate by means of genetic algorithm analysis aid in sustainable energy decision-making, Renew. Sustain. Energy Rev. 13
application, Energy Build. 216 (2020) 109945, https://doi.org/10.1016/j. (2009) 2263–2278, https://doi.org/10.1016/j.rser.2009.06.021.
enbuild.2020.109945. [73] En, Bs. “15978: 2011, Sustainability of construction works. Assessment of
[54] F. Harkouss, F. Fardoun, P.H. Biwole, Passive design optimization of low energy environmental performance of buildings, Calculation method (2011) (n.d.).
buildings in different climates, Energy 165 (2018) 591–613, https://doi.org/ [74] Y. Jung, J. Kim, H. Lee, Multi-criteria evaluation of medium-sized residential
10.1016/j.energy.2018.09.019. building with micro-CHP system in South Korea, Energy Build. 193 (2019)
[55] Guidelines for the Design and Performance Evaluation of Eco-Friendly Houses, 201–215, https://doi.org/10.1016/j.enbuild.2019.03.051.
Minister of Land, Infrastructure and Transport, 2009 (n.d.). [75] R. Santos, A.A. Costa, J.D. Silvestre, L. Pyl, Integration of LCA and LCC analysis
[56] Sketchup official web site (n.d.), https://www.sketchup.com. within a BIM-based environment, Autom. ConStruct. 103 (2019) 127–149, https://
[57] Trnsys document_05_MultizoneBuilding_5.2.4.1. Definition of a New(opaque) doi.org/10.1016/j.autcon.2019.02.011.
Construction Type, (n.d.). [76] T.H. Kim, C.U. Chae, Environmental impact analysis of acidification and
[58] Y. Jung, J. Kim, H. Lee, Multi-criteria evaluation of medium-sized residential eutrophication due to emissions from the production of concrete, Sustain. Times 8
building with micro-CHP system in South Korea, Energy Build. 193 (2019), (2016) 1–20, https://doi.org/10.3390/su8060578.
https://doi.org/10.1016/j.enbuild.2019.03.051. [77] Manfred Hegger, et al., Construction Materials Manual, Walter de Gruyter, 2013
[59] TRNSYS18_Sensible and Latent Loads for Occupancy Types., (n.d.). (n.d.).
[60] J.H. Yoo, K.H. Kim, Development of methodology for estimating electricity use in [78] Jeroen B. Guinée, Erwin Lindeijer (Eds.), Handbook on Life Cycle Assessment:
residential sectors using national statistics survey data from South Korea, Energy Operational Guide to the ISO Standards, vol. 7, Springer Science & Business Media,
Build. 75 (2014) 402–409, https://doi.org/10.1016/j.enbuild.2014.02.033. 2002 (n.d.).
[61] Standard, I. S. O. “7730, Ergonomics of the Thermal Environment-Analytical [79] I. Dincer, A. Abu-Rayash, Sustainability Modeling, 2020, https://doi.org/10.1016/
Determination and Interpretation of Thermal Comfort Using Calculation of the b978-0-12-819556-7.00006-1.
PMV and PPD Indices and Local Thermal Comfort Criteria, (2005), 2005 (n.d.). [80] Korea Life Cycle Inventory Database, Korea Environmental Industry and
[62] S.S. Garud, I.A. Karimi, M. Kraft, Design of computer experiments: a review, Technology Institute. Official web site: http://www.epd.or.kr/lci/lciDb.do.
Comput. Chem. Eng. 106 (2017) 71–95, https://doi.org/10.1016/j. [81] S. Roh, S. Tae, R. Kim, Analysis of embodied environmental impacts of Korean
compchemeng.2017.05.010. apartment buildings considering major building materials, Sustain. Times 10
[63] J. Kim, Y. Jung, H. Lee, Optimization of dynamic poly-generation system and (2018), https://doi.org/10.3390/su10061693.
evaluation of system performance in building application, Energy Convers. Manag. [82] S. Roh, S. Tae, S.J. Suk, G. Ford, Evaluating the embodied environmental impacts
201 (2019) 112128, https://doi.org/10.1016/j.enconman.2019.112128. of major building tasks and materials of apartment buildings in Korea, Renew.
[64] (Draft) Notice of Legislation, Standard Adjustment of Apartment House Ven- Sustain. Energy Rev. 73 (2017) 135–144, https://doi.org/10.1016/j.
Tilation Equipment (Article 11, Clause 1), Minister of Land, Infrastructure and rser.2017.01.081.
Transport Notice, 2013 (n.d.). [83] Construction Standard Market Unit Price Application in the First Half of 2019,
[65] Energy-saving Design Standard of Building _ Indoor Temperature and Humidity Ministry of Land, Infrastructure and Transport, Korea Institute of Construction
Standard for Calculation of Capacity of Heating and Heating Equipment, Minis- Ter Technology, 2019 (n.d.).
of Land, Infrastructure and Transport Notice 2015-596, 2015 (n.d.). [84] Korea Electric Power Corporation(Kepco), Official website: (n.d.), http://cyber.
[66] Trnsys document_05_MultizoneBuilding_5.2.4.1. Definition of a New(opaque) kepco.co.kr/ckepco/front/jsp/CY/E/E/CYEEHP00101.jsp n.d.
Construction Type, (n.d.). [85] Seoul city gas, Official website: https://www.seoulgas.co.kr/front/payment/ga
[67] S. Ahn, S. Han, D. Lee, J. Kim, D. Ju, D. Hwang, T.S.O. Air-conditioning, sPayTable.do.
R. Engineers, O. Korea, A preliminary study on prioritizing energy saving measures [86] Korea Water Resources Corporation (KOWACO), Ministry of Construction and
based on sensitivity analysis of determinants on building energy consumption, Transportation (MOCT), Official website: http://kwater.or.kr.
Proc. SAREK Annu. Conf. (2012). [87] Backtel demolition industry, Official web site (n.d.), http://www.backtel.co.kr/b
ase_1/03detail/03detail03.php.

18