Energy & Buildings 178 (2018) 216–227

Contents lists available at ScienceDirect

Energy & Buildings

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

A simulation-based optimization method for designing energy

efficient buildings
Sadik Yigit a,∗, Beliz Ozorhon b
a
Department of Civil Engineering, Istanbul Kultur University, Istanbul, Turkey
b
Department of Civil Engineering, Bogazici University, Istanbul, Turkey

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

Article history: Buildings have a remarkable impact on the environment, therefore finding efficient design configurations
Received 15 March 2018 satisfying conflicting criteria such as, economic and environmental performance has become an important
Revised 14 August 2018
task. The aim of this study is to propose a methodology to aid designers in satisfying the requirements of
Accepted 28 August 2018
government regulations and green building certification programs, while optimizing the energy consump-
Available online 5 September 2018
tion and maintaining the thermal comfort. In this context, a software package combining a tailor-made
Keywords: thermal simulation software and Matlab Optimtool is developed to implement proposed methodology.
Energy simulation The developed software package offers an effective method to perform large number of simulations to
Optimization find optimal building configuration. The software package was tested on a reference building that rep-
Genetic algorithm resents a typical residential building in Turkey. The testing process is conducted on a five story building
Heat balance method located in Istanbul and building information such as building size, location and occupation schedule were
used within the developed software. Based on an extensive market search for building materials, cost
functions for each parameter were developed as well as a database required for optimization process. A
genetic algorithm optimization technique was utilized to minimize the objective function and find the
optimal building configuration for the selected building. Development of a simulation-based optimization
method fulfills the need for a tool that assists designers to find better design alternatives at the concep-
tual design stage. The software package requires least amount of data input for energy simulation process
to improve usability. A tailor-made energy simulation module was developed to significantly reduce opti-
mization time period. Besides, instead of coupling two separate software packages, performing the energy
simulation and optimization processes on a single platform (MATLAB) reduces the time required to find
optimal design and eliminates compatibility issues. Developing simulation-based optimization software
on a single platform increased the flexibility and user-friendliness of the software. The effectiveness of
the approach for finding optimal building configuration is demonstrated in the presented test cases.
© 2018 Elsevier B.V. All rights reserved.

1. Introduction local standards force construction sector to build energy efficient

Buildings (residential/commercial/institutional) use approxi-
mately 30%–40% of the total energy consumed in the world
[1, 2]. Almost 50% of the energy used is consumed for the air
conditioning of the buildings and about 40% of the world’s mate-
rial consumption is used for housing projects [3]. European Union
endeavor to improve energy efficiency in buildings and aim to
reach the goal of all buildings to be nearly zero energy building
by 2020 [4]. In addition, regulations such as Energy Performance
of Buildings Directive (EPBD), American Society of Heating, Refrig-
eration, and Air-Conditioning Engineers (ASHRAE), Handbook and
∗
Corresponding author.
E-mail addresses: s.yigit@iku.edu.tr (S. Yigit), beliz.ozorhon@boun.edu.tr (B.
Ozorhon).
buildings.
Energy performance of the buildings substantially depends on
decisions in the early phases of the design process. Designing a
building is a complex process; in which experts from many disci-
plines contribute to make decisions to find a design solution that
meets all performance criteria [5]. Thus, finding more efficient de-
sign alternatives that fulfill multiple conflicting criteria is a diffi-
cult task. Assessing the energy performance of the buildings and
deciding on the materials and systems to be used in the build-
ings require complex methods and interrelated calculations. Green
buildings save energy, minimize the environmental impacts and
provide healthier environments compared to conventional build-
ings. On the other hand, green buildings are not considered as an
attractive option, because most of the technologies and materials
used in green buildings have higher initial costs [2]. A careful de-

https://doi.org/10.1016/j.enbuild.2018.08.045
0378-7788/© 2018 Elsevier B.V. All rights reserved.
 S. Yigit, B. Ozorhon / Energy & Buildings 178 (2018) 216–227
sign process and material selection are required to decrease, initial
costs and in the long term operational costs of buildings.
Due to abovementioned reasons, over the past two decades,
building energy optimization has been widely discussed and num-
ber of studies in this area sharply increased [6]. The researchers
developed simulation-based optimization methods to overcome
the complexity of these calculations. Considering the recent ad-
vancements in computer science, simulation-based methods have
become an effective alternative to improve building performance.
The majority of the current literature on building energy opti-
mization focuses on coupling an optimization tool and commer-
cial building energy simulation software to optimize the build-
ing energy efficiency. Energy Plus and TRNSYS are the most pop-
ular building energy simulation programs and Matlab optimization
toolbox is the most popular optimization tool used by researchers
studying in this area.
Recently there has been significantly increasing interest in
building energy optimization methods within academic communi-
ties. The first attempt of simulation based optimization methods
was in 70s, to optimize thermal performance of the office build-
ings with computer based models [7]. However, most of the stud-
ies related to building thermal design optimization were carried
out in 20 0 0s and number of studies sharply increased in 2005
[6]. Vast majority of the researchers, focused on coupling detailed
simulation software with optimization software to find energy-
optimal configurations for buildings in a shorter period of time [1,
4, 8-15]. Proposed simulation methods were developed by combin-
ing detailed simulation software such as, DOE-2, TRNSYS, Energy-
Plus, IDA ICE and ATHENA with optimization software like Matlab,
GenOpt etc. The developed simulation-based optimization methods
are accepted as effective methods for building energy optimiza-
tion there are still difficulties such as, long run-times, labor re-
quirements, practicality, flexibility and coupling efficiency. The de-
veloped simulation-based optimization methods reduced runtime
periods. However, due to number of design variables, coupling ef-
ficiency and long simulation periods, optimization periods of the
proposed methods are still too long [13, 16, 17].
Some authors used simplified building simulation methods to
speed up the optimization process [8, 18-24]. Some other authors
used surrogate models to overcome this problem [25, 16, 17]. Yang
et al. [26] conducted a review study to investigate applicability of
genetic algorithm based optimization methods for building ther-
mal design. Kwong et al. [27] published a review, which provides
a comprehensive review of the energy efficiency improvement po-
tentials in tropical buildings by considering thermal comfort of oc-
cupants. Nguyen et al. [6] carried out an extensive review study to
investigate simulation-based building optimization techniques.
The literature survey confirms that simulation-based optimiza-
tion is certainly a promising method for building energy perfor-
mance optimization. However, there are still difficulties in cou-
pling efficiency considering time, labor, practicality and flexibility
[6]. Asadi et al. [16] stated that, 950 simulation runs for training
took 3 days to be completed. Magnier and Hagihighat [17] indi-
cated that, a system consisting of GA and TRNSYS needs 10 years
to optimize a building. Hasan et al. [10] proposed an optimization
system by combining IDA ICE and GenOpt and the developed sys-
tem needed 73 h to run 6400 cases. In addition, the performance
of the coupling interfaces between simulation tools and optimiza-
tion packages is one of the important factors that negatively af-
fect the performance of the simulation based optimization meth-
ods [6]. Nguyen et al. [6] indicated that validation and reliability
of the surrogate models is problematic and using surrogate mod-
els increases the uncertainty and the risk of accumulative errors
in the optimization process. In addition, development of surrogate
models greatly increases preprocessing time and requires expertise
[16, 17, 25]. Due to simplifications and assumptions in the simpli-
217
fied simulation methods, fundamental calculations are hidden and
it is impossible to reveal effects of simple changes in building [28].
Coakley et al. [29] conducted a study on simulation methods and
divided into three category; black box, grey box and detailed sim-
ulation. The study demonstrated that in black box models there is
no explicit link between model inputs and physical design parame-
ters so, it is impossible to compute effects of design or operational
changes.
In a large majority of the studies multiple software packages,
which require high expertise in optimization methods and energy
simulation were used. As mentioned above the current study aims
to introduce an efficient, user-friendly, time-saving method, which
requires less labor and expertise. A standalone software is de-
veloped to implement the proposed methodology. The introduced
method is applied to find energy-optimal design for a residen-
tial building in Istanbul. Using single software instead of combin-
ing multiple software eliminates coupling problems and greatly re-
duces optimization time period.
2. Methodology of the study
In this study a complete, flexible and practical software package
that integrates material selection, design, and budget to find the
optimal design for the buildings is proposed. The software pack-
age consists an energy simulation module and a genetic algorithm
optimization module. The modules are integrated by performing
the energy simulation, optimization processes and financial calcu-
lations on a single platform (Matlab). The computer used for the
optimization process has an Intel i7-2600 central processing unit
and 16 GB ram. The process of building energy optimization is pre-
sented in Fig. 1.
The steps of the proposed methodology are listed below:
i The developed software reads the input data consisting param-
eters such as weather condition data, building location, building
geometry, optimization constraints, user requirements etc.
ii The genetic algorithm optimization module creates a random
population consisting building configuration combinations. Only
at the very first cycle of the optimization process the genetic
algorithm module creates a random population and in follow-
ing cycles the module creates populations by implementing
crossover, elitism, selection and mutation processes.
iii The genetic algorithm optimization module sends building con-
figurations array to energy simulation module and the en-
ergy simulation module calculates energy demands by using re-
ceived configurations and input data.
iv The energy simulation module completes the simulation pro-
cess and sends the results array back to optimization module
for evaluation.
v The optimization module evaluates the received energy simu-
lation results array. If the results meet one of the termination
criteria the process ends and the software yields an optimal or
near optimal solution. If the termination criteria are not met,
the cycle goes back to step 2.
2.1. Determination of decision variables
Simulation-based optimization methods mostly depend on de-
cision variables taken into consideration. Therefore, for a proper
thermal calculation decision variables are required to be deter-
mined cautiously. The number of decision variables directly af-
fects the manageability of the calculations and optimization pe-
riod. Consequently, one of the main challenges of simulation-based
optimization methods is to ensure that all the important decision
variables are taken into consideration while keeping the calcula-
tion process manageable [4]. A literature review is required to se-
lect decision variables used in this study.
218 S. Yigit, B. Ozorhon / Energy & Buildings 178 (2018) 216–227
Fig. 1. Framework of the methodology.
Ferrara et al. [4] selected building components such as, thermal
resistance of walls, roof and slab internal insulation, window type,
roof type, window size, thermal resistance of walls external insu-
lation, thermal resistance of walls, roof additional insulation, heat-
ing/cooling system and ventilation system as decision variables. In
another study conducted in 2015 selected decision variables can
be listed as; solar absorbance of external plaster, infrared emit-
tance of external plastering, thickness of thermal insulation, thick-
ness of bricks, brick density and thermal transmittance of win-
dows [14]. Building orientation, aspect ratio of the building, win-
dow type, window to wall ratio, wall type, layers of wall, roof type
and each layer of roof were chosen as the variables of optimization
process in a study conducted to optimize green buildings thermal
design [1]. In a research developed to optimize buildings in Colom-
bia used wood components, adhesive sealants, paint, carpet, roofs,
glasses and windows as variables of optimization model [2].
According to a research aiming to review articles published in
the field of simulation-based optimization method studies, energy
efficiency measures affecting the energy performance can be listed
as; external wall and roof color, ventilation strategy, external wall
type, power of gas stove, discharge coefficient bedrooms, thickness
of the ceiling concrete slab, crack window, thickness of roof insu-
lation, maximum number of occupants, thickness of internal mass,
window type, density of slab concrete, brick thickness, conductiv-
ity of EPS insulation, crack window bedrooms, thickness wooden
floor, maximum equipment power, brick density and window size
[6]. The list covers almost all the design variables that affect the
energy performance of the buildings.
As a result of the literature review, decision variables that
affect energy efficiency of the buildings are determined. To define
the decision variables a component description format for all ele-
ments forming the building envelope is suggested. The description
format is formed of details such as: layer thicknesses, thermal
properties, material types and properties, service lives and initial
costs. The selected energy decision variables are presented in
Table 1. The decision variables used in the developed simulation-
based optimization method can be discrete, continuous or categor-
ical.
2.2. Building energy simulation
The energy simulation module developed in this study is based
on heat balance method. Heat balance method is an hourly dy-
namic calculation method in which solar heat gains and inter-
nal heat gains are calculated in detail. The building’s internal sur-
face temperatures are identified separately and natural ventilation,
shading, lighting and other internal loads are taken into consider-
ation. Heat balance method reduces the number of assumptions
made and its models are closest to real physical buildings. The
heat balance method takes its name from application of first law
of thermodynamics, “energy is conserved” in the inner and outer
surfaces and the zone air [30]. Pedersen et al. [28] published the
formulation of heat balance method for cooling load calculations
and it is accepted to be scientifically the most rigorous method.
Kusuda [31] conducted the first study in which all the elements of
heat balance method were used in a method called NBSLD. Heat
balance method was also used in energy analysis programs such
as, BLAST and TARP [30, 32].
The developed software use conduction transfer functions (CTF)
to formulate the conduction process of exterior walls, roof and
floor. Layer thicknesses, material heat capacity and density are
required for CTF calculations. An external computer program called
PRF/RTF Generator is implemented to calculate CTFs of building
components and the calculated CTFs are stored in a database [33].
The CTF values can also be obtained from ASHRAE handbooks.
A clear sky model is implemented to calculate diffuse and direct
solar radiation [34]. Heat loads of the natural ventilation and
infiltration are calculated using methods provided from ASHRAE
Handbook [30]. A single zone building model is used to speed up
the simulation process and the equations of heat balance method
are solved using an iteration process introduced by Pedersen et al.
[28].
 S. Yigit, B. Ozorhon / Energy & Buildings 178 (2018) 216–227 219

Table 1
Determined energy efficiency measures.

Parameter description Parameter name Unit Variation

WT Window type [–] Choice between options

WSS1 Window size (Surface 1) [m2 ] Continuous
WSS2 Window size (Surface 2) [m2 ] Continuous
WSS3 Window size (Surface 3) [m2 ] Continuous
WSS4 Window size (Surface 4) [m2 ] Continuous
ST Slab type [–] Choice between options
SIT I Slab Ins. I Thickness [m] Discrete
SIT II Slab Ins. II Thickness (If exist) [m] Discrete
RT Roof type [–] Choice between options
RIT I Roof ins. I Thickness [m] Discrete
RIT II Roof ins. II Thickness (If exist) [m] Discrete
WT Wall type (S1, S2, S3, S4) [–] Choice between options
WI I Wall insulation I Thickness [m] Discrete
WI II Wall insulation II Thickness (If exist) [m] Discrete
BWC Building wall absorption coefficient [–] Continuous
BRC Building roof absorption coefficient [–] Continuous

receives all the members of the population and calculates their

heat loads. After the energy consumption of the each individual
in the population is calculated, the GA optimization tool evaluates
the fitness of the members. If the stop criteria are met GA tool ter-
minates the optimization process and displays the Pareto optimal
result. If the stop criteria are not met, GA optimization tool cre-
ates a new population using selection, elitism, crossover and mu-
tation. Then, the new population is sent back to energy simulation
module. The cycle repeats until the stop criteria and constraints
are met.

3. Prototype software

Prototype software (E-Mat) is developed based on the method-

Fig. 2. Genetic algorithm energy optimization process.
2.3. Genetic algorithm optimization methodology
In this study, the GA optimization tool in Matlab OptimTool
is utilized to automatically run building energy simulations and
find optimal building thermal design. Genetic Algorithm (GA) is
a search method, which is based on simplified natural evolution
theory and can be categorized as a meta-heuristic with global per-
spectives [35]. Using genetic algorithm optimization methods for
thermal design of the buildings has become popular among the
researchers since early 20 0 0s [6].
The process of building energy optimization is presented in the
Fig. 2. At the start of the optimization process GA creates a ran-
dom population. Each individual in the population are decoded and
sent to energy simulation module. The energy simulation module
ology presented in previous section. The developed prototype soft-
ware is provided to support the designers at the preliminary stages
of the design process. The prototype software is developed in Mat-
lab (Matlab 2012b). Matlab consists many toolboxes for technical
computing and supports object-oriented programming. Optimtool
is the optimization toolbox of Matlab, which provides functions for
finding parameters that minimize objectives while satisfying con-
straints. The genetic algorithm in the OptimTool is utilized to au-
tomatically run the developed energy simulation module and find
the optimal solution.
A Microsoft Excel sheet is utilized to create a database and a
user interface. Matlab graphical output tools are utilized to present
the results of the analysis. The data required to run the building
simulation such as; economic constants, unit prices, weather data,
building components data, building usage schedules, internal loads,
indoor temperatures and orientation of the building are inserted
to Microsoft Excel sheet. Microsoft Excel automatically fills other
forms to numerically define building geometry and configuration.
The energy simulation module requires least amount of data in-
put. It is easy to use and simulation processes require less time
compared to other commercial software. E-Mat reads the MS Ex-
cel sheet once and obtains the required data for energy simulation
and optimization processes. Then, completes the optimization pro-
cess and outputs the optimal building configuration.
3.1. Implementation of the software
E-Mat requires building components database, economic data,
weather data, location data, building geometry, GA parameters and
building usage schedules to run optimization process. MS Excel
form used as user interface of the software is presented in the Fig.
3. The GA parameters needed to perform optimization process can
be listed as; number of objectives, number of constraints, muta-
220 S. Yigit, B. Ozorhon / Energy & Buildings 178 (2018) 216–227

Fig. 3. User interface for entering energy simulation input data.

tion probability, crossover probability, population size, maximum
number of generations, function tolerance, elitism rate, number of
variables and boundary conditions. The number of wall types, roof
types, foundation types and window types are dependent on the
market research or user’s will. The window area at each surface
of the building is defined considering regulations, owner’s will and
architectural design. GA parameters such as population size, max-
imum generation, elitism and crossover probability constants are
decided based on the user’s experience, literature and Matlab rec-
ommendations.
3.2. Validation of building energy simulation software
The developed energy simulation module runs yearly energy
analysis in a shorter period of time with a limited amount of in-
put data. Besides, the energy simulation module is expected to give
reasonable results. To validate the results of building energy simu-
lation module, the results from the developed software and a de-
tailed energy analysis tool (EnergyPlus) are compared. Energy Plus
is an open source software used to calculate and analyze indoor
environment and energy demand of the buildings [36]. EnergyPlus
uses a heat balance based model to carry out thermal simulations
of the buildings. The program carries out a very detailed simula-
tion process to calculate all the energy inputs and outputs includ-
ing lighting, heating, cooling and appliances. On the other hand,
EnergyPlus has long run-times and requires a detailed input data
for building description, materials and HVAC systems.
In this study, a completely new software module for building
energy simulation is proposed. Therefore, the validation of the en-
ergy simulation module is one of the most important aspects of
the study. A commercial prototype building model developed by
U.S. Department of Energy is used in the validation analysis. The
 S. Yigit, B. Ozorhon / Energy & Buildings 178 (2018) 216–227
Table 2
Test case component layers and basic properties.
Construction Layers
External walls Steel-frame walls (2 × 4 16IN o.c.)
0.4 in. Stucco
5/8 in. gypsum board
Wall Insulation
5/8 in. gypsum board
Roof Roof membrane
Roof insulation
Metal decking
Foundation 20 cm thick concrete slab poured directly on to the earth
Windows Double Glazing 8 mm thick
prototype buildings are a set of structure that covers almost 80% of
the commercial buildings in United States. The energy analyzes of
the prototype buildings are published by DOE to support develop-
ment of building energy codes and standards [37]. The software is
tested by running yearly energy simulations of the prototype build-
ing in 10 different climate regions.
3.2.1. Test case for validation analysis
A four story mid-rise apartment published by DOE is selected
for validation analysis. The total floor area of the building is
3131 m2 , floor to ceiling height is 3 m and the average total win-
dow area is 20% of the total wall area. Total wall area is 1519 m2
and the total window area is 303.8 m2 . Each floor of the building
has 8 apartments except the ground floor, which has 7 apartments
and a lobby. The building component layers and basic properties
are presented in Table 3. Natural gas is used for heating and the
heating system efficiency is 0.8. Air conditioning units use electric-
ity and the average efficiency is about 3.45. The prototype building
is heated to 21 °C and cooled to 24 °C. It is assumed that an ideal
controller controls the heating and cooling when it is necessary
and the power needed for cooling and heating system is available.
The weather data and building usage schedule are acquired from
DOE website. The schedule is utilized to find internal heat loads
such as, appliances and lighting loads.
3.2.2. Validation results
The prototype building is simulated in 10 different climatic re-
gions to validate the developed simulation module and the results
are presented in Figs. S.1a and S.1b. The results of analyses are
compared to EnergyPlus simulation results. It can be said that, the
vast majority of analyses have given reasonable results. The re-
sults of the simulations in, Memphis, Baltimore, Miami, Phoenix
and Boise Idaho are beyond expectations. The results of the de-
veloped energy simulation module are very similar to EnergyPlus
simulation results, which are carried out based on 2013 ASHRAE
90.1. The simulation results in Chicago, Vancouver and Houston are
also compared to results of EnergyPlus energy simulations and it
can be said that the results are reasonable. However, the results
for San Francisco and Fairbanks are less accurate due to extraordi-
nary weather conditions and fluctuating weather temperature.
There are significant differences between the simulation results
for ASHRAE 20 04, 20 07, 2010 and 2013. For this reason, it is un-
derstood that energy simulation methods are a developing field
of study. Latest ASHRAE standard’s calculations are more detailed
and accurate compared to older standards. However, new methods
have longer runtimes and require more input data for simulations.
The results showed that the developed software gives reasonable
results for heating and cooling load calculations compared to Ener-
gyPlus results. The presented monthly energy consumption profiles
also show similar behavior. The results prove that the developed
221
Dimensions Tilt angle
2 × (46.3 m × 12 m) Vertical
2 × (17 m × 12 m)
Total area = 1519 m2
46.3 m × 17 m Horizontal
Total area = 787.1 m2
46.3 m × 17 m Horizontal
Total area = 787.1 m2
Total area = 303.8 m2 Vertical
software gives acceptable results and it is a reliable tool to be used
in this study.
4. Test cases
The first objective of the testing process is to show how de-
sign problems are defined in the developed software package. The
second objective is to evaluate the results obtained. Finally, the
test cases are used to identify weaknesses and strengths of pro-
posed methodology to offer changes and improvements in the pro-
totype software. In order to test the prototype software a resi-
dential building in Istanbul is selected. The objective of optimiza-
tion problem is the minimization of the energy required to provide
thermal comfort in the reference building.
4.1. Reference building
In Turkey, the urban regeneration law enforced in May 2012 is
being applied. For this reason, the construction of new buildings in
Turkey has gained a great acceleration. Since 2002, 98,602 build-
ings have been constructed in Istanbul and 88,277 of the build-
ings built are residential buildings; resulting in 89.5% of the build-
ings. Around 28% of the residential buildings built in Istanbul have
5 floors and 20% have 6 floors [38]. In this context, the building
used in the testing process represents a large part of buildings in
Istanbul. The building selected for the test cases is a real 5-storey
building located in Istanbul at 41.01° latitude and 28.77° longitude.
The reference building sits on an area of 300 m2 and has 5
normal floors and 1 attic floor. The total area of ground floor is
300 m2 and other floors are 320 m2 . The height of each storey
(floor to floor height) is 3.20 m. The plans and northwest side
view drawings of the reference building is presented in the Fig.
4. The reference building is assumed to be cooled to 24 °C and
heated to 21 °C. Since the building is a typical residential build-
ing, it is considered that there is no mechanical ventilation. Heat
loads of the appliances and equipment are neglected. The refer-
ence building usage schedule has been determined in accordance
with the ASHRAE standards [37]. Due to the limits of the proto-
type software, the reference building has been regarded as a single
zone. However, this simplification makes the software practical and
shortens the calculation time.
4.2. Software database formation
In the following sections, the components defined in the pro-
totype software database are described. E-Mat creates design al-
ternatives by using the building component properties stored in
database. The unit prices of all materials are acquired from the
2017 unit price list of the Ministry of Environment and Urbaniza-
tion.
222 S. Yigit, B. Ozorhon / Energy & Buildings 178 (2018) 216–227

Fig. 4. Plans and cross-section of the reference building.

Table 3 Table 4
Exterior wall types and layer thicknesses. Roof types and layer thicknesses.

Type Layers Thickness (mm) Type Layers Thickness (mm)

Wall type I Gypsum plaster 20 Roof type I Gypsum plaster 20

Brick wall 100 Concrete slab 200
Cement plaster 20 XPS foamboard 20–100
Rock wool 20–100 Protective concrete 50
Cement bonded particle board 30–60 Gravel 30
Wall type II Gypsum plaster 20 Roof type II Gypsum plaster 20
Brick wall 100 Concrete slab 200
Cement plaster 20 XPS foamboard 20–100
Thermal insulation (EPS) 20–100 Screed 50
Cement plaster 20 Roof type III Gypsum plaster 20
Wall type III Gypsum plaster 20 Concrete slab 200
Brick wall 100 Rock wool 20–100
Cement plaster 20 XPS foamboard 20–100
Rock wool 20–100 Protective concrete 50
Thermal insulation (EPS) 20–100 Roof type IV Gypsum plaster 20
Cement plaster 20 Concrete slab 200
Wall type IV Gypsum plaster 20 XPS foamboard 20–100
Brick wall 100 Protective concrete 50
Cement plaster 20 Roof tiles 40
Thermal insulation (EPS) 20–100
Aluminum composite panel 4

4.2.1. Outer wall types
Four types of exterior wall are defined in the prototype soft-
ware’s database; W T1, W T2, W T3 and W T4. The layers and layer
thicknesses of the exterior wall types are presented in the Table 3.
The most commonly used exterior wall types in the Turkish con-
struction sector are selected. The main material used in the outer
walls is common brick. EPS, rock wool, cement bonded particle
board and aluminum composite panels are preferred for insulation.
4.2.2. Roof types
Four types of roofs are defined in the prototype software’s
database: RT1, RT2, RT3 and RT4. The layers and layer thicknesses
of the roof types are presented in Table 4. The most commonly
used material types and roof systems are selected. XPS foamboard,
rock wool and gravel are preferred for insulation.
4.2.3. Slab types
Two types of slabs are defined in the prototype software’s
database; ST1 and ST2. The layers and layer thickness of the roof
types are presented in the Table 5. Rock wool is preferred for in-
sulation.
4.2.4. Window types
A database of glass and frame combinations was created for
the windows. Frames made of aluminum, polyvinyl chloride, wood
and glass types with different thicknesses and coatings are defined
in the database. In addition, Solar Heat Gain Coefficient (SHGC),
 S. Yigit, B. Ozorhon / Energy & Buildings 178 (2018) 216–227
Table 5
Roof types and layer thicknesses.
Type Layers Thickness (mm)
Slab type I Screed 50
Foundation slab 500
Protective concrete 50
Blockage 150
Slab type II Screed 50
Rock wool 40–100
Foundation slab 500
Protective concrete 50
Blockage 150
which are required to calculate amount of solar radiation transmit-
ting through the windows are defined in the database. 54 different
window type combinations are obtained and inserted to prototype
database by using the elements and properties presented in the
Table 6.
4.3. Meteorological data
The air temperature and humidity data required for the build-
ing energy simulation are acquired from Turkish State Meteorolog-
ical Service. Monthly average temperature, average humidity ratios
and standard deviations of the data are presented in the Table 7.
4.4. Genetic algorithm parameters
The parameters of the GA are determined according to build-
ings geometry and prototype database. Considering the land con-
straints and dense housing in İstanbul, orientation and geometry
of the building are fixed. The CTFs for each building components
(walls, roofs and slabs) were calculated prior to the optimization
process and for each component a reference number was given.
The design variables and their upper and lower bounds are pre-
sented in Table 8. Considering that the reference building is a resi-
dential building, the windows without joints are removed from the
windows lists by setting window types lower bound to 10.
GA toolbox options are also determined for optimization pro-
cess. The determination of the options can only be set according
to user’s experience and problem size. The determined GA options
inserted into E-Mat user interface form are presented in Table 9.
After making a couple of trial runs, the size of the population and
maximum number of generations is determined to be 100. The se-
lection function, elitism coefficient, crossover function, crossover
probability coefficient and mutation function are set according to
user experience and Matlab recommendations. The objective of
the optimization problem is the minimization of total heating and
cooling energy consumption.
4.5. Case planning
The calculated lowest and highest initial cost for the reference
building is calculated to be 850,441TL and 1,306,200TL respectively.
The total initial cost of the building envelope is taken as a con-
straint and the budget is assumed to be 1.05, 1.10, 1.25 and 1.40
times the lowest initial cost possible for each case. The aim is to
determine a configuration with lowest energy consumption pos-
sible with a budget constraint. The constraints for each case are
presented in Table 10.
5. Analysis results
The developed software run 10,0 0 0 thermal simulations and
took about 3.5 h to complete optimization process. Reference
building optimization results for each case is presented in Table
223
11 and the drawings of the selected structure cross sections are
presented in Figs. S.3a, S.3b and S.3c. In the following sections the
results of analysis for each case are described.
5.1. Case 1 analysis results
The initial budget constraint for case 1 is assumed to be
234,990 $ and the aim is to minimize total energy consumed to
maintain the thermal comfort in the reference building. The results
of the genetic algorithm optimization process are presented in Fig.
S.2a. At first 10 generations the genetic algorithm optimization tool
could not find results meeting the initial cost constraint. After 10th
generation, the toolbox started to show results for the optimization
problem. From generation 60 to 100 the best individual in the pop-
ulation has shown nearly no improvements. The genetic algorithm
tool exceeded the 100 generations limit and terminated.
Optimal building envelope configuration for case 1 is presented
in Table 11. The windows area facing to 225° and 405° are about
92.3 m2 . 92.16 m2 is the lower boundary condition for these two
surfaces. The windows facing to 315° and 135° are 64.57 m2 and
103.53 m2 respectively. The software selected nearly the smallest
windows for the reference building except for the surface directed
to 135°. In addition, for the roof and walls the software selected
the highest absorbance values 0.93 and 0.99 respectively. The ref-
erence building optimal window type is selected to be window
type 18. The 18th window is double-glazing Low-E window with
wooden joints. Glass thickness is 8 mm, space between glasses
is 16 mm and thermal transmittance of the selected window is
2 W/m2 K. The unit price for the 18th window is 61.89 $/m2 .
152nd wall is selected to be exterior walls. The selected exte-
rior wall is a traditional brick wall with 100 mm thermal insulation
(EPS). The unit cost of the 152nd wall is 43.68 $/m2 . The software
selected roof type 16th for the reference building. The selected
roof has 95 mm XPS insulation and the unit price of the selected
roof is 180.93 $/m2 . The 1st slab is selected to be the foundation
of the reference building. The selected slab has 150 mm blockage,
50 mm protective concrete, 500 mm foundation slab and 50 mm
screed and the unit price for the selected slab is 217.56 $/m2 .
The layers of the selected building components are presented in
Table 12. The calculated yearly electricity consumption for cool-
ing is 40,957 KWh and calculated yearly natural gas consumption
for heating is 26,840 KWh. The buildings envelope total cost is
234,907 $.
5.2. Case 2 analysis results
The initial budget constraint for case 2 is assumed to be
246,180 $. The results of the genetic algorithm optimization process
are presented in Fig. S.2b. From generation 60 to 100 the best indi-
vidual in the population has shown no improvements. The average
relative change of the best value is very low between generations
60 and 100. However, the optimization process did not terminate
because tolerance function value is very low (10−8 ). The genetic
algorithm tool exceeded 100 generations limit and terminated.
Optimal building envelope configuration for case 2 is presented
in Table 11. The windows area facing to 225° and 45° are about
92.2 m2 . The windows facing to 315° and 135° are 64.37 m2 and
64.33 m2 respectively. The software selected almost the smallest
windows for all surfaces. The software selected the highest ab-
sorbance value (0.99) for the walls and lowest absorbance coef-
ficient possible (0.2) for the roof. The reference building optimal
window type is selected to be the 18th window. The 18th window
is double-glazing Low-E window with wooden joints. Glass thick-
ness is 8 mm, space between glasses is 16 mm and thermal trans-
mittance of the selected window is 2.0 W/m2 K. The unit price for
the 18th window is 61.90 $/m2 .
224 S. Yigit, B. Ozorhon / Energy & Buildings 178 (2018) 216–227

Table 6
Window types used in the database.

Frame types Glass types Space between glasses (m)

Without joints Single glazing 0.006

Double glazing 0.009
Woodwork Double glazing low-E 0.012
PVC 2c joint
PVC 3c joint
Aluminium joint 0.016
Aluminium joint with insulation bridge

Table 7
Weather temperature data used in building simulation.

Month Average temperature (°C) Std. dev. of temperature data Average humidity (%) Std. dev. of humidity data

1 6.7 4.0 81.7 11.9

2 6.9 3.7 81.9 12.0
3 8.9 3.6 82.0 11.5
4 12.1 3.6 81.4 11.5
5 17.9 4.2 79.4 10.8
6 22.3 3.3 75.7 11.6
7 25.2 3.1 75.6 10.9
8 25.4 3.1 78.5 10.5
9 21.2 3.5 80.9 11.5
10 16.3 3.7 83.1 11.5
11 12.3 3.9 82.1 12.4
12 9.2 4.6 80.7 12.1

Table 8
Determined genetic algorithm upper and lower boundaries.

Variables Lower bound Upper bound

Wall type (X1 ) 1 459

Roof type (X2 ) 1 340
Foundation type (X3 ) 1 14
Window type (X4 ) 10 54
Surface 1 window area (m2 ) (X5 ) 92.16 322.56
Surface 2 window area (m2 ) (X6 ) 92.16 322.56
Surface 3 window area (m2 ) (X7 ) 64.32 225.12
Surface 4 window area (m2 ) (X8 ) 64.32 225.12
Wall absorbance coefficient (X9 ) 0.2 1
Roof absorbance coefficient (X10 ) 0.2 1

Table 9
Determined genetic algorithm optimization options.

Genetic algorithm options Description Value

Population size Size of the population (Number of individuals) 100

Selection
Maximum generations
Elitism
Tolerance function
Crossover function
Crossover fraction
Mutation function
Selection of individuals for the next generation Stochastic uniform
Maximum number of iteration 100
How many individual in the current generation are guaranteed to survive 0.05∗ Population size
If the average relative change in the best fitness function value is less or equal to Funtol value 1.00∗ E−08
Crossover Constraint dependent
The fraction of the population created by crossover function 0.8
Mutation Constraint dependent

Table 10 roof is 198.47 $/m2 . The 2nd slab is selected to be the foundation of
Initial budget constraints for each test
the reference building, which has 150 mm blockage, 100 mm lean
case.
concrete, 500 mm foundation slab, 40 mm rock wool and 50 mm
Test case Constraints screed. The unit price of the selected slab is 228.92 $/m2 . The
Case 1 Initial cost ≤ 234,990 $ layers of the selected building components are presented in the
Case 2 Initial cost ≤ 246,180 $ Table 13. The calculated yearly electricity consumption for cooling
Case 3 Initial cost ≤ 279,750 $
is 46,716 KWh and calculated yearly natural gas consumption for
Case 4 Initial Ccost ≤ 313,320 $
heating is 18,864 KWh. The buildings envelope initial total cost is
246,180 $.

153rd wall is selected to be the exterior walls. 153rd wall is a
traditional brick wall with 100 mm thermal insulation (EPS). The
unit cost of the wall is 43.89 $/m2 . The software selected roof type
289th for the reference building. The selected roof has 90 mm XPS
and 100 mm rock wool for insulation. The unit price of the selected
5.3. Case 3 analysis results
The initial budget constraint for case 3 is assumed to be
270,342 $. The results of the genetic algorithm optimization pro-
cess are presented in Fig. S.2c. From generation 60 to 94 the best
 S. Yigit, B. Ozorhon / Energy & Buildings 178 (2018) 216–227 225

Table 11 Table 14
Optimization results for reference building. The layers of the selected building components.

Building components Selected components/Values Type Layers Thicknesses (mm)

Case 1 Case 2 Case 3 Case 4
Wall type 442 Gypsum plaster 20
External wall type 152 153 442 442 Brick wall 100
Roof types 16 289 289 306 Cement plaster 20
Slab type 1 2 2 2 Rock wool 100
Window type 18 18 36 36 Thermal insulation (EPS) 100
Surface 1 - window area (m2 ) 92.21 92.18 92.16 92.18 Cement plaster 20
Surface 2 - window area (m2 ) 92.31 92.19 92.16 92.17 Roof type 289 Gypsum plaster 20
Surface 3 - window area (m2 ) 64.57 64.37 64.32 64.33 Concrete slab 200
Surface 4 - window area (m2 ) 103.53 64.33 64.32 64.32 Rock wool 100
Wall absorbance coefficient 0.99 0.99 0.47 0.72 XPS foamboard 90
Roof absorbance coefficient 0.93 0.21 0.2 0.23 Protective concrete 50
Slab type 2 Screed 50
Rock wool 40
Table 12 Foundation slab 500
The layers of the selected building components. Lean concrete 100
Blockage 150
Type Layers Thicknesses (mm)

Wall type 152 Gypsum plaster 20

Brick wall 100
Cement plaster 20
Thermal insulation (EPS) 95
lation and 100 mm rock wool. The unit price of the selected roof is
Cement plaster 20 198.47 $/m2 . The selected slab has 150 mm blockage, 100 mm lean
Roof type 16 Gypsum plaster 20 concrete, 500 mm foundation slab, 40 mm rock wool and 50 mm
Concrete slab 200 screed. The unit price of the selected slab is 228.92 $/m2 . The
XPS foamboard 95
layers of the selected building components are presented in the
Protective concrete 50
Gravel 30 Table 14. The calculated yearly electricity consumption for cooling
Slab type 1 Screed 50 is 42,413 KWh and calculated yearly natural gas consumption for
Foundation slab 500 heating is 20,900 KWh. The buildings envelope initial total cost
Protective concrete 50 is 270,342 $. Although the initial budget is increased considerably,
Blockage 150
there is not a considerable reduction in the energy consumed for
heating and cooling purposes.
Table 13
The layers of the selected building components.

Type Layers Thicknesses (mm)

Wall type 153 Gypsumplaster 20 5.4. Case 4 analysis results

Brick wall 100
Cement plaster 20 The budget constraint for case 4 is assumed to be 313,320 $. The
Thermal insulation (EPS) 100
Cement plaster 20
results of the genetic algorithm optimization process are presented
Roof type 289 Gypsum plaster 20 in Fig. S.2d. Similar to case 1 and case 2 the genetic algorithm tool
Concrete slab 200 exceeded 100 generations limit and terminated.
Rock wool 100 Optimal building envelope configuration for case 4 is presented
XPS foamboard 90
in Table 11. The windows area facing to 225° and 45° are 92.18 m2
Protective concrete 50
Slab type 2 Screed 50 and 92.17 m2 respectively. The windows facing to 315° and 135°
Rock wool 40 are about 64.30 m2 . The 36th window is selected for the refer-
Foundation slab 500 ence building. The 36th window is double-glazing Low-E window
Lean concrete 100 with PVC 3c joints. Glass thickness is 8 mm, space between glasses
Blockage 150
is 16 mm and thermal transmittance of the selected window is
1.9 W/m2 K. The unit price for the 36th window is 66.87 $/m2 .
In this case 442nd wall is selected. The selected wall is a
member of the population has shown no improvements. The op- traditional brick wall with 100 mm thermal insulation (EPS) and
timization process is terminated at generation 94 because aver- 100 mm rock wool. The unit cost of the wall is 61.99 $/m2 . The
age relative change of the best value was lower than the tolerance software selected roof type 306th and the selected roof has 95 mm
function (10−8 ). XPS and 100 mm rock wool for insulation. The layer properties
Optimal building envelope configuration for case 3 is presented are presented in Table 15. The unit price for the selected roof
in Table 11. The windows area facing to 225° and 45° are 92.16 m2 , is 198.83 $/m2 . The 2nd slab is selected to be the foundation of
which is the lowest window area possible. The windows directed the reference building. The selected slab has 150 mm blockage,
to 315° and 135° are 64.32 m2 . 36th window is selected to be 10 0 mm lean concrete, 50 0 mm foundation slab, 40 mm rock wool
the reference building optimal window type. The 36th window is and 50 mm screed. The layers of the selected building components
double-glazing Low-E window with PVC 3c joints. Glass thickness are presented in the Table 15. The unit price of the selected slab
is 8 mm, space between glasses is 16 mm and thermal transmit- is 228.92 $/m2 . The calculated yearly electricity consumption for
tance of the selected window is 1.9 W/m2 K. The unit price for the cooling is 43,924 KWh and calculated yearly natural gas consump-
36th window is 66.87 $/m2 . tion for heating is 19,346 KWh. The buildings envelope total initial
442nd wall is selected to be the exterior wall, which is a cost is 270,500 $. The results of the case 4 are very similar to case
traditional brick wall with 100 mm thermal insulation (EPS) and 3. Although the initial budget is increased considerably, there is
100 mm rock wool. The unit cost of the 442nd wall is 61.99 $/m2 . not a considerable reduction in the energy consumption for heat-
The software selected roof type 289th, which has 90 mm XPS insu- ing and cooling purposes.
226 S. Yigit, B. Ozorhon / Energy & Buildings 178 (2018) 216–227

Table 15
The layers of the selected building components.

Type Layers Thicknesses (mm)

Wall type 442 Gypsum plaster 20

Brick wall 100
Cement plaster 20
Rock wool 100
Thermal insulation (EPS) 100
Cement plaster 20
Roof type 306 Gypsum plaster 20
Concrete slab 200
Rock wool 100
XPS foamboard 95
Protective concrete 50
Slab type 2 Screed 50
Rock wool 40
Foundation slab 500
Lean concrete 100
Blockage 150
Fig. 5. Initial cost - energy consumption relation.
Table 16 cial energy simulation software are not designed for optimization
Results of the test cases.
processes, which require numerous analyses in a short time pe-
Cases Case 1 Case 2 Case 3 Case 4 riod. The most commonly used energy simulation softwares are
Budget constraint ($) 234,990 246,180 270,342 313,320 open source programs, which are not user friendly and require
Cooling load (KWh) 40,957 46,716 42,413 43,924 high expertise [40]. The popular software such as TRNSYS and En-
Heating load (KWh) 26,840 18,864 20,900 19,346 ergyPlus require 3rd party softwares for data input [36, 41]. The
Configuration cost ($) 234,907 246,180 270,342 270,500
developed prototype software requires minimum amount of expe-
rience in building energy simulation. Except from the EXCEL form
which is used as graphical user interface software does not require
6. Discussions and evaluation 3rd party software. Detailed software require a very detailed CAD
model for the simulations and methods implementing surrogate
In literature, the major part of the studies in simulation-based models require excessive amount of input data for training [4,14–
optimization methods couple a commercial building energy sim- 17,25]. The software offered requires least amount of input data,
ulation software and an optimization software to carry out opti- which greatly saves time for designers. The designers are required
mization analysis [1, 4, 14-16]. On the other hand, in this study to input only the data presented in Fig. 3. Using simplified models
a completely new dynamic building energy simulation module is greatly reduces the time and expertise required for the optimiza-
developed. Therefore, before using the developed energy simula- tion process. However, combination of optimization software with
tion software, validation runs are required to be carried out. Us- simplified simulation software requires input and output files to
ing a prototype building analyzed by an impartial expert increases communicate with each other [20, 23]. Besides, using too simpli-
the credibility of the validation analysis. Besides, EnergyPlus is the fied static methods lack accuracy [24]. The developed software was
most popular software among the researchers studying in simu- developed on Matlab, new modules for HVAC and renewable en-
lation based optimization methods [6]. Therefore, EnergyPlus is a ergy, can be combined with the software without a problem which
suitable choice for comparison. demonstrates the flexibility of the proposed method.
In this study, a test case process has been conducted includ- The results of the testing and validation analysis showed that
ing 4 cases to test the developed software. The results of the case the developed software is a reliable tool for solving complex design
runs are presented in Table 16. It took 3.5–12 h to complete op- problems. Unlike other studies in literature, in this study a stan-
timization processes depending on the constraint ranges. As men- dalone simulation-based optimization software is developed. Us-
tioned in the previous sections, compared to other studies the op- ing a standalone software eliminated coupling problems. Develop-
timization time periods significantly reduced [10, 16, 17, 25]. The ing a tailor-made energy simulation software for, building energy
energy consumption of the building is minimized under a budget optimization significantly reduces optimization time period. In ad-
constraint. Similar to other studies, the test cases results showed dition, The results obtained from testing process are in line with
that the optimal design configurations cost only a small amount previous studies in the literature. Therefore, the results of the test
more than the minimum initial cost possible [4, 39]. cases indicate that the proposed methodology and the software are
Consistent with the results of other studies, the optimization successful.
analysis carried out in the test cases showed that the software se-
lected small and high efficiency windows [1, 4]. In addition, in par- 7. Conclusions
allel with previous studies, in the external walls and roof the soft-
ware selected thickest insulation possible [10]. The reason is that, The purpose of this study is to propose a plain, flexible, effi-
selecting thicker insulations does not have a significant impact on cient and user-friendly methodology that supports the designers at
initial cost, which means thick insulations have a high benefit-cost the preliminary design phase of the building projects to make de-
ratio. The relation between Initial cost and energy consumption is cisions related to energy performance of the buildings. Designers
presented in Fig. 5. The results of the testing analysis demonstrated aim to reduce life cycle costs and energy consumptions to design
that initial budgets, higher than approximately 1.25 times the low- cost effective and greener buildings. However, while designing cost
est initial cost possible has no effect on the energy performance of effective and green buildings designers must satisfy the require-
the buildings. In case 3 the budget is 1.25 times the lowest budget ments of governmental regulations and users of the buildings. In
possible therefore, the case 3 is determined to be the preferable this study, a prototype software package is developed to optimize
case of this study. energy performance of the buildings. The developed prototype uti-
Commercial energy simulation software packages carry out very lizes a genetic algorithm optimization tool to optimize buildings
detailed thermal analyses which have long run times. The commer- considering energy performance.
 S. Yigit, B. Ozorhon / Energy & Buildings 178 (2018) 216–227
The developed prototype was implemented in 4 test case runs.
A real 5-storey building located in Istanbul was selected for test
cases. In each case, a different level of budget constraint was used.
Test cases demonstrated the implementation of the software. The
test cases demonstrated utilization of the developed software and
the resultant building configurations were evaluated. By imple-
menting the test cases, weaknesses and strengths of the proposed
methodology were identified. The results of the test cases demon-
strated that, the proposed methodology and the software can be
an effective tool in aiding the designers, at the preliminary design
phase of the buildings.
Although there are similar studies on the subject, there are still
difficulties in coupling efficiency considering time, labor, practical-
ity and flexibility. The developed prototype is a user-friendly soft-
ware package, which requires least amount of input data. Previ-
ous studies have predominantly combined a number of software
to create their models. Gathering both simulation and optimization
tools on the same platform (Matlab) eliminated the coupling prob-
lems and greatly sped up the optimization process. The prototype
is designed to require least amount of input data so; simulation
preparations are significantly shortened. Therefore, the developed
software is suitable for evaluating and optimizing large number of
buildings in a short period of time. Using widely known software
(MS Excel) for database creation is a major advantage. In addition,
a single database can be used in several projects. Unlike other en-
ergy simulation and optimization software packages, the developed
software does not require any expertise to implement. Compared
to other studies, a very comprehensive validation analysis was con-
ducted in this study.
To sum up, most of the design decisions affecting the energy
performance of the buildings are made in the preliminary design
phase of the buildings. Therefore, the design variables affecting
the building energy performance must be carefully decided. Imple-
menting simulation-based software at the preliminary design stage
of the building is a great help to designers. Further studies can
be conducted to increase accuracy of the building energy simula-
tion module by implementing more detailed models for infiltra-
tion, ventilation and internal loads. In addition, parallel computing
methods can be implemented to speed up the optimization pro-
cess and reduce optimization time period.
Supplementary materials
Supplementary material associated with this article can be
found, in the online version, at doi: 10.1016/j.enbuild.2018.08.045.
References
[1] W. Wang, R. Zmeureanu, H. Rivard, Applying multi-objective genetic algo-
rithms in green building design optimization, Build. Environ. 40 (11) (2005)
1512–1525.
[2] D. Castro-Lacouture, J.A. Sefair, L. Flórez, A.L. Medaglia, Optimization model for
the selection of materials using a LEED-based green building rating system in
Colombia, Build. Environ. 44 (6) (2009) 1162–1170.
[3] R. Pulselli, E. Simoncini, F. Pulselli, S. Bastianoni, Emergy analysis of building
manufacturing, maintenance and use: em-building indices to evaluate housing
sustainability, Energy Build. 39 (5) (2007) 620–628.
[4] M. Ferrara, E. Fabrizio, J. Virgone, M. Filippi, A simulation-based optimization
method for cost-optimal analysis of nearly Zero Energy Buildings, Energy Build.
84 (2014) 442–457.
[5] F. Yang, D. Bouchlaghem, Genetic algorithm-based multiobjective optimization
for building design, Architect. Eng. Des. Manag. 6 (1) (2010) 68–82.
[6] A.T. Nguyen, S. Reiter, P. Rigo, A review on simulation-based optimization
methods applied to building performance analysis, Appl. Energy 113 (2014)
1043–1058.
[7] A. Wilson, A. Templeman, An approach to the optimum thermal design of of-
fice buildings, Build. Environ. 11 (1) (1976) 39–50.
[8] L.G. Caldas, L.K. Norford, A design optimization tool based on a genetic algo-
rithm, Autom. Constr. 11 (2) (2002) 173–184.
[9] N. Djuric, V. Novakovic, J. Holst, Z. Mitrovic, Optimization of energy consump-
tion in buildings with hydronic heating systems considering thermal comfort
by use of computer-based tools, Energy Build. 39 (4) (2007) 471–477.
227
[10] A. Hasan, M. Vuolle, Minimisation of life cycle cost of a detached house us-
ing combined simulation and optimization, Build. Environ. 43 (12) (2008)
2022–2034.
[11] F.P. Chantrelle, H. Lahmidi, W. Keilholz, M. El Mankibi, P. Michel, Development
of a multicriteria tool for optimizing the renovation of buildings, Appl. Energy
88 (4) (2011) 1386–1394.
[12] S. Bambrook, A. Sproul, D. Jacob, Design optimisation for a low energy home
in Sydney, Energy Build. 43 (7) (2011) 1702–1711.
[13] M. Hamdy, A. Hasan, K. Siren, Applying a multi-objective optimization ap-
proach for design of low-emission cost-effective dwellings, Build. Environ. 46
(1) (2011) 109–123.
[14] F. Ascione, N. Bianco, C. De Stasio, G.M. Mauro, G.P. Vanoli, A new methodol-
ogy for cost-optimal analysis by means of the multi-objective optimization of
building energy performance, Energy Build. 88 (2015) 78–90.
[15] J. Kim, T. Hong, J. Jeong, C. Koo, K. Jeong, An optimization model for select-
ing the optimal green systems by considering the thermal comfort and energy
consumption, Appl. Energy 169 (2016) 682–695.
[16] E. Asadi, M.G. da Silva, C.H. Antunes, L. Dias, L. Glicksman, Multi-objective op-
timization for building retrofit: A model using genetic algorithm and artificial
neural network and an application, Energy Build. 81 (2014) 444–456.
[17] L Magnier, F Haghighat, Multiobjective optimization of building design using
TRNSYS simulations, genetic algorithm, and artificial neural network, Build En-
viron. 45 (3) (2010) 739–746.
[18] N. Bouchlaghem, K. Letherman, Numerical optimization applied to the thermal
design of buildings, Build. Environ. 25 (2) (1990) 117–124.
[19] J.A. Wright, H.A. Loosemore, R. Farmani, Optimization of building thermal de-
sign and control by multi-criterion genetic algorithm, Energy Build. 34 (9)
(2002) 959–972.
[20] E. Znouda, N. Ghrab-Morcos, A. Hadj-Alouane, Optimization of Mediterranean
building design using genetic algorithms, Energy Build. 39 (2) (2007) 148–153.
[21] M. Sahu, B. Bhattacharjee, S. Kaushik, Thermal design of air-conditioned build-
ing for tropical climate using admittance method and genetic algorithm, En-
ergy Build. 53 (2012) 6.
[22] X. Gong, Y. Akashi, D. Sumiyoshi, Optimization of passive design measures
for residential buildings in different Chinese areas, Build. Environ. 58 (2012)
46–57.
[23] M. Hamdy, A. Hasan, K. Siren, A multi-stage optimization method for cost op-
timal and nearly-zero-energy building solutions in line with the EPBD-recast
2010, Energy Build. 56 (2013) 189–203.
[24] S.N. Murray, B.P. Walsh, D. Kelliher, D. O’Sullivan, Multi-variable optimization
of thermal energy efficiency retrofitting of buildings using static modelling and
genetic algorithms: a case study, Build. Environ. 75 (2014) 98–107.
[25] L. Zhou, F. Haghighat, Optimization of ventilation system design and opera-
tion in office environment, Part I: methodology, Build. Environ. 44 (4) (2009)
651–656.
[26] L. Yang, H. Yan, J.C. Lam, Thermal comfort and building energy consumption
implications a review, Appl. Energy 115 (2014) 164–173.
[27] Q.J. Kwong, N.M. Adam, B. Sahari, Thermal comfort assessment and potential
for energy efficiency enhancement in modern tropical buildings: a review, En-
ergy Build. 68 (2014) 547–557.
[28] C.O. Pedersen, D.E. Fisher, R.J. Liesen, Development of a heat balance procedure
for calculating cooling loads, American Society of Heating, Refrigerating and
Air-Conditioning Engineers, Inc., 1997.
[29] D. Coakley, P. Raftery, M. Keane, A review of methods to match building energy
simulation models to measured data, Renew. Sustan. Energy Rev. 37 (2014)
123–141.
[30] E. ASHRAE, ASHRAE Handbook Fundamentals, 30329, American Society of
Heating, Refrigerating and Air-Conditioning Engineers, Atlanta, GA, 2013 ISBN:
978-1-936504-46-6.
[31] T. Kusuda, NBSLD, Computer Program for Heating and Cooling Loads in Build-
ings, Center for Building Technology, Washington, DC, 1974 Final Report Na-
tional Bureau of Standards.
[32] GN Walton, Thermal Analysis Research Program Reference Manual, National
Bureau of Standards, 1983.
[33] I. Lu, D. Fisher, Application of conduction transfer functions and periodic re-
sponse factors in cooling load calculation procedures, ASHRAE Trans. 110 (2)
(2004).
[34] J.D. Spitler, Load Calculation Application Manual SI Edition, American Society
of Heating Refrigerating and Air-Conditioning Engineers, 2013.
[35] M. Palonen, A. Hasan, K. Siren, A genetic algorithm for optimization of building
envelope and HVAC system parameters, in: Proc. of the 11th IBPSA Conference,
Glasgow, Scotland, 2009.
[36] Energy Plus, U.S. Department of Energy - EnergyPlus, 2017 https://energyplus.
net/ . Access Date: December 2017.
[37] DOE, U.S. Department of Energy Commercial Prototype Building Models, 2017
https://www.energycodes.gov/development/commercial/prototypemodels.
[38] TUIK, Turkish Statistical Institute - Building Occupancy Permit Data, 2017 http:
//www.tuik.gov.tr/ , accessed: September.
[39] J. Morrissey, R. Horne, Life cycle cost implications of energy efficiency mea-
sures in new residential buildings, Energy Build. 43 (4) (2011) 915–924.
[40] D. Zhu, et al., A detailed loads comparison of three building energy modeling
programs: EnergyPlus, DeST and DOE-2.1 E, Build. Simul. 6 (3) (2013) Tsinghua
University Press.
[41] Transient System Simulation Tool, https://trnsys.com/. Access Date: December
2017.