Beruflich Dokumente
Kultur Dokumente
Applied Energy
journal homepage: www.elsevier.com/locate/apenergy
h i g h l i g h t s
A site location potential model for urban energy supply planning is proposed.
The model deduces urban energy plants’ suitability and renewable energy availability.
The model was constructed as an ANN with an input–output dataset from the E-GIS DB.
The model produces planning supportive materials, such as energy potential maps.
This model addresses the technical methodology for urban energy supply planning.
a r t i c l e i n f o a b s t r a c t
Article history: This study proposes a site location potential model for urban energy supply plants and renewable energy
Received 28 May 2013 availability using an environment and energy geographical information system (E-GIS) database (DB) and
Received in revised form 23 December 2013 an artificial neural network (ANN). This model addresses the technical methodology of examining the
Accepted 29 December 2013
potential for the suitability of urban energy supply plants and renewable energy latency in a region for
Available online 25 January 2014
support material for urban energy supply planning in the draft plan development stage. The applicability
of this model is examined by applying it for a planned city in the Republic of Korea, where urban planning
Keywords:
is in process. The results from this study are as follows:
Low carbon green city
Urban planning support system (PSS) (1) The E-GIS DB was integrated with geography, climate, and energy-related information to construct
Energy supply planning an ANN model that can manage, in an integrated manner, the factors that affect the site location of
Environment and energy geographic
the energy supply plants.
information system (E-GIS)
(2) The input dataset included the topography, land cover classification, accessibility, water usability,
Artificial neural network (ANN)
Energy potential map
and energy demand, and the target dataset included the system capacity of domestically installed
energy supply plants.
(3) The site location potential model of the ANN for the urban energy supply plants and renewable
energy availability was deduced, and the Levenberg–Marquardt (trainlm) and scaled conjugate
gradient (trainscg) algorithms were used. The potentiality class map was constructed for 10 types
of energy supply systems and renewable energy resources.
(4) The applicability of this energy model was tested in the Gwang-myung/Si-heung public housing
district area, a domestic ‘planned city’ of the Republic of Korea. The most appropriate urban energy
supply systems for the research area were considered to be the general hydraulic power and solar
power based on the topographic conditions and profitable locations for solar resources in Korea.
Wind power generation was found to be the least suitable.
(5) In terms of the wind energy potential, the technical wind power generation by horizontal – axis
wind turbines is unattainable even in the area that has the maximum wind speed, and at least
a 10-kW rated power wind turbine should be installed for vertical – axis wind turbines in the
research area of interest. In terms of the solar energy potential, the maximum electric power gen-
eration potential is 413, 186 MJ/monthmesh, which is applied by mono-crystalline bulk PV.
⇑ Corresponding author. Tel.: +82 51 200 7609; fax: +82 51 294 2256.
E-mail address: jjyee@dau.ac.kr (J.-J. Yee).
0306-2619/$ - see front matter Ó 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.apenergy.2013.12.060
100 I.-A. Yeo, J.-J. Yee / Applied Energy 119 (2014) 99–117
1. Introduction supply planning, which can be used for decision making about
which types of energy plants are suitable on which locations and
According to the ‘World Urbanization Prospects’ report (2011 which types of natural energy resources are available in urban
revision) of the United Nations [1], 50% of the total world popula- areas.
tion resides in urban areas, and this value is higher in most devel- Based on these perspectives, the technology discussed in this
oped countries. It is predicted that the urbanization trend will paper proposes the development of a site location supportive
continue. The speed and degree of urbanization depend on the re- map for urban energy supply plants and renewable energy avail-
gion and country (Table 1). Because urban development and ability, which is one of the site location assistive technologies for
growth cannot be stopped, it is desirable to create an environmen- supporting energy supply planning. The scope of this study is to
tally friendly city that functions as a city in the urban life cycle in apply the environment and energy geographic information
terms of ‘sustainable development’; this city would use efficient (E-GIS) DB1 to artificial neural network techniques and to analyze
energy, circulate natural resources, and provide a comfortable the potential of the site location for various energy supply plants
living environment. and the availability of renewable energy resources in the urban area.
From this perspective, an environmentally friendly city can After taking these perspectives into consideration, this study
achieve comprehensive low energy consumption by developing develops a site location potential model for urban energy supply
an energy structure of compact energy demand and reasonable en- plants and renewable energy availability, which is a general-pur-
ergy supply in the urban planning stage and consistent control for pose urban energy supply planning support model for an environ-
demand side management (DSM), such as load management and mentally friendly city. This model will address the technical
energy efficiency, and supply side management (SSM) in the urban methodology of examining the potential for the suitability of urban
operation stage. Considering that the previously developed struc- energy supply plants and renewable energy latency in a region by
ture of the urban energy demand and supply is difficult to change making the site location for an urban energy supply planning sup-
or improve, demand-driven energy planning is needed at the urban port material in the draft plan development stage, which is the ba-
planning stage. In other words, the foundation of the lower energy sis for optimized planning through further decision making in
structure should be set in the urban planning stage by predicting terms of economic and urban policies. Potential practical users of
the energy demand to reduce excessive energy production. To this model are urban planners, energy planners, or engineers,
establish energy supply planning at this stage, urban energy supply who are supplied with potential maps of the urban energy supply
facility planning should optimally combine an economical, stable, plants and renewable energy availability as a mesh unit. Other po-
and environmentally friendly supply, and the energy demand tential users include people in the energy business or citizens be-
should be satisfied. The energy supply facility’s location should cause the issue of locating energy supply facility site is related to
be based on the examination of the present condition of the urban economic and urban policies.
energy supply facility in the existing city and the evaluation of the The research is presented in 3 stages. Section 2 examines the
feasibility or potential of various urban energy resources to maxi- existing research trends regarding the regional energy potential
mize the utilization of regional natural resources of a city and clean and site location planning support systems; it investigates the cur-
energy resources. rent urban supply planning of the Republic of Korea and sets the
To address energy demand stability and to make energy supply scope of the methodology of environmentally friendly urban en-
facilities economically and environmentally friendly through en- ergy supply planning that this study proposes. Section 3 proposes
ergy supply planning, a link between energy scenarios and site an artificial neural network (ANN) model of the potential of urban
location support technology is required. Based on the energy sce- energy supply plants and renewable energy availability by inte-
narios, alternative scenarios in terms of the urban policy and econ- grating the current energy-related legal system and standards
omy are suggested, and based on the site location supportive and energy plant facilities, which affect energy facility site plan-
technology, the availability of urban natural resources and energy ning, into the E-GIS DB of the Republic of Korea. The details of
is evaluated in terms of the technical aspects. Considering that eco- the E-GIS construction model, including the algorithms employed,
nomic or policy evaluation is possible for alternative energy sce- are described, and Section 4 describes the results of testing the
narios based on a technical review of whether the site location ANN model of the potential for energy supply plants and renew-
planning of urban energy supply facilities suffice for efficient land able energy availability for a case study area by utilizing the
use and environmentally friendly energy resources, the major pri- E-GIS DB of a planned city in the Republic of Korea.
ority in the planning process is to collect information on urban en-
ergy resource availability and site-location potential in terms of
the technical aspects. Materials such as the ‘potential map of the 2. Materials and scope of the article
site location for energy supply plants and natural energy resources
availability’ are especially helpful for integrated urban energy 2.1. Literature review
studied at the regional and national levels. According to existing For the wind energy technical potential, Nguyen [9] studied the
studies, the definition of ‘energy potential’ varies. Some definitions wind energy density in Vietnam under a domestic wind technology
include resource availability, theoretically determined energy, the condition, and Aydin et al. [10] derived the priority site map for
energy output by a specific technology, and economically feasible wind turbine construction planning while considering the wind
energy. The National Renewable Energy Laboratory (NREL) charac- energy potential and acceptable land in western Turkey. Hossain
terizes the energy potential as ‘resource’, ‘technical’, ‘economic’, et al. [11] constructed wind energy potential maps that were based
and ‘market’, forming a pyramid structure [4]. The ‘resource poten- on parameters such as the yearly mean wind speed at the hub
tial’ is found by the theoretical environmental physics potential height, the Weibull distribution of the wind speed, the energy out-
and energy carriers in the resource. The ‘technical potential’ is put of the turbine, and the plant load factor (PLF), at a resolution of
found by the resource potential that is limited by the performances 1 km, to develop a wind farm in India. Mari et al. [12] made wind
of the energy systems, topography, and land cover factors. The energy potential maps that were based on the yearly mean wind
‘economic potential’ is found by the technical potential that is lim- speed, full load hours (FLH), and the annual energy production
ited by the cost of the applied technology and fuel. The ‘market po- (AEP) for a resolution of 2 km in the web-oriented GIS-based DSS
tential’ is found by the economic potential that is limited by the developed as a part of the WIND-GIS project for supporting the re-
effect of the policy implementation, regulatory restrictions, invest- gional renewable energy policy. Sliz-Szkliniarz and Vogt [13] found
ment, and regional energy resources factors. Here, the upper the available locations for wind farm sites based on the spatial,
potential can be reasonably reached by establishing a lower poten- ecological policies, and the restriction standards of the potential
tial, which makes it easier to use efficient energy from natural factors of the wind resources.
resources. Larentis et al. [14] examined the hydro energy potential with
Extensive studies on estimating the resource potential based on available land use area by exploring the dam location potential
resource investigation and finding the technical potential applied based on the regional stream flow data, selecting a site for hydro
by the statistics of energy-related technology and information on power systems (dam axis spotting and powerhouse spotting) and
topographic limitations have been conducted, mainly in the United optimizing the system allocation.
States and Europe. There have been studies on the evaluation of the For the geothermal energy technical potential, Ondreka et al.
economy under specific technical conditions and energy system [15] presented the specific heat extraction of borehole heat
allocation by least cost analysis. A natural/renewable resource po- exchangers (BHEs) that were influenced by subsurface conditions
tential map is usually constructed for a national project, whereas a and the use of BHEs with a heat pump for heating. Nam and Ooka
technical potential map reflects technical characteristics at the re- [16] presented the methodology of finding the energy potential
gional level. from a system through the thermal properties of the ground and
The NREL developed the renewable energy resource distribu- the heat extraction rate for optimizing the system design for a
tion of solar, wind, geothermal, hydrogen, and biomass per state ground source heat pump (GSHP) and derived the potential map
or county unit and created resource potential maps that utilize of the geothermal resources (geology and ground water velocity,
boundary files such as political, infrastructure/utility, federal lands, ground water level and pumping regulation) in Japan and of the
parks, census data, and natural geographic features using Google location of available wells for direct heat exchange.
Maps. The technical potential map includes information on avail- Studies on biomass, where many research studies have been
able land or urban open space, the system capacity potential, and conducted because of the diversity of resources, have revolved
the energy generation potential, which finds the renewable energy around resource investigation rather than technical aspects. For
resources potential, system power, topographical limitations, and the bio energy technical potential, Perpiñá et al. [17] examined
environment and land use constraints. the distribution of the agricultural and forest biomass residues
For decision making that is related to urban energy supply with a resolution of 1 km and determined the potential location
planning, the materials on urban resources and the technical po- of the bioenergy tapping region and the bioenergy plant. Ragaglini
tential should be supported. The absence of these materials is det- et al. [18] studied land suitability for biodiesel production by sun-
rimental when an integrated urban energy planning process must flowers by considering potential sunflower yields, the energy effi-
be systemized. Resources and the technical potential are indis- ciency, the supply chain, and cost information for Tuscany in Italy.
pensable factors for estimating the economic potential, which is Hiloidhari and Baruah [19], with the help of satellite images, inves-
moderately affected, and the market potential, which is depen- tigated the potential of a decentralized electricity generation out-
dent on the national or regional economy and the political put by extracting available space for rice straw residues from rice
situation. The energy potential that is addressed in this study is croplands in the Lakhimpur district of Assam in India. Tenerelli
characterized by the technical potential, which can find the plan- and Carver [20] constructed a land capability model using multi-
ning of supportive materials with objective results using engi- criteria approaches for the soil climate and topographic factors,
neering methodology. to assess the energy conversion potential of perennial crops for
Research studies on the technical potential that was combined the Yorkshire and Humber region in the northern UK. Höhn et al.
with GIS-based spatial technology to support urban renewable en- [21] developed a bio-methane potential map by integrating
ergy and clean energy supply planning are shown in Table 2. For amounts of available feedstock and spatial distribution in the
the solar energy technical potential, Gadsden and Rylatt [5] studied southern Finland and suggested site locations of biogas plants by
the methodology of urban planners’ potential to use urban solar a road network analysis that were optimized in transportation
energy for a passive solar, solar hot water, and photovoltaic (PV) effect.
system. Janke [6] applied the multi-criteria decision-making meth- Urban natural energy planning is continuously supported by the
od to select a farm site for wind and solar energy generation in development of individual resource planning and integrated plan-
Colorado in the United States using resource and land cover infor- ning supporting tools. Bravo et al. [22], in addition to Janke and
mation. Arnette and Zobel [7] analyzed the renewable energy Arnette and Zobel [6,7], examined the system capacity and energy
potential based on the resource intensity of wind, solar, and bio- output of the renewable energy available area based on constraints
mass energy and the constraints of topography and regulations such as land use, altitude, and potential productivity, as shown in
for the Appalachian area in the southern part of the United States. Table 2. Ramachandra and Shruthi [23] developed a methodology
Charabi and Gastli [8] studied land suitability for large PV farms in to find the map of the seasonal irradiation, wind speed, biomass re-
Oman while using the fuzzy multi-criteria method. sources, and small hydropower capacity, and Belmonte et al. [24]
102 I.-A. Yeo, J.-J. Yee / Applied Energy 119 (2014) 99–117
Table 2
Research trends of the technical potential map of renewable energy.
p
[18] S: Soil group, weather conditions, and grain Biodiesel yield, NER (energy efficiency),
yield GHG saving, and land suitability
p
[19] S: Rice crop area, grain yield, residue Village wise surplus rice straw residue
production ratio, availability factor of rice biomass potential/power potential
straw, lower heating value of rice straw,
overall conversion efficiency of the power
plants, and power plant operating time
p p p
[20] W: Rainfall, degree day, soil (texture, wetness, Land capability map for energy crops
pH, and depth), and slope
S: Land cover, arable cereals, horticulture,
improved grassland, set-aside grass, and
agriculture land capability
C: Water bodies, flooding risk, erosion risk,
and protected areas
p p
[21] (Equation) S: Available feedstock energy, feedstock Bio-methane potential intensity, potential
collection area, amount of potential biomass, plant locations
theoretical installed power, annual running
time of plant
C: Weighted distance allocated to a facility
from road line
a
C: Constraints.
b
S: Selectives.
c
W: Weights.
suggested a quantitative index and map that examines irradiation, terms of the supply; and the ‘Basic Plan for New and Renewable
wind, hydro, and biomass as energy sources. Energy Technology Development and Usage/Distribution’. Local
In terms of data utilization, the remote sensing data are increas- energy planning includes details on renewable energy, integrated
ingly applied to identify resources, such as irradiation and biomass, energy, and the unused energy supply, while considering the
serving as the technological development of the GIS-based data trends and prospects of the energy demand and supply by predict-
and satellite data acquisition. Martins et al. [25] derived a solar en- ing the energy demand for local planning to induce a stable energy
ergy resource map by utilizing the satellite irradiation model of supply and environmentally friendly energy use.
Brazil and NREL irradiation data by the SWERA project and came The current urban energy planning process of South Korea fol-
up with a user-friendly renewable energy DB. Nguyen and Pearce lows the ‘Regulations regarding the establishment of an energy
[26] investigated the PV potential modeled by yearly irradiation usage plan and a consultation process’ from Article 20 [28] in the
through satellite data and suggested a methodology to support site Enforcement Ordinance of the Energy Use Rationalization Act.
selection for PV system installment based on the census, land use, According to the above regulations, business proprietors of na-
and transmission line information. Straub and Koch [27] proposed tional land development, energy development, and the construc-
a methodology to calculate the timber volume density according to tion of social overhead capital facilities (SOC) conduct energy
the classification of trees using airborne laser scanning and multi- demand forecasts and supply these plans to administrative divi-
spectral line scanner data and offered detailed information on sions at the metropolitan city, county, and district levels, which
forest residues as potential energy resources and the spatial distri- are the levels of urban management planning. The proprietor then
bution of biomass energy. establishes and submits an energy use plan, which includes an im-
pact analysis of the energy consumption on CO2 emissions, meth-
2.2. Current energy planning in the urban planning process of the ods to enhance the energy use efficiency, and methods to reduce
Republic of Korea CO2 emissions through energy use rationalization, to the Minister
of Knowledge and Economy. This process is performed to conserve
2.2.1. Urban energy planning process urban energy and manage CO2 emissions, which is achieved by
The national energy plan of the Republic of Korea constitutes submitting an energy conservation plan for each building after
the ‘Basic Plan on Energy Use Rationalization’, which induces en- the approval of the urban plan.
ergy-saving low CO2 emission in terms of the demand; the ‘Pri- According to the above-stated regulations, the energy usage
mary Energy Plan’ of oil, natural gas, coal, and overseas plan is distinguished as ‘national land development’ for a district
resources; the ‘Basic Plan on Electricity Demand and Supply’ in or site planning project, ‘energy development’ for urban facilities,
Table 3
Classification of the energy use plan establishment area.
Classification Projects
Group I: National land development Urban development
Industrial complex development
Tourism complex development
Development promotion district/integrated regional development
Group II: Energy development Mining development
Power plant construction
Gas business
Group III: SOC facility construction Harbor construction
Railway construction
Airport construction
104 I.-A. Yeo, J.-J. Yee / Applied Energy 119 (2014) 99–117
and ‘special facility’ for facilities such as the SOC facility (Table 3). power supply (UPS) system for emergency power generators, mon-
This study addresses a range of ‘‘urban planning’’ concepts in itoring/control systems, security facilities, and important computer
‘‘Group I: national land development’’ from the perspective of gen- systems for a large-scale system, is included to provide reserve
eral energy planning. The urban development ranges for special energy in an emergency or contingency situation.
use, such as industrial or tourism complexes, are not considered. The existing energy supply plan can be summarized as the pro-
cess of predicting the urban energy demand, investigating the
amount of energy supply that is linked with the energy supply net-
2.2.2. Status of urban energy supply planning work near the project area, examining the feasibility of the induc-
The current urban energy supply planning of the Republic of tion for each energy supply, and establishing a plan for a new
Korea consists of a power supply plan, a heat supply plan, and an energy supply facility. However, the following points suggest the
emergency energy supply plan, which are based on the urban en- necessities for reinforcing the energy supply plan in urban energy
ergy demand prediction. Energy supply facility planning that pro- planning with the existing plan for a new energy supply facility.
vides the city with heat and electricity is organized into a project
area when the supply reserve is insufficient from the power supply r The energy sources that satisfy the legal standard or intro-
network, and an existing heat supply network around the project duction standard have been listed without providing an appro-
area is examined with respect to heat and electricity. At this time, priate combination of the energy systems.
a plan is established and executed after investigating the feasibility s The new and renewable energy plan in a building group unit
of the construction area depending on the characteristics of the focuses on building use and is simply dependent on legally
urban development project and the consideration standard of the advised standards [29].
energy supply facilities. t An integrated consideration is insufficient for the factors that
The heat supply plan is classified as direct heat, such as inte- are essential for the optimal site location of urban energy sup-
grated energy, and fuel supply, such as LNG and LPG. The existing ply systems (i.e., the natural energy resource potential should
direct heat supply plan begins with executing feasibility studies for be considered for planning renewable energy plants).
community energy under the characteristics of urban development
and the consideration standard of the ‘Base Plan on the Integrated Therefore, this study proposed a methodology for a site location
Energy Supply’. The community energy supply (CES) assignment potential model for urban energy supply plants and renewable en-
and notification standards are given in Table 4. A new integrated ergy availability that considers the appropriate combinations of ur-
energy supply facility is hosted in or around the project area after ban energy supply facilities such as renewable energy, integrated
assigning and announcing the applied area of the integrated energy energy, and unused energy for lower energy consumption, lower
supply for the area that meets the standards, as shown in Table 4, CO2, and stable energy demand and supply in urban energy supply
and examining the CES project’s feasibility. The heat supply by the planning.
fuel is examined by the available fuel sources in the region accord-
ing to the ‘fuel use regulation notice’. The appropriate fuel sources
for the planned area are selected after investigating the environ- 2.3. Current energy planning in the urban planning process of the
mental friendliness, stability, balance of supply and demand, re- Republic of Korea
lated laws, and other properties for each fuel and heating
method applied in the energy supply facility. Depending on the se- The environmentally friendly city planning that this study pro-
lected fuel, a detailed pipe network and a pipe construction scale posed suggests that controlling urban planning together with the
are calculated after examining the feasibility of the supply facility E-GIS DB is important in addressing the issues of low energy and
and the pipe arrangement area. ecology simultaneously. Once the urban physical structure is
The power supply plan is established for substation construc- formed according to urban development, the urban climate takes
tion in the project area after investigating the power supply net- on urban climate characteristics that are different from pre-devel-
work around the project area. The power supply plan from the opment, and the real energy demand appears after the urban oper-
substation to consumers is set up, and the supply capacity and ation stage. When the urban energy demand forecasting algorithm
equipment plan for detached housing, commercial areas, public is integrated with the E-GIS DB, which is constructed in the urban
facilities, parks, and street lamps are fixed power supply targets planning stage, complete energy and environment information at
that are dependent on the uses of the power supply. the time of urban development can be acquired in advance. If pro-
The reserve power facility plan is included in the existing en- vided with a more precisely predicted value of the energy demand,
ergy supply plan and prevents a secondary accident caused by alternative energy scenarios can optimize the cost effectiveness
lights, elevators, the water supply, and air-conditioning devices. (LCC), CO2 emission reduction (LCCO2), renewable energy expan-
This plan ensures minimum security power for energy supply mea- sion, and energy self-sufficiency as substitutes for the existing
sures under emergencies such as power failure or a natural disas- baseline of the energy supply plan. When optimal site selection
ter. Energy supply facility planning, such as the installation of a for urban energy facilities is completed according to the quantita-
storage battery for a small-scale system and an uninterruptible tive evaluation of energy supply scenarios and the site location
Table 4
Standards for the evaluation of the integrated energy supply (Ministry of Commerce, Industry and Energy of the Rep. of Korea).
Classification Thermal load, max. (Gcal/h) Heat use (Gcal/year) Heat density (Gcal/km2 h)
District cooling and heating When an independent heat source facility is required 150 300,000 30
When there is a usable heat source within 5 km 30 60,000 30
CES_district cooling and heating 30 60,000 30
CES_industrial complexa – – 60
a
Supply scale (area of industry using many steams) is more than 50,000 m2; the fuel usage amount is more than 50,000 TOE; and heat production capacity > electricity
production capacity.
I.-A. Yeo, J.-J. Yee / Applied Energy 119 (2014) 99–117 105
potential for urban energy supply plants and renewable energy 3.2. Use of E-GIS DB for ANN
availability, the optimal urban design process can be used to
support decision making for urban planners, policy makers, and re- The E-GIS DB is basic information from the Energy Integrated
lated bodies. Planning Support System (EnerISS) [2,3] for environmentally
friendly urban planning, which was developed by the authors.
The E-GIS DB was initially structured in EnerISS Modeler, where
urban GIS was collected, compiled in a mesh DB, and modeled in
2D/3D space, and accumulates the forecasted value of the urban
3. Proposal of a site location potential model for the urban climate and energy demand in EnerISS Solver; it deduces the clas-
energy supply plants and renewable energy availability using ses of the urban energy supply plants‘ availability and the natural
the E-GIS DB and ANN energy generation potential in a mesh unit in EnerISS Evaluator,
which has been used to present input data for each module imple-
3.1. E-GIS DB construction for the factors for artificial neural network mentation and to obtain feedback from the outputs. The E-GIS DB
analysis supports each planning process, such as urban environment plan-
ning, urban energy demand forecasting, and energy supply plan-
In this study, the concept that was suggested to provide sup- ning, by providing the result of each module as visual maps and
portive technology when finding an optimal site location of an ur- quantitative values (Fig. 1).
ban energy supply plant in the planning stage of an The E-GIS DB addressed in this study is divided into three con-
environmentally friendly city is a site location potential model cepts, such as (a) input and output datasets for constructing ANN
and the derived planning supportive materials, which includes models to be built in Evaluator, (b) input datasets for implement-
the technical potential for a site, such as the classes of the urban ing the ANN models in a project area of urban energy supply plan-
energy supply plants‘ availability and the natural energy genera- ning, and (c) the output dataset obtained from implementing the
tion, which were deduced from the non-linear relations between ANN models in a project area. E-GIS DB as (a) input and output
various environment and energy planning variables and the power datasets for constructing ANN models is used for deriving algo-
supply capacity of the energy plants or natural energy resources. rithms of the classes of the urban energy supply plants‘ availability
To address this issue, the proposed methodologies in this study and the natural energy generation potential. Nation-wide data of
are as follows: environment and energy information with the land use level, such
First is the utilization of environment and energy planning vari- as energy supply plants, geography, meteorology, and land use, are
ables with a mesh unit in E-GIS DB, which is applied in a series of collected and linked with the mesh DB to develop ANN algorithms
processes of modeling urban space, predicting urban microclimate that support environmentally friendly energy plant planning in the
and energy demand, and supporting energy supply planning, as de- urban energy supply planning stage (Fig. 2). E-GIS DB as (b) input
scribed in Section 2.3. Second is the adoption of artificial intelli- datasets for implementing the proposed ANN models includes ur-
gence (AI) computing for deducing non-linear algorithms ban GIS and the forecasted energy demand as a result of applying
between environment and energy planning variables and the site Modeler and Solver in a project area of urban energy supply plan-
location potential. Third is the linkage of the obtained result with ning, which is constructed by region-wide data for entering into
the mesh unit that was derived from the site location potential the proposed ANN algorithm (Fig. 3). E-GIS DB as (c) output data
algorithms to a planning city by giving feedback on the E-GIS DB from the ANN model implementation is the result of applying Eval-
and planning materials, such as the visual maps and quantitative uator, which is deduced as the classes of the urban energy supply
values, which can support finding energy supply plant sites. The plants’ availability and the natural energy generation potential in a
proposed methodologies are specified through Sections 3.2–3.4. project area; these are finally suggested as planning supportive
Viewer
Integrated Fig .3
Fig .4
Output
E-GIS DB Validation
Input Energy consumption
(Spreadsheet & Visual
unit DB
Compare Maps)
Slope
Altitude
Road
Power plant capacity (KW)
0 - 250,000
Direction 250,000 - 700,000
700,000-1,200,000
Fig. 2. E-GIS DB as input and output datasets for constructing ANN models to be built in Evaluator.
Building Information Urban environment and energy planning and management Climate
information
Field Value
FID 1223
shape Polygon
ID 1224
XMIN 186400
XMAX 186600
YMIN 436000
YMAX 436200
case no. 1224 Electricity demand
water temp. 25.5
water area ratio 0
soil area ratio 0.013469
DB tree area ratio 0
DB
tree height 6
BCR 0.100891
build width 11.598349
build height 72.981503
build mater 2
build use type 2
Land and topography 200 m air cond type 3 Heat demand
roof green area 0
wall green area 0
air-cond temp 26
soil emissivity 0.2
roof emissivity 0.06
wall emissivity 0.3
Fig. 3. E-GIS DB as input datasets for implementing the ANN models in a project area of urban energy supply planning.
materials for energy supply planning by the feedback given in (b) ear relations between the environment and energy planning
(Fig. 4). information.
The E-GIS DB as (a) input and output datasets for constructing As an approach to the non-linear modeling, soft computing
ANN models to be built in Evaluator was described for ‘data prep- methods such as ANN, fuzzy, and adaptive neuro-fuzzy inference
aration’ in Section 3.3. The E-GIS DB as in (b) and (c) was used for systems have good performances on non-linear modeling with
input datasets to the proposed ANN and produced an output data- high accuracy and fast processing [35]. The ANNs are especially
set by an applicability study on the research subject area in considered to be the main field of AI, which includes expert
Section 4. systems, fuzzy systems, genetic algorithms (GA), and all competent
tools for non-linear function fitting, pattern recognition, clustering,
3.3. Development of an ANN for a site location potential model time series analysis, simulations, and optimizations. An ANN has
been practically applied in a wide range of engineering fields, such
3.3.1. Background of an ANN application in comparative studies as industrial processes, power plants, building energy management
The ability to predict future energy supply potential in the ur- systems, and electric machinery systems, for the past 20 years [36].
ban planning stage derived from the environment and energy plan- In the urban energy field, such as the prediction and optimization
ning information is essential for supporting decision making that is of energy consumption [37–46], prediction and control of power
related to energy supply planning. Most conditions that affect ur- generation systems (combustion process, thermal plant) [47–49],
ban energy consumption represent non-linearities, such as rela- wind and solar resources and power (wind speed [50–56], wind
tions between urban social and physical factors, including land energy [57], solar radiation [58–65], solar energy [66,67]), model-
use, population, building density, high rise buildings, building ing and performance prediction of PV systems (solar steam gener-
use, and urban geometry [30], the combination of urban geometry ator) [68,69], and prediction of building heating and cooling load
and natural energy acquisition [31–33], and the combination of [70–73], various research has demonstrated outstanding predic-
building elements and energy consumption [34]. Therefore, urban tion performances of applied ANN methods compared with those
energy supply potential should be well captured with the non-lin- of conventional linear or non-linear regression models [74–80].
I.-A. Yeo, J.-J. Yee / Applied Energy 119 (2014) 99–117 107
Training
(1)
(2) Validation
Inputting ANN structure Deduction of Output Data: Site location potential map
E-GIS DB of the potential
applied area into the map for the
deduced ANN applied area
(3) (4) class
1
8 5 2
16 5
3
4
5
6
7
8
Table 5
Selection of the input dataset for the ANN algorithm on the site location potential for urban energy supply plants.
Table 6
Format and scope of the input dataset for the ANN algorithm based on the E-GIS DB of the installed mesh of urban energy supply plants.
from the Korea Meteorological Administration (KMA) for 252 for 600 km2, the capital city of Korea. The temporal resolution is,
points measured in the Automatic Weather System (AWS) and over 24 h, 1 min-average air temperature for every hour, 10 min-
Automatic Synoptic Observation System (ASOS)4 over 20 years. average wind velocity for every hour, instantaneous (1 s) wind direc-
The spatial resolution of the AWS and ASSOS network of Korea is tion with maximum wind speed among 86,400 measured values, and
shown in Fig. 6. The nation-wide data of 252 AWS/ASSOS points is 1 h cumulative precipitation. The air temperature used in this study
distributed differently from region to region by population density was calculated as monthly average for the data collection period
and urbanization. For example, 30 points are distributed in Seoul city from monthly average for a year followed by the 1 min-average mea-
sured from AWS at every 3 h on time (8 times a day) during 3:00–
24:00 KST (Korean standard time), same as the officially released
4
The quality of AWS data is managed first by checking the measuring instruments statistics by Korea Meteorological Administration.
periodically to serve real-time monitoring of local dangerous weather conditions and
producing meteorological database for disaster prevention and second by imple-
menting overall quality control with handling outliers and missing values from the 3.3.3.2. ANN modeling. The characteristics of the input dataset, such
Korea Meteorological Administration. as the topography, land cover, accessibility, and water usability,
110 I.-A. Yeo, J.-J. Yee / Applied Energy 119 (2014) 99–117
Table 8
Classification of the applied target dataset grade based on the capacities of the energy supply plant (KW).
Class (Decile) Hydraulic power Thermal power Co-generation Solar power Wind power Waste heat Mixed combustion
General Small Pumping-up Coal-fired Oil-fired
1 (D1–D2) 412,000 6000 1,000,000 4,000,000 1,800,000 511,800 452,409 98,000 9880 110,000
135,600 2960 820,000 3,380,000 1,400,000 96,190 1856 32,160 7600 9350
2 (D2–D3) 135,600 2960 820,000 3,380,000 1,400,000 96,190 1856 32,160 7600 9350
96,000 2326 680,000 2,760,000 650,000 76,140 1067 12,036 4984 3410
3 (D3–D4) 96,000 2326 680,000 2,760,000 650,000 76,140 1067 12,036 4984 3410
89,200 1952 620,000 2,140,000 528,600 60,980 1045 7170 3291 1885
4 (D4–D5) 89,200 1952 620,000 2,140,000 528,600 60,980 1045 7170 3291 1885
80,080 1417 600,000 1,296,000 500,000 50,448 1022 3500 2236 950
5 (D5–D6) 80,080 1417 600,000 1,296,000 500,000 50,448 1022 3500 2236 950
60,000 1200 586,667 800,000 400,000 42,900 1000 2750 1850 750
6 (D6–D7) 60,000 1200 586,667 800,000 400,000 42,900 1000 2750 1850 750
57,952 998 573,333 680,000 250,000 32,980 576 2264 1000 450
7 (D7–D8) 57,952 998 573,333 680,000 250,000 32,980 576 2264 1000 450
45,500 800 560,000 420,000 200,000 24,000 500 1540 977 380
8 (D8–D9) 45,500 800 560,000 420,000 200,000 24,000 500 1540 977 380
25,020 462 440,000 385,000 150,000 21,000 440 732 954 310
9 (D9–D10) 25,020 462 440,000 385,000 150,000 21,000 440 732 954 310
19,670 332 240,000 130,000 137,500 19,180 300 624 745 240
10 (D10–0) 19,670 332 240,000 130,000 137,500 19,180 300 624 745 240
0 0 0 0 0 0 0 0 0 0
1 1 1 3 1 1 1 1 2 2 8
1 2 1 2 1 1 1 1 5 3 8
1 3 1 1 3 2 1 1 3
4 8
1 3 1 3 2 1 1 2 5
Class
1 3 3 1 1 2 1 1 1 5 9 1
1 3 3 2 4 1 1 1 5 2
6 9
1 3 4 2 1 1 2 2 5
3
… … 4
1 3 4 2 2 1 1 1 5
5
1 3 4 3 2 4 1 3 2 2398 8 6
1 3 4 3 4 4 1 3 1 7
2399 8 8
2 1 2 2 1 2 2 1 3
9
2400 5
2 2 1 1 2 2 1 1 4 10
Fig. 8. Structure of the deduced ANN algorithm for the site location potential of the urban energy supply plants.
(SVR) method [90]. However, this study does not address the opti- energy generation. For this reason, in practice, the final site selec-
mized solutions of non-linear relations by hybrid AI systems. The tion is accomplished by reflecting the decision makers’ judgments
methodology that is approached in this study is not finding the on landscape violations, land price affordability, and target policies
most optimal site itself for locating energy plants but instead sup- as well as the technical potential for energy supply plant locations.
porting optimal planning by providing the technical potentials for Hence, future studies should conduct optimizations for global
classes of the urban energy supply plants‘ availability and natural modeling by comparing the proposed ANN and other non-linear
112 I.-A. Yeo, J.-J. Yee / Applied Energy 119 (2014) 99–117
Table 9
Trained ANN structure for the site location potential of urban energy supply plants.
Energy plant type Training function Network structure (Hidden layer Output layer)
Hydraulic power General trainlm 31
Small trainscg 31
Pumping-up trainlm 21
Thermal power Coal-fired trainlm 21
Oil-fired trainscg 31
Co-generation – trainlm 31
Solar power – trainscg 31
Wind power – trainscg 41
Waste heat – trainlm 41
Mixed combustion – trainscg 21
Table 10
Trained ANN structure for the site location potential of renewable energy availability.
Classification Layer
Inputa Hidden Targeta
0 0
Solar power Latitude (° ), longitude (° ), altitude (m), month (–), average sunshine duration (hrs), and average temp (°C) Single Solar radiation (MJ)
Wind power Latitude (° 0 ), longitude (° 0 ), altitude (m), month (–), monthly average wind speed (m/s) Single Wind speed (m/s)
a
Altitude, temperature, and wind speed were displayed with the measured value of the Automatic Weather Station up to one decimal place.
Building
of
existing
Gyunggi-do, city
Republic of Central commerce
Korea 8km district
Low density
apartment district
12km
Low density
apartment district
Public housing
district of Gwang-
myung· Si-heung
(a) Geographical location (b) Case study area (c) Subject area development plan
Fig. 9. Outline of the urban development plan for the research area of interest.
Fig. 10. The site location potential map for the urban energy supply plants of the research area of interest.
oil-fired thermal power, and solar power. The research area of the research area had a higher wind speed than the other parts.
interest is suitable for hydraulic power generation because of the An area is discerned to be appropriate for wind power generation
high water level caused by the surrounding mountainous environ- if the yearly mean wind speed is over 5.0 m/s, the wind flow in
ment and is appropriate for solar power generation mainly in the the main direction occurs frequently, and there is little variation
surrounding mountain area. Wind power generation was consid- in the wind direction for the use of wind power resources. The rec-
ered to be inappropriate for most of the region because the annual ommended candidate location should have a yearly mean wind
mean wind speed is approximately 1 m/s. speed of over 4 m/s for a large-scale wind farm and over 2.0–
The solar and wind resource potential map created by applying 2.5 m/s for building units.
the ANN algorithm in Section 3 with the input dataset of the AWS The capacity of domestic wind turbines varies from 0.4 kW to
weather information on the research area is shown in Fig. 11. The 750 kW. Tables 12 and 13 show the specifications for horizontal-
wind speed was 1.7 m/s in the winter and was the minimum value axis wind turbines and vertical-axis wind turbines, respectively,
for the year. The mean wind speed was 2.0 m/s from Mar. to Oct. for present domestic technologies. According to the wind speed
and 3.4 m/s in Aug. in the area of interest. The western part of in Fig. 11, the 750-kW rated power of a horizontal-axis wind
114 I.-A. Yeo, J.-J. Yee / Applied Energy 119 (2014) 99–117
Table 12
Technical specifications for domestic horizontal-axis wind turbines.
Table 13
Technical specifications for domestic vertical-axis wind turbines.
Table 14
Technical potential of wind electric power generation for the research area of interest.
Wind turbine type Blade/rotor diameter (m) Wind speed (m/s) Monthly generation potentiala (kW/monthmesh)
HAWT 750 kW 47.0 Min.: 1.7 (Jan.) 29.31
VAWT b 3 kW 1.5 3.0 Avg.: 2.0 (May) 0.07
10 kW 3.0 6.0 Max.: 3.3 (Aug.) 0.05
50 kW 7.3 15.9 2.23
a
Applied by the max. wind speed.
b
Wind power of VAWT is considered to have 50% of the efficiency of HAWT.
Table 15
Technical potential of PV electric power generation for the research area of interest.
PV technology Efficiency g The potentially utilizable surface area CA Mean monthly solar radiation SRm Monthly Generation potentiala GPm
(%) (m2/mesh) (MJ/m2/month) (MJ/monthmesh)
Bulk Mono 14–20 Min.: 400 4.490–6.490 285,856–413,186
crystalline
Poly 13–15 Avg.: 3,745 235,411–340,271
crystalline
a-Si 6–9 Max.: 37,875 126,113–182,288
Thin film CdTe 9–11 168,151–243,051
CIS/CIGS 10–12 184,966–267,356
a
Applied by the average value of each g and CA.
turbine is acceptable in an area in which there is a maximum wind GPm ¼ SRm CA AF g ð6Þ
speed, such as the western part of the research area, and a vertical-
where GPm is the Electric power generation potential per month
axis wind turbine is suitable in areas that have a mean wind speed
(MJ/monthmesh); SRm the Monthly solar radiation received per
of 2 m/s. Table 14 shows the results of the wind electric power
unit horizontal area (MJ/m2/month); CA the Calculated total poten-
generation based on the wind resources map in Fig. 11 and the
tial area of installment (m2/mesh); AF the The area factor that indi-
domestic wind turbine technological level applied by Eq. (2) and
cates what fraction of the calculated areas can be covered by solar
is based on existing research that calculated the power output
panels (70% in the case study area) and g the efficiency with which
from wind turbines [91]. The result indicates that technical wind
the solar system converts sunlight into electricity (%).
power generation by horizontal-axis wind turbines is unattainable
In terms of the technical energy potential, the availability of a
even in areas that have the maximum wind speed, and a wind
site location for an urban energy supply plant and wind/solar en-
turbine that has at least a 10-kW rated vertical-axis should be
ergy was examined in an environmentally friendly city using the
installed in the research area of interest.
results of this study. A site location potential model for urban en-
ergy supply plants and renewable energy availability was exam-
Prated
P ¼ Prated CF ¼ Prated 0:087V 2 ð5Þ ined separately, but the integrated optimal site location should
D
be found by overlay analysis in the future. Therefore, future studies
where P is the Power output from a single turbine (kW); Prated the should specify the methodology for finding the optimal site loca-
Rated power of the turbine (kW); CF the Capacity factor; V the tion potential classes when integrated with the technical potential
Monthly averaged wind speed (m/s) and D the Turbine diameter for renewable energy.
(m).
To make solar resources such as solar power generation or 5. Conclusions
building integrated photovoltaic (BIPV) profitable, they must be in-
stalled in an area that has a high intensity of daylight and long day- This study suggested a site location potential model for urban
light hours. The recommended size of a PV panel is at least 30 m2 energy supply plants and renewable energy availability based on
(6 kW rated level) for an environment in which the minimum wind the increasing necessity of urban planning that creates an ‘energy
speed is over approximately 0.5 m/s and the humidity is below saving environmentally friendly city’ in urban energy supply plan-
80%. ning in terms of ‘sustainable development’. The applicability of this
According to Fig. 11, the monthly solar radiation of the research model was examined by applying it for a planned city in the
area varied from a minimum of 4.490 MJ/m2 to a maximum of Republic of Korea, where urban planning is in progress. The results
6.490 MJ/m2, and the building footprint area in the mesh ranges from this study are described as follows.
from a minimum of 400 to a maximum of 37,875 m2, with a mean
value of 3745 m2 as a result of examining the E-GIS DB of the ap- (1) Section 2 examined existing research trends in terms of the
plied area to consider the installable surface area of the building technical energy potential and site location planning
roof. Table 15 shows the result of the monthly mean PV electric support systems and showed the need to reinforce the
power generation in a mesh that is based on the present domestic energy supply plan in environmentally friendly urban
PV technology, as applied by Eq. (3), and that is based on existing energy planning based on an appropriate combination of
research that calculated the yearly electric power output from PV energy systems. An integrated system is insufficient for the
[8], which is shown in Table 15. As a result of applying the mean optimal site location of urban energy supply systems.
value of the efficiency g (%) of each PV technology and the poten- (2) Section 3 integrated the E-GIS DB, which contains the
tially utilizable building roof area (m2/mesh), the maximum elec- geographic and climatic conditions of the area and energy-
tric power generation potential was 413,186 MJ/monthmesh related information to manage the integrated factors that
with mono crystalline bulk PV. affect the site location of the energy supply plants. For the
116 I.-A. Yeo, J.-J. Yee / Applied Energy 119 (2014) 99–117
[32] Masmoudi S, Mazouz S. Relation of geometry, vegetation and thermal comfort [62] Khatib T, Mohamed A, Sopian K. A review of solar energy modeling techniques.
around buildings in urban settings, the case of hot arid regions, Energy and Renew Sustain Energy Rev 2012;16(5):2864–9.
Buildings. Energy Build 2004;36(7):710–9. [63] Ozgoren M, Bilgili M, Sahin B. Estimation of global solar radiation using ANN
[33] Marciotto ER, Oliveira AP, Hanna SR. Modeling study of the aspect ratio over Turkey. Expert Syst Appl 2012;39(5):5043–51.
influence on urban canopy energy fluxes with a modified wall-canyon energy [64] Yadav AK, Chandel SS. Solar radiation prediction using Artificial Neural
budget scheme. Build Environ 2010;45(11):2497–505. Network techniques: a review. Renew Sustain Energy Rev; in Press. Corrected
[34] Ashie Y, Ca VT, Asaeda T. Building canopy model for the analysis of urban Proof.
climate. J Wind Eng Ind Aerodyn 1999;81(1–3):237–48. [65] S
ß ahin M, Kaya Y, Uyar M. Comparison of ANN and MLR models for estimating
[35] Mahjoobi J, Mosabbeb EA. Prediction of significant wave height using solar radiation in Turkey using NOAA/AVHRR data. Adv Space Res
regressive support vector machines. Ocean Eng 2009;36(5):339–47. 2013;51(5):891–904.
[36] Mavromatidis G, Acha S, Shah N. Diagnostic tools of energy performance for [66] Kalogirou SA, Bojic M. Artificial neural networks for the prediction of the
supermarkets using artificial neural network algorithms. Energy Build energy consumption of a passive solar building. Energy 2000;25(5):479–91.
2013;62(7):304–14. [67] Mohandes MA. Modeling global solar radiation using particle swarm
[37] Islam SM, Al-Alawi SM, Ellithy KA. Forecasting monthly electric load and optimization (PSO). Sol Energy 2012;86(11):3137–45.
energy for a fast growing utility using an artificial neural network. Electr [68] Kalogirou SA, Panteliou S. Thermosiphon solar domestic water heating
Power Syst Res 1995;34(1):1–9. systems: long-term performance prediction using artificial neural networks.
[38] Sözen A, Arcaklioğlu E, Özkaymak M. Turkey‘s net energy consumption. Appl Sol Energy 2000;69(2):163–74.
Energy 2005;81(2):209–21. [69] Mellit A, Kalogirou SA, Hontoria L, Shaari S. Artificial intelligence techniques
[39] Hamzaçebi C. Forecasting of Turkey’s net electricity energy consumption on for sizing photovoltaic systems: a review. Renew Sustain Energy Rev
sectoral bases. Energy Policy 2007;35(3):2009–16. 2009;13(2):406–19.
[40] Sözen A, Gülseven Z, Arcaklioğlu E. Forecasting based on sectoral energy [70] González PA, Zamarreño JM. Prediction of hourly energy consumption in
consumption of GHGs in Turkey and mitigation policies. Energy Policy buildings based on a feedback artificial neural network. Energy Build
2007;35(12):6491–505. 2005;37(6):595–601.
[41] Midilli KE, Dincer I, Rosen MA. Artificial neural network analysis of world [71] Yao Y, Lian Z, Hou Z, Liu W. An innovative air-conditioning load forecasting
green energy use. Energy Policy 2007;35(3):1731–43. model based on RBF neural network and combined residual error correction.
[42] Sözen A. Future projection of the energy dependency of Turkey using artificial Int J Refrig 2006;29(4):528–38.
neural network. Energy Policy 2009;37(11):4827–33. [72] Neto AH, Fiorelli FAS. Comparison between detailed model simulation and
[43] Ekonomou L. Greek long-term energy consumption prediction using artificial artificial neural network for forecasting building energy consumption. Energy
neural networks. Energy 2010;35(2):512–7. Build 2008;40(12):2169–76.
[44] Safa M, Samarasinghe S. Determination and modelling of energy consumption [73] Dombaycı OA. The prediction of heating energy consumption in a model house
in wheat production using neural networks: ‘‘a case study in Canterbury by using artificial neural networks in Denizli–Turkey. Adv Eng Softw
province, New Zealand’’. Energy 2011;36(8):5140–7. 2010;41(2):141–7.
[45] Kankal M, Akpınar A, Kömürcü MI, Özsßahin TSß. Modeling and forecasting of [74] Azadeh A, Ghaderi SF, Tarverdian S, Saberi M. Integration of artificial neural
Turkey‘s energy consumption using socio-economic and demographic networks and genetic algorithm to predict electrical energy consumption.
variables. Appl Energy 2011;88(5):1927–39. Appl Math Comput 2007;186(2):1731–41.
[46] Bilgili M, Sahin B, Yasar A, Simsek E. Electric energy demands of Turkey in [75] Pao HT. Forecasting electricity market pricing using artificial neural networks.
residential and industrial sectors. Renew Sustain Energy Rev Energy Conv Manag 2007;48(3):907–12.
2012;16(1):404–14. [76] Ermis K, Erek A, Dincer I. Heat transfer analysis of phase change process in a
[47] Chang CS, Wang F, Liew AC. Simulated hardware design of artificial neural finned-tube thermal energy storage system using artificial neural network. Int
networks for adaptive plant control. Electr Power Syst Res 1996;37(3):231–40. J Heat Mass Transf 2007;50(15–16):3163–75.
[48] Tso SK, Gu XP, Zeng QY, Lo KL. Deriving a transient stability index by neural [77] Azadeh A, Ghaderi SF, Sohrabkhani S. Annual electricity consumption
networks for power-system security assessment. Eng Appl Artif Intell forecasting by neural network in high energy consuming industrial sectors.
1998;11(6):771–9. Energy Conv Manage 2008;49(8):2272–8.
[49] Karapidakis ES. Machine learning for frequency estimation of power systems. [78] Hong WC. Electric load forecasting by support vector model. Appl Math Model
Appl Soft Comput 2007;7(1):105–14. 2009;33(5):2444–54.
[50] Bechrakis DA, Deane JP, McKeogh EJ. Wind resource assessment of an area [79] Olanrewaju OA, Jimoh AA, Kholopane PA. Integrated IDA–ANN–DEA for
using short term data correlated to a long term data set. Sol Energy assessment and optimization of energy consumption in industrial sectors.
2004;76(6):725–32. Energy 2012;46(1):629–35.
[51] Cadenas E, Rivera W. Short term wind speed forecasting in La Venta, Oaxaca, [80] Azadeh A, Babazadeh R, Asadzadeh SM. Optimum estimation and forecasting
México, using artificial neural networks. Renew Energy 2009;34(1):274–8. of renewable energy consumption by artificial neural networks. Renew Sustain
[52] Fadare DA. The application of artificial neural networks to mapping of wind Energy Rev 2013;27:605–12.
speed profile for energy application in Nigeria. Appl Energy [81] Hong WC. Hybrid evolutionary algorithms in a SVR-based electric load
2010;87(3):934–42. forecasting model. Int J Electr Power Energy Syst 2009;31(7–8):409–17.
[53] Cadenas E, Jaramillo OA, Rivera W. Analysis and forecasting of wind velocity in [82] Magnier L, Haghighat F. Multiobjective optimization of building design using
chetumal, quintana roo, using the single exponential smoothing method. TRNSYS simulations, genetic algorithm, and artificial neural network. Build
Renew Energy 2010;35(5):925–30. Environ 2010;45(3):739–46.
[54] Velázquez S, Cart JA, Matías JM. Comparison between ANNs and linear MCP [83] Lee KH, Jun SO, Pak KH, Lee DH, Lee KW, Park JP. Numerical optimization of site
algorithms in the long-term estimation of the cost per kW h produced by a selection for offshore wind turbine installation using genetic algorithm. Curr
wind turbine at a candidate site: a case study in the Canary Islands. Appl Appl Phys 2010;10(2):S302–6.
Energy 2011;88(11):3869–81. [84] Iliadis L, Maris F, Tachos S. Soft computing techniques toward modeling the
[55] Sheela G, Deepa SN. An intelligent computing model for wind speed prediction water supplies of Cyprus. Neural Networks 2011;24(8):836–41.
in renewable energy systems. Procedia Eng 2012;30:380–5. [85] Pedro HTC, Coimbra CFM. Assessment of forecasting techniques for solar
[56] Jung S, Kwon SD. Weighted error functions in artificial neural networks for power production with no exogenous inputs. Sol Energy 2012;86(7):2017–28.
improved wind energy potential estimation. Appl Energy 2013;111:778–90. [86] Abdelsalam AA, Eldesouky AA, Sallam AA. Classification of power system
[57] Mabel MC, Fernandez E. Analysis of wind power generation and prediction disturbances using linear Kalman filter and fuzzy-expert system. Int J Electr
using ANN: a case study. Renew Energy 2008;33(5):986–92. Power Energy Syst 2012;43(1):688–95.
[58] Benghanem M, Mellit A, Alamri SN. ANN-based modelling and estimation of [87] Wei Z, Li X, Xu L, Cheng Y. Comparative study of computational intelligence
daily global solar radiation data: a case study. Energy Conv Manage approaches for NOx reduction of coal-fired boiler. Energy 2013;55:683–92.
2009;50(7):1644–55. [88] Kavaklioglu K. Robust electricity consumption modeling of Turkey using
[59] Fadare DA. Modelling of solar energy potential in Nigeria using an artificial singular value decomposition. Int J Electr Power Energy Syst 2014;54:268–76.
neural network model. Appl Energy 2009;86(9):1410–22. [89] Beale MH, Hagan MT, Demuth HB. Neural Network Toolbox™ Getting Started
[60] S
ß enkal O. Modeling of solar radiation using remote sensing and artificial neural Guide R2013b. The MathWorks Inc.; 2013.
network in Turkey. Energy 2010;35(12):4795–801. [90] Kavaklioglu K. Modeling and prediction of Turkey‘s electricity consumption
[61] Hossain R, Ooa AMT, Shawkat Alia ABM. Historical weather data supported using support vector regression. Appl Energy 2011;88(1):368–75.
hybrid renewable energy forecasting using artificial neural network (ANN). [91] Yue CD, Yang MH. Exploring the potential of wind energy for a coastal state.
Energy Procedia 2012;14:1035–40. Energy Policy 2009;37(10):3925–40.