SHORT COMMUNICATION
Identi
fication of optimal strategies for sustainable
energy management in Taiwan
S. K. Ning
1, S. C. Yeh
2,*
,†, J. C. Chen
3, M. C. Hung
1, Q. G. Lin
4, Y. P. Cai
4and Y. H. Yeh
2 1Department of Civil and Environmental Engineering, National University of Kaohsiung, Kaohsiung, Taiwan2
Graduate Institute of Environmental Education, National Taiwan Normal University, Taipei, Taiwan
3Department of Environmental Engineering and Science, Fooyin University, Kaohsiung, Taiwan 4Center for studies in Energy and Environment, University of Regina, Regina, SK, Canada
ABSTRACT
In this study, an optimization model was developed for identifying optimal strategies in adjusting the existing fossil fuel-based energy structure in Taiwan. In this model, minimization of the total system cost was adopted as the objective func-tion, which was subject to a series of constraints related to energy demand, greenhouse gas (GHG) emission restricfunc-tion, and energy balance. Feasibility of several potential energy structures was also evaluated through tradeoff analysis between en-ergy system costs and GHG emission targets. Three scenarios were established under several GHG emission restriction tar-gets and potential nuclear power expansion options. Under the three scenarios, optimal energy allocation patterns were generated. In terms of the total energy system cost, the scenario that restricted GHG emissions and nuclear power growth would result in the highest one, with an average annual increase of 4.2% over the planning horizon. Also, the results indi-cated that the energy supply structure would be directly influenced by energy cost and GHG emission reduction targets. Scenario 2 would lead to the greatest dependence on clean energy, which would take up 41.8% in 2025. In comparison, with no restriction on nuclear energy, it would replace several energy sources and contribute to 34.0% of the total energy consumption. Significant reduction in GHG emission could be identified under scenario 2 due to the replacement of con-ventional fossil fuels with clean energies. Under scenario 3, GHG emission would be significantly reduced due to the adop-tion of nuclear power. After 2015, energy structure in Taiwan would be slightly adjusted due to synthetic impacts of energy demand growth and GHG emission restriction. The results also indicated that further studies would be necessarily needed for evaluating impacts and feasibilities of clean energy and nuclear power utilization in Taiwan. Copyright © 2011 John Wiley & Sons, Ltd.
KEY WORDS
energy structure; energy model; GHG reduction; optimization; Taiwan Correspondence
*Dr. Shin-Cheng Yeh, Graduate Institute of Environmental Education, National Taiwan Normal University; 88, Sec. 4, Tingjou Rd., Taipei, Taiwan 11677.
†E-mail: scyeh@ntnu.edu.tw
Received 9 January 2011; Revised 19 June 2011; Accepted 9 July 2011
1. INTRODUCTION
Over the past decades, global energy demand has been un-dergoing a steady growth. Particularly, consumptions of fossil fuel have increased significantly in many economi-cally prosperous regions such as Taiwan. In order to sup-ply energy resources in a sustainable and economical manner, optimal management of energy systems is desired. However, there are a number of factors and issues that need to be considered in the managing process, posing many challenges for decision makers. For example, energy price is subject to many political, economic, and environ-mental conditions, and is hard to be quantified in the man-agement practices [1]. Moreover, since greenhouse gas
(GHG) emission and climate change have become a major concern across the world, adjustment of the current fossil fuel-based energy structure to a low-carbon ones is essential [2]. Thus, how to satisfy the needs of energy uses at the least cost while ensuring sustainable energy supply has been the most crucial objective of energy planning for decision makers at multiple scales [3,4]. This is particularly impera-tive in Taiwan due to its high dependency on imported fossil fuel and its necessity to foster a green energy system that is economical, low-carbon emission, and high ef fi-ciency. Therefore, development of effective tools for sup-porting management of energy resources, technologies, and services is desired, which could provide an synthetic mechanism to identify optimal strategies for supporting
new energy utilization, regional energy allocation, GHG emission reduction, and sustainable development.
Previously, the issues of optimal energy combinations were studied with respect to several potential energy resources through examining their supply and consump-tion costs and efficiencies. Also, the advantages and short-comings of utilizing renewable energies and fossil fuels were analyzed. For example, Jenkins (1997) determined the optimal scale of biofuel energy applications and the as-sociatedfixed and variable costs [5]. Rana et al. (1998) proposed a case study in which renewable energy sources were incorporated into existing power generation systems [6]. In their research, biogas, biomass, and solar energy were used as the alternative sources of electricity in Madhy Pradesh Valley in India. It was shown that this decentra-lized mixed system had better economic efficiency than the traditional centralized power supply system. Dornburg and Faaij (2001) employed systems analysis techniques to examine the relationships among characteristics of bio-masses, logistics, ranges of heat supply, combustion and gasification unit designs, and appliance scales [7]. The ef-ficiency issues in terms of energy use and economics were investigated. Yoshida et al. (2003) compared the efficiency and CO2 emissions of a variety of biomass conversion techniques using LCA-based techniques [8]. Tradeoff be-tween CO2emissions and total cost of technologies was analyzed for determining the most cost-effective renewable energy techniques with consideration of GHG emissions. Sørensen (2010) analyzed possible optimal energy strate-gies to meet energy demands in a specific region with the consideration of diverse energy sources [9]. At the same time, optimal analysis was the major focus in many re-search works. The factors being taken into account in-cluded technical/operational characteristics, terminal energy demand, and environmental quality requirements. The minimized cost and the maximized performance were among the most concerned objectives in these studies. For example, Iniyan et al. (2000) identified optimal allocation patterns of solar, wind, and biomass energy using a socio-economic factors-based optimization model with the objective for fulfilling the energy demands in different end-use sectors [10]. Iniyan and Sumathy (2000) examined the future direction of biomass energy development using linear programming techniques [11]. An optimal renew-able energy mathematical model was developed for identi-fying the best energy policy with cost minimization of renewable energy at afixed power output as the objective and social acceptance, system reliability, energy demands, and renewable energy development potentials as the con-straints. The follow-up of this research was conducted and the Delphi survey techniques were used for incorporat-ing opinions of many experts into the assessment and eval-uation of relevant policies [12]. Suganthi and Williams (2000) used an optimization model to determine the alloca-tion strategy of commercialized renewable energy systems in India, in which sensitivity analysis was performed in terms of energy demand, technology development, system reliability, and manpower [13]. Schmidt et al. (2010)
assessed the combined heat and power potentials through using a mixed integer programming model for supporting optimal siting of bioenergy plants [14]. Drozdz (2003) ad-vanced suggestions to the modernization of Poland’s energy system, including proper adoption of energy appliances and selection of fuels [15]. Mathur et al. (2004) evaluated the growth trend of electricity generated from renewable energy sources in the future for India by considering the dynamic input–output balance of energy during the period of construction of the new power systems [16]. The results showed that the small-sized wind power systems would be the one growing fastest in the near fu-ture. Jebaraja and Iniyanb (2006) reviewed the models of renewable and sustainable energy planning, including en-ergy planning models, enen-ergy supply and demand models, prediction models, renewable energy models, emission re-duction models, and optimization models [17]. Iniyan et al. (2006) developed a modified econometric mathematical (MEM) model for predicting the demand of coal, oil, and electricity, then derived the optimal allocation of commercial energy using the mathematical programming energy-economy-environment model with environmental constraints [18]. Becerra-López and Golding (2008) uti-lized multi-objective programming techniques tofind the optimal conversion capacity of renewable energy and nat-ural gas. The total cost and environmental impacts were the major referential factors in the regional energy system development planning [19]. Cai et al. (2010) developed an energy model integrating systems analysis and environ-mental economics theories. Energy supply and demand, energy cost, air pollution, and GHG emissions were the major variables in the scenarios designed for decision makers and users of energy. This model was successfully applied to the evaluation and promotion of energy policies of Waterloo Metropolitan Area in Ontario, Canada. These studies show the importance of energy models for national and regional planning [20]. Moreover, in many studies, en-vironmental and social constraints were considered for identifying optimal energy allocations. In addition to com-mon air pollutants such as SOxand NOx, CO2emissions are becoming a major concern as global warming and cli-mate change have attracted more and more attention. To abide by the GHG reduction targets, many countries have revised their energy policy or modified their energy struc-tures. In the integrated studies of energy use and CO2 emis-sions, optimization models were frequently employed for performance evaluation. For example, Ghafghazi et al. (2010) compared energy costs and GHG emission profiles of a district heating project under various energy demand scenarios. It was shown that the natural gas boiler option would provide energy with less costs and higher GHG emissions [21]. Brownsword et al. (2005) established an energy supply and demand model for simulating the varia-tion of energy demand with respect to space and time, and assessing the GHG reduction performance [22]. An em-bedded optimization model was capable of identifying the most cost-effective GHG reduction scheme. Pehnt (2006) applied LCA to examine the environmental
advantage of electricity power systems and heating sys-tems using traditional fossil fuel, hydropower, solar en-ergy, wind, geothermal enen-ergy, and wood as fuel sources [23]. The results indicated that renewable energy would lead to better environmental performances. Gilau et al. (2007) compared the economic advantages of different mixture of energy sources including diesel, solar power-diesel, wind-power-diesel, and solar power-wind-diesel and found that the developmental cost for renewable energy could be as low as zero if the negative externality of CO2emissions were reduced with a 87% cut of the total CO2emissions [24]. A renewable energy-based system would be more cost effective. Granovskii et al. (2007) discussed the differ-ences in air pollution produced by the use of renewable en-ergy sources instead of natural gas for electricity and hydrogen production [25]. Renewable energy sources per-formed better in air pollution reduction. Blesl et al. (2007) assessed the achievements of GHG reduction via improv-ing energy efficiency for the individual sectors and the na-tion as a whole in Germany [26]. Efficiency promotion in the residential sector would contribute to significant GHG reduction while that in the transportation sector would cost the most. Lafforgue et al. (2008) analyzed the shares of fossil fuel, renewable energy, and clean energy with the viewpoint of carbon sinks [27]. Thornely et al. (2009) de-veloped a set of 42 indicators in technical, economical, en-vironmental, and social aspects for evaluating the performance of multiple biofuels [28]. It was found that the general co-combustion system can emit less CO2than traditional power generation systems, while large gasifier systems can emit the least CO2. Daniel et al. (2009) built up an optimization model for minimizing cost following the concept of energyflow with multiple energy alterna-tives and CO2emission constraints [29]. Dougherty et al. (2009) evaluated the cost, energy production, and GHG re-duction of shifting a traditional power system to a hydro-gen electricity power system with different hydrohydro-gen generation techniques and transportation systems [30]. Chau et al. (2009) carried out a technique-economy analy-sis of an energy saving house using a wooden pellets com-bustion heating system [31]. It was found that this system can reduce emission for as much as 3,000 tonnes of CO2 -equivalent. Soimakallio et al. (2009) examined the GHG reduction of the transportation sector when replacing fossil fuels with different kinds of biofuels [32].
More recently, because the high uncertainty inherent in the predictions in the energy models brings challenges to the accuracy of the results, many uncertainty analysis tools such as statistics and fuzzy mathematics have been employed in related studies. Lin and Huang (2009) inte-grated mixed integer programming, interval parametric programming, and two-stage stochastic programming and developed a dynamic inexact stochastic energy systems planning model and applied it to energy management of Beijing, China [33]. This model can reflect the uncertainty in GHG reduction corresponding to different management and power capacity expansion scenarios for decision makers’ convenient reference. Li et al. (2010) developed
a multistage interval-stochastic regional-scale energy model for analyzing many energy policies that can help conduct the dynamic power system planning, in which probability distribution and interval values were used to deal with the uncertainty in the system planning [34]. Cai et al. (2009) integrated internal linear programming and chance-constrained programming for planning a renewable energy system [35]. A variety of decision alternatives were carried out under different settings of economic policy and system reliability. Cai et al. (2009) then developed a fuzzy-random interval programming capable of reflecting differ-ent patterns of uncertainty in the objective function as well as constraints. Formulating the upper and lower bounds of the interval parameters as fuzzy-random vari-ables can make the results of optimization models more robust [36]. Cai et al. (2010) further developed an interval-parameter superiority–inferiority-based regional energy management (REM) model for supporting regional energy management systems planning under uncertainty. It can be used for generating decision alternatives and thus help resource managers identify desired policies under var-ious economic and system-reliability constraints [37]. Lin and Huang (2009) examined the interrelationships between several community level energy management systems through integrating interval parameters and mixed integer programming [38]. As the dynamic characteristics of the environmental, social, and economic problems in the com-munity can be formulated, the model can support energy systems analysis and environmental management. Lin et al. (2009) built up an interval-fuzzy two-stage stochastic energy systems planning model, in which fuzzy numbers, probability distributions, and discrete time intervals were employed for reflecting the complex uncertainties of the systems and dealing with the inherent two-stage stochastic decision-making issues in the energy systems [39].
Taiwan is located in East Asia, with a total area of 36.0 thousand square kilometers. It is a crowded island with a population of 23.1 million people. In the past two decades, the total supply of primary energy source had increased steadily from 58,520.0 MLOE (Mega Liter Oil Equivalent) in 1990 to 138,050 MLOE in 2009. The average growth rate in this period was as high as 4.6%. In 2009, coal, oil, and natural gas occupied 30.5, 51.8, and 8.4% of the total energy supply, respectively. The remaining percentage was composed of 0.3% of hydropower, 0.1% of solar and wind, and 8.7% of nuclear power [40]. It was indicated that Tai-wan relied greatly on oil and coal heavily. Moreover, only 0.6% of the total energy resources were domestically gen-erated and over 99.7% were imported externally. Thus, Taiwan’s economic growth could be easily influenced by thefluctuating energy prices in the international markets. According to International Energy Agency [41], GHG emission in Taiwan has reached the amount of 276.2 mil-lion tonnes CO2-eq, which was the 22
nd
place in the world. In 2000, the government in Taiwan announced a ‘no-nuclear community’ policy, and this meant nuclear power projects could not be expanded in the near future. After 2008, ‘no-nuclear community’ policy has still been
adopted in the near future. How to diversify energy supply sources and meet GHG reduction goals at the same time has been the major challenge of Taiwan’s energy policy. Thus, in order to improve energy efficiency and energy diversity with the consideration of environmental protection and en-ergy security, Taiwan has initiated a number of programs for energy sustainable development, covering renewable energy utilization, energy conservation, high energy con-sumption efficiency, high productive value, low emission, and low dependency [42]. In this study, it is anticipated that the potential energy structures can be identified through us-ing systems analysis techniques while takus-ing into account multiple objectives and constrains such as energy demands, application costs, and environmental protection. The previous studies reflected that a well-developed en-ergy model could help effectively evaluate various enen-ergy strategies for regional development. Also, the replacement of fossil fuel with renewable energy could be an effective way for GHG reduction. However, since the costs of cleaner energy alternatives are comparatively high, the overall cost effectiveness is still necessary with the consid-eration of GHG emission benefits. Thus, tradeoff analysis between economic and environmental targets needs to be considered in energy system planning. The complex inter-relationships and competition among various renewable energy sources and technologies need to be analyzed using an energy model incorporating energy technology, cost, and environmental goals so that multiple and comprehen-sive investigations can be made. A large-scale optimization model will be developed in this study, which is thefirst one in Taiwan. The results could provide decision support to decision makers in adjusting energy structures to achieve lower energy costs, lower GHG emissions, and higher en-ergy reliability.
2. MODELING FORMULATION
In this study, energy supply in Taiwan can be divided into two categories: fuel and electricity. With further consider-ation of renewability and carbon intensity, the energy sup-ply can be also categorized into fossil fuel energy (coal, oil, and LNG), clean energy (wind power, geothermal, hydro power, solar power, ocean energy, and bio-energy), and nuclear energy. On the other hand, end-users include in-dustrial, transportation, agricultural, municipal, and com-mercial sectors. The detailed energy supply–demand framework is illustrated in Figure 1.
A large-scale optimization model was formulated. Min-imization of the total energy supply cost is the objective of the model. The demands of fuels and electricity of each of the sectors and the targets of GHG reduction are formu-lated as the constraints. The objective of this optimization model is minimization of the total energy cost which is the summation of related costs from fossil fuel, clean en-ergy, and nuclear energy. These costs include any funds spent for purchasing the raw materials and the following treatments and transformations, which can be presented as follows:
MinZ¼ TFC þ TRC þ TOC (1) where TFC is the total cost of fossil fuel, TRC is the total cost of clean energy, and TOC is the total cost of nuclear energy. Detailed energy sources considered in the model are presented in Table I.
Each part of the costs is detailed as follows:
TFC¼X 17 i¼1 FUit FUCit ð Þ þX17 i¼1 X3 w¼1
FEUwit FEUCwit
ð Þ; 8t
(2) where FUitis the total amount of energy use of the ith fos-sil fuel in the tth year (PJ); FUCitis the unit cost of the ith fossil fuel in the tth year (NT$/PJ); FEUwitis the electricity generated by the ith fossil fuel in the tth year with the wth loading status (Mkwh); FEUCwitis the unit cost of power generation by the ith fossil fuel in the tth year with the wth loading status (NT$/Mkwh). TRC¼X 9 j¼1 RUjt RUCjt þX9 j¼1 X3 w¼1 REUwjt REUCwjt ; 8t (3) where RUjt is the total amount of energy use of the jth clean energy source in the tth year (PJ); RUCjtis the unit cost of the jth clean energy source in the tth year (NT$/ PJ); REUwjtis the electricity generated by the jth clean en-ergy source in the tth year with the wth loading status (Mkwh); REUCwjtis the unit cost of power generation by the jth clean energy source in the tth year with the wth loading status (NT$/Mkwh); TOC¼X 1 k¼1 X3 w¼1 OEUwkt OEUCwkt ð Þ; 8t (4)
where OEUwktis the electricity generated by the kth‘other’ energy source in the tth year with the wth loading status (Mkwh); OEUCwkt is the unit cost of power generation by the kth‘other’ energy source in the tth year with the wth loading status (NT$/Mkwh). The cost of fossil fuel is assumed to increase gradually owing to depletion of nature resources. On the other hand, the cost of clean energy is considered to decrease due to the development of technol-ogies. Table II presents energy costs.
The constraints are presented as follows:
1. Constraints of fuel demands, Equations (5) to (7) are employed for confirming that the fuel demands can be satisfied. The pattern of fuel supply was further classified to three groups of coal (Equation 5), oil (Equation 6), and natural gas (Equation 7) on the basis of current energy structure. The terms of left-hand side on the equations represent that all possible energy types could be used for demands of fuel type on the right-hand side:
X6 i¼1 FUitþ X9 j¼1 RUjt⩾ X d′ X d FDd′dt Cd′; 8t (5) X15 i¼7 FUitþ X9 j¼1 RUjt⩾ X d′ X d FDd′dt Od′; 8t (6) X17 i¼16 FUitþ X9 j¼1 RUjt⩾ X d′ X d FDd′dt Gd′; 8t (7)
where FDd’dtis the energy demand of the d’th subsector of the dth sector in the tth year (PJ); Cd’ is the percentage of coal in the energy demand of the d’th subsector; Od’ is the percentage of oil in the energy demand of the d’th sub-sector; Gd’ is the percentage of natural gas in the energy demand of the d’th subsector.
2. Constraints of electricity demands, Equation (8) is for confirming that the sectoral electricity demands can be satisfied. It covers fossil energy, clean energy, and nuclear energy: X17 i¼1 FEUwitþ X9 j¼1 REUwjtþ X1 k¼1 OEUwkt ⩾X d’ X d EDwd’dt; 8t; w (8)
where EDwd’dtis the wth load pattern of electricity demand of the d’th subsector of the dth sector in the sth month in the tth year (Kwh); w =1 means the basic load, applicable for fossil fuel and nuclear energy; w =2 means the medium load, applicable for fossil fuel and clean energy; w =3 means the peak load, applicable for fossil fuel.
3. Environmental constraints, Equation (9) is for con-firming that the annual GHG emission will not exceed the target in each year:
Figure 1. Energy supply–demand structure in Taiwan.
Table I. The list of energy sources in this research.
i=fossil fuel j=clean
energy
k=nuclear energy
1 Steam coal 1 Wind power 1 Nuclear
power 2 Sub-bituminous coal 2 Solar power
3 Anthracite 3 Hydro power
4 Coke 4 Tidal power
5 Coke oven gas 5 Geothermal power
6 Steam furnace gas 6 Current generation
7 LPG 7 RDF
8 Naphtha 8 Bio-diesel
9 Motor gasoline 9 Bio-ethanol
10 Jet fuel 11 Diesel oil 12 Fuel oil 13 Lubricants 14 Asphalts 15 Others 16 LNG (domestic) 17 LNG (imported)
X17 i¼1 FUit COFUi ð Þ þX17 i¼1 X3 w¼1
FEUwit COFEUi
ð Þ þX9 j¼1 RUjt CORUj þX9 j¼1 X3 w¼1 REUwjt COREUj þX1 k¼1 X3 w¼1 OEUwkt COOEUk ð Þ⩽EOt;8t (9)
where COFUiis the GHG emission factor of the ith fossil fuel (kg/MJ); COFEUwi is the GHG emission factor of the ith fossil fuel if used for electricity generation with the wth load pattern (kg/kwh); CORUjis the GHG emis-sion factor of the jth clean energy source (kg/MJ); COR-EUwjis the GHG emission factor of the jth clean energy source if used for electricity generation with the wth load
pattern (kg/kwh); COOEUwkis the GHG emission factor of the kth nuclear energy source if used for electricity gen-eration with the wth load pattern (kg/kwh); EOtis the GHG emission target of the tth year.
4. Non-negativity constraints:
FUit⩾0; FEUwit⩾0; RUjt⩾0: (10)
3. RESULT ANALYSIS AND
DISCUSSIONS
This study set up three reality-based scenarios, taking into consideration Taiwan’s increasing demand for energy, the need to reduce GHG emissions, and the fact that the issue of nuclear energy is emotionally loaded and politically po-larizing in Taiwan. The average growth rate of energy de-mand is about 3% in the recent ten years in Taiwan. Sustainable energy development policy was proposed by the Ministry of Economic Affairs (Taiwan) in 2008 to bal-ance the objectives of energy security, economic develop-ment, and environment protection, and consider the need of future generations. The targets are improving energy ef-ficiency, developing clean energy, and securing stable en-ergy supply. The strategy therefore included cleaner energy supply and rationalized energy demand to meet the GHG emission reduction goal: the total emission could return to the 2000 level in 2025. Summarizing the energy policy in the future, three planning scenarios were set for evaluating the strategy feasibility. Scenario 1 was set up based on the current situation, i.e., satisfying the growing energy demand and setting no limit to GHG emissions while maintaining current nuclear power output. Scenario 2 referred to as satisfying the growing energy demand while maintaining the current nuclear power output, but re-quiring that by 2025, GHG emissions be gradually reduced to the level in 2000. Scenario 3 then required that by 2025, GHG emissions be drawn down to the 2000 level and set no cap on the growth of nuclear power. The three scenarios are detailed in Table III. In order to fully reflect energy structure in reality and the forms energy demands take, en-ergy demands were divided into two broad categories of fuel and electricity, and the annual energy supplies were required to satisfy the demands of all sectors. Detailed en-ergy demands and the related parameters are presented in Table IV and V (including various energy costs, GHG emission amount, and average annual increasing rates of energy costs). The costs of different energy sources were estimated from local and international predictions and allowed for variations through the years. Among them the cost of fossil fuel was assumed to grow annually and that of new energy to decline annually, as forecast by rele-vant researches. In addition, based on the current practice in Taiwan, electricity demand was separated into three load conditions including low, medium, and peak demands which are set at 60, 27.5, and 12.5%, respectively. Because of their natural limits, some energy sources can only be used for medium and peak load conditions. 2009 was set
Table II. Assumptions of energy cost variation. Energy type Energy cost (fuel) Energy cost (electricity)
Steam coal1 1.75% 1.75%
Anthracite1 1.75%
Coke1 1.75%
Coke oven gas1 1.75%
Steam furnace gas1 1.75%
LPG1 2.40% Naphtha1 2.40% Motor gasoline1 2.40% Jet fuel1 2.40% Diesel oil1 2.40% 2.40% Fuel oil1 2.40% 2.40% Lubricants1 2.40% Asphalts1 2.40% Other1 2.40% LNG (domestic)1 4.85% LNG (imported) 4.85% 4.85% RDF2 0.78% Bio-diesel2 1.59% Bio-ethanol2 1.59% Sub-bituminous coal1 1.75% Wind power2 0.70% Solar power2 3.56% Hydro power2 0.0% Tidal power2 0.0% Geothermal power2 0.19% Current generation2 0.0% RDF power2 0.78% Nuclear power1 4.85% Note: 1
In year 2008–2017 on the long-term energy demand forecast and power development plan summary. Bureau of Energy, MOEA, Taiwan.
2IEA Bioenergy. Potential Contribution of Bioenergy to the
World’s Future Energy Demand. International Energy Agency 2007.
as the base year in this study. The projected period was from 2010 to 2025.
According to the model formulation and scenario set-ting, the results of (i) the cost of energy supply, (ii) the changing energy structure, (iii) the trend of clean energy, and (iv) the trend of GHG emissions derived from this model are discussed in the following:
3.1. The cost of energy supply
The results show that total energy costs will grow annually in every scenario, as Figure 2 illustrates. Scenario 2 is the costliest due to the double limits set by the compliance with GHG emission standards and the restriction on the lower cost nuclear power. With an average annual increase of 4.23%, the total energy cost here is projected to reach 2.698 trillion NTD in 2025, an 86% jump from 2010. The total energy costs in 2025 from scenario 1 and 3 would be 2.403 and 2.416 trillion, respectively, with average annual increases of 3.5 and 3.7% (Table VI). Significantly, the total energy cost in scenario 3 outstrips the cost in sce-nario 1 in 2023; this is mainly because scesce-nario 3 deploys wind power, which has zero GHG emissions but higher per unit costs, as a medium load electricity source. It would seem that nuclear power is the inevitable option if one hopes to reduce GHG emissions without a drastic increase in cost. Nevertheless, the increasing trend of the total system cost would be gradually slowed down if the costs
associated with renewable energy utilization are lower than those of the conventional ones.
3.2. The change of energy structure
Figure 3 shows the results of the energy supply from the three scenarios. The results of scenario 1 (Figure 3(a)) indi-cate that in the projected future coal, oil, liquefied natural gas, nuclear energy, and clean energy continue to supply 30.0, 40.0, 7.0, 8.0, and 15.0%, respectively, of total en-ergy needs. With the exception of clean enen-ergy, these fig-ures correspond to the current energy structure in Taiwan, which demonstrates that the model of this study closely reflects local reality. The reason why the clean en-ergy share is higher here is its use in power generation. In order to minimize total fuel costs, the model decides to re-place the current oil for power generation with the cheaper clean energy (RDF). Furthermore, in scenario 2 (Figure 3 (b)), clean energy is clearly the largest portion (41.8%) in 2025. This is because with the restriction on the growth of nuclear power, clean energy is relied upon as the source for fuel and electricity in order to lower GHG emissions. Among them, power generation from biomass and wind energy would be the primary options. However, as things stand in Taiwan, it will be an uphill battle to raise clean en-ergy output to the level this model suggests. Scenario 3 corresponds to scenario 2 in showing that nuclear energy is able to replace most coal-fired electricity generation, the major emitter of GHG. The low cost of nuclear power makes economic sense in this model. In terms of energy al-location structure, nuclear power has accounted for over 32.2% of the total power generation since 2010. This pro-portion will increase to 34% in the near future. Also, coal consumption would decrease to 6.5% since the beginning of the planning period. However, this allocation patterns are greatly different compared with the baseline case. Such shift might not be achieved under current energy supply situations in Taiwan. In addition, the difference nuclear power makes in lowering GHG emissions could allow for a much lower supply of clean energy, therefore lessen the pressure on its short-term supply capacity. Scenario 3 is clearly superior to scenario 2 in feasibility and cost. The catch is, it goes against the slogan‘Nuclear-Free Home-land,’ which is the policy target of the Taiwanese govern-ment proclaimed several years ago.
Table IV. Fuel and electricity demands for various sectors in base year. Sector Fuel demand (PJ/year)[42] Electricity demand (109Gkwh/year) [42] Energy Sector 124.62 18.74 Industrial Sector 1181.41 109.64 Transportation Sector 549.73 1.11 Agricultural Sector 13.50 2.58 Residential Sector 59.25 45.16 Commercial Sector 82.35 43.05
Non-energy use Sector 133.38
-Table III. Detailed considerations under scenarios 1 to 3.
Growth rate of energy demand GHG emission
Nuclear power generation Scenario 1 3% in the initial year, decreasing
0.2% annually, reaching 0% in 2025
No limitation Remain the same as
the current state Scenario 2 3% in the initial year, decreasing
0.2% annually, reaching 0% in 2025
Decreasing steadily to 2025 with the same level as in 2000 (223.0 million ton CO2-eq, decrease 1.8% annually)
Remain the same as the current state Scenario 3 3% in the initial year, decreasing
0.2% annually, reaching 0% in 2025
Decreasing steadily to 2025 with the same level as in 2000 (223.0 million ton-CO2-eq, decrease 1.8% annually)
Can develop without a cap
3.3. The trend of clean energy
The contribution of clean energy to the energy structure is determined by price competition and the requirement of
GHG reduction. According to the simulation results in Figure 4(a), scenario 1 requires two types of clean energy, RDF and biodiesel, with RDF mostly used in power gener-ation in medium load. According to the costs of different
Table V. Input parameters in base year.
Energy type
Fuel Electric power
Unit cost in base year (NTD/MJ)
GHG emission (kg-CO2eq/MJ)
Unit cost in base year (NTD/kwh) GHG emission (kg-CO2eq/kwh) Steam coal 0.1113[42] 0.0946[48] 1.0426[49] 0.9564[48] Anthracite 0.1557[42] 0.0983[48] Coke 0.3200[43] 0.1070[48]
Coke oven gas 0.1000[43] 0.0444[48]
Steam furnace gas 0.1000[43] 0.2600[48]
LPG 0.4363[44] 0.0631[48] Naphtha 0.7281[45] 0.0733[48] Motor gasoline 0.5840[44] 0.0693[48] Jet fuel 0.5270[44] 0.0715[48] Diesel oil 0.4837[44] 0.0741[48] 4.9209[49] 0.7119[48] Fuel oil 0.4232[44] 0.0774[48] 4.6258[49] 0.8219[48] Lubricants 1.2448[44] 0.0733[48] Asphalts 0.4121[44] 0.0807[48] Other 0.5856[44] 0.0733[48] Self-produced LNG 0.3607[42] 0.0561[48] LNG 0.3605[42] 0.0642[48] 3.1357[49] 0.5598[48] RDF 0.7760[46] 0.0740[46] Bio-diesel 0.8700[47] 0.0039[47] Bio-ethanol 1.3600[47] 0.0039[47] Sub-bituminous coal 1.0139[49] 0.9709[48] Wind power 2.3834[42] 0.0000 Solar power 11.1190[42] 0.0000 Hydro power 1.5393[42] 0.0000 Tidal power 15.0000[42] 0.0000 Geothermal power 5.1838[42] 0.0000 Current generation 15.0000[42] 0.0000 RDF power 2.0879[42] 0.7480[46] Nuclear power 0.6532[42] 0.0000
fuel types collected in this study, RDF is more economi-cally efficient than traditional fossil fuel and other clean energy sources. Since GHG reduction is not required in scenario 1, RDF is a viable replacement. The problem is that at this stage, Taiwan does not have the capacity to fully supply the amount called for in this simulation result. On the other hand, with the increasing cost of fossil fuel and the decreasing cost of clean energy, the cost of biodie-sel will fall below the costs of automobile and airplane fuels by 2020 and 2022 respectively, and therefore, biodie-sel will meet the needs of some industrial sectors and the aviation industry.
Figure 4(b) illustrates the impact of GHG reduction and the restriction on nuclear power to the energy structure. Scenario 2 demands the highest supply of clean energy, most of which is biodiesel, to replace automobile and avi-ation fuels, diesel, and bunker oil. Although this change in supply structure would meet the requirement of GHG re-duction, it also entails higher cost. The results also reveal negligible growth in hydroelectricity and wind power. This is because in this study, hydroelectricity is considered to have no potential for further development in Taiwan, and in practice, wind power can only generate power in me-dium load, and that limits its demand and its growth poten-tial. On the other hand, LPG would replace fuel oil in 2013 due to the restrictions on GHG emission. Then, diesel from biomass would replace jet fuel oil to a certain degree. This is because of the competition between LPG and biodiesel. In detail, with the consideration of both a minimized sys-tem cost and a minimized GHG emission, the benefit for replacing fuel oil with LPG would be 1.72kg/NTD, which would be higher than those associated with the replace-ment of biodiesel (0.28kg/NTD). In order to satisfy GHG emission restriction targets, LPG would befirstly adopted due to its economic competitiveness. After 2024 or 2025, all of the fuel oil would be replaced by biodiesel. Thus, in order to further reduce GHG emission, coal would be replaced by RDF.
As scenario 3 sets no limit on the expansion of nuclear power and its high utilization will cut down GHG emis-sions drastically, Figure 4(c) shows only moderate supply of clean energy. It should be noted that although RDF power emits more GHG per unit than wind power does, be-cause of its lower cost, RDF enjoys some advantage in the first few years when GHG emission standards are less stringent. Furthermore, hydroelectric power’s absence in
scenario 3 is due to its lack of competitiveness in compar-ison with nuclear power. While neither emits GHG, nu-clear power has the edge in cost. In addition, because of its lower price, biodiesel is set to partially replace other fuels for industrial and aviation uses after 2020; the mech-anism is the same as in scenario 1. By 2021, all of the jet fuel oil would be replaced by biodiesel, saving a certain amount of GHG emission quota. The saved quota could be used for adopting RDF for power generation at a re-duced cost and a raised GHG emission ratio, leading to retardness in the replacement of RDF. However, along with the gradual increase of GHG emission restriction, after the year of 2021, RDF would be replaced by wind energy for power generation.
Based on the above analyses, it is found that although fuel cost is lower in scenarios that do not set GHG stan-dards, the tradeoff is the exacerbation of GHG emissions. Moreover, the supply of fossil fuel is dwindling and its price increasingly unstable. In addition to the benefit of lower GHG emissions, developing clean energy and nu-clear power seems to be the only viable option for a stable and secured energy future. However, scenario 2 indicates that a fast leap in clean energy supply is required if we are to limit nuclear power while achieving GHG reduction. It will be a major challenge for energy supply in Taiwan. Still, with the further expansion of nuclear energy come the problems of nuclear wastes and whether the society is ready to accept it. Therefore nuclear is probably only a short-term solution for the transitional period. In the face of the coming structural changes, further studies are needed in order to devise a nimble strategy for clean en-ergy and nuclear power. An optimal solution is still elusive.
3.4. The trend of GHG emissions
Figure 5 presents the simulation results of GHG emissions in different scenarios. Scenario 1 shows that without an emission cap, total GHG will reach 397 million tonnes by 2025, 67.85 million tonnes more than the 2010 emis-sions and averages an annual increase of 2.1%. Scenario 2 shows that by adjusting current energy structure, it is possible to meet the GHG emission targets without resort-ing to more nuclear power, but this requires devotresort-ing more capital to energy production. In comparison, without a curb on nuclear power, GHG emissions clearly trend
Table VI. Growth of power generation cost and GHG emission under scenarios 1 to 3. GHG emission
(million tonne) Annual
growth rate (%)
The total power generation cost (trillion NT$) Annual growth rate (%) 2010 2025 2010 2025 Scenario 1 329 397 2.1% 1.44 2.403 3.47% Scenario 2 292 223 1.775% 1.45 2.698 4.23% Scenario 3 237 223 0.004% 1.40 2.416 3.7%
downwards, as seen in scenario 3. However, with growing demand and stricter reduction targets, after 2014 other sources of clean energy must be employed for further stan-dards to be met. Correspondingly, the two targets of GHG emission restriction and system cost minimization could be
achieved under scenario 3 due to the adoption of nuclear power. Comparatively, the system cost under scenario 3 is similar to that under scenario 1.
The analysis of GHG emissions in individual scenar-ios shows (Figure 6(a)(b)) that from 2010 to 2025, in
(a) scenario 1
(b) scenario 2
(c) scenario 3
scenario 1, with no cap on GHG, emissions from all en-ergy sources rise constantly. Scenario 2 relies on con-siderable supply of clean energy to replace traditional sources for fuel and power generation in order to achieve GHG reduction; as a result, emissions from oil decline dramatically. In scenario 3, GHG emissions rise and then fall in the projected period (Figure 5),
the reason being that without the requirement to limit nuclear energy, initially GHG emissions rose signi fi-cantly, but after 2015, due to increasing energy de-mand, GHG reduction targets will force a reshape of energy structure. As for the comparison between 2010 and 2025 (Figure 6(a) and (b)), only clean energy reduces its contribution to GHG by a large amount as power
(a) scenario 1
(b) scenario 2
(c) scenario 3
generated in medium load switches its source from RDF to wind power. In Table VII, energy structures under dif-ferent scenarios are presented.
4. CONCLUSIONS
In this study, an optimization model was developed for supporting identification of optimal strategies in adjusting existing energy structure in Taiwan. In this model, minimization of the total system cost was con-sidered as the objective function, which was subject to a series of constraints such as energy demands of vari-ous economic sectors, energy balance, and GHG emis-sion restrictions. Three scenarios were considered based on various restrictions of GHG emission and gen-eration patterns of nuclear power. Under the three sce-narios, different energy allocation strategies were generated. In terms of the total cost of energy supply, the scenario that restricted GHG emissions and nuclear power growth resulted in the highest one, with an aver-age annual increase of 4.2%. However, if nuclear power could be expanded, the system cost would be re-duced further before 2023 because of the relative inex-pensiveness of nuclear energy. After 2023, the cost
would be higher compared with that under scenario 1 (i.e., no limits on either GHG or nuclear power) due to more investments in wind power to meet GHG re-duction targets. The energy supply structure is directly influenced by energy cost and GHG emission reduc-tion, the most stringent scenario (i.e., scenario 2) entails the greatest dependence on clean energy, which would take up 41.8% in 2025. In comparison, with no restrictions on nuclear energy, it would further replace energy sources and would account for 34.0% of the to-tal energy.
GHG emissions would increase continuously under the scenario without restrictions. Comparatively, under scenario 2 (i.e., massive adoption of cleaner energies), GHG emissions from oil products would decrease greatly. Under scenario 3, GHG emission would be re-duced correspondingly due to the utilization of nuclear power. The development of clean energy and nuclear power would help reduce GHG emissions and to diver-sify energy supply. However, due to its cost, technolog-ical constraints, and the limitation of geography and environment, its supply capacity could hardly be ex-panded in the short term. Moreover, nuclear energy would give rise to the problems of social acceptability and nuclear wastes. Further studies are needed before we can arrive at a deft strategy for clean energy and nu-clear power in the face of the changing energy structure in Taiwan.
ACKNOWLEDGEMENTS
This work was supported by the National Science Council of Taiwan (contract no.: NSC 98-2514-S-003-006-NE). The authors would like to extend special thanks to the ed-itor and the anonymous reviewers for their constructive comments and suggestions in improving the quality of this paper.
Figure 5. The trend of greenhouse gas emission over the planning period under scenarios 1 to 3.
Table VII. Modeling results under three scenarios.
Energy type
Scenario 1 Scenario 2 Scenario 3
2010 2025 2010 2025 2010 2025 Coal 27.8% 29.6% 29.4% 26.3% 6.5% 6.9% Oil 40.0% 38.3% 42.2% 14.5% 39.3% 40.2% LNG 7.6% 7.6% 8.0% 8.5% 7.4% 7.9% Nuclear 9.8% 8.0% 10.4% 8.9% 32.2% 34.4% Clean Energy 14.8% 16.5% 10.1% 41.8% 14.6% 10.6%
REFERENCES
1. IEA. World energy outlook Executive Summary. International Energy Agency: Paris, France, 2009. 2. IEA Bioenergy. Potential Contribution of Bioenergy to
the World’s Future Energy Demand. International Energy Agency: Paris, France, 2007.
3. WEC. World Energy and Climate Policy: 2009 Assessment. World Energy Council, London, United Kingdom, 2009.
4. WEC. Deciding the Future: Energy Policy Scenarios to 2050. World Energy Council, London, United Kingdom, 2007.
5. Jenkins BM. A comment on the optimal sizing of a biomass utilization facility under constant and variable cost scaling. Biomass & Bioenergy 1997;13:1–9.
6. Rana S, Chandra R, Singh SP, Sodha MS. Optimal mix of renewable energy resources to meet the electrical energy demand in villages of Madhya Pradesh. Energy Convers. Mgmt 1998; 39(3/4): 203–216.
7. Dornburg V, Faaij APC. Efficiency and economy of wood-fired biomass energy systems in relation to scale regarding heat and power generation using combustion and gasification technologies. Biomass & Bioenergy 2001;21:91–108.
8. Yoshida Y, Dowaki K, Matsumura Y, Matsuhashi R, Li D, Ishitani H, Komiyama H. Comprehensive com-parison of efficiency and CO2 emissions between biomass energy conversion technologies- position of supercritical water gasification in biomass technolo-gies. Biomass & bioenergy 2003;25:257–272.
(a)Year 2010
(b)Year 2025
9. Sørensen B. A sustainable energy future: Construction of demand and renewable energy supply scenarios. International Journal of Energy Research 2010; 32:436–470.
10. Iniyan S, Suganthi L, Jagadeesan TR, Samuel AA. Reliability based socio economic optimal renewable energy model for India. Renewable Energy 2000; 19:291–297.
11. Iniyan S, Sumathy K. An optimal renewable energy model for various end-uses. Energy 2000;25:563–575. 12. Iniyan S, Sumathy K. The application of a Delphi technique in the linear programming optimization of future renewable energy options for India. Biomass & Bioenergy 2003;24:39–50.
13. Suganthi L, Williams A. Renewable energy in India— a modeling study for 2020–2021. Energy Policy 2000; 28:1095–1109.
14. Schmidt J, Leduc S, Dotzauer E, Kindermann G, Schmid E. Potential of biomass-fired combined heat and power plants considering the spatial distribution of biomass supply and heat demand. International Journal of Energy Research 2010;34:970–985. 15. Drozdz M. Polish energy-system modernization.
Applied Energy 2003;76:267–277.
16. Mathur J, Bansal NK, Wagner HJ. Dynamic energy analysis to assess maximum growth rates in develop-ing power generation capacity: case study of India. Energy Policy 2004;32:281–287.
17. Jebaraja S, Iniyanb S. A review of energy models. Renewable and Sustainable Energy Reviews 2006; 10:281–311.
18. Iniyan S, Suganthi L, Samuel AA. Energy models for commercial energy prediction and substitution of renewable energy sources. Energy Policy 2006; 34:2640–2653.
19. Becerra-López HR, Golding P. Multi-objective optimization for capacity expansion of regional power-generations systems: Case study of far west Texas. Energy Conversion and Management 2008; 49:1433–1445.
20. Cai YP, Huang GH, Yang ZF, Lin QG, Bass B, Tan Q. Development of an optimization model for energy systems planning in the Region of Waterloo. International Journal of Energy Research 2008; 32:988–1005.
21. Ghafghazi S, Sowlati T, Sokhansanj S, Melin S. Techno-economic analysis of renewable energy source options for a district heating project. Inter-national Journal of Energy Research 2010; 34: 1109–1120.
22. Brownsword RA, Fleming PD, Powell JC, Pearsall N. Sustainable cities – modelling urban energy supply and demand. Applied Energy 2005;82:167–180.
23. Pehnt M. Dynamic life cycle assessment (LCA) of renewable energy technologies. Renewable Energy 2006;31:55–71.
24. Gilau AM, Buskirk RV, Small MJ. Enabling optimal energy options under the Clean Development Mecha-nism. Energy Policy 2007;35:5526–5534.
25. Granovskii M, Dincer I, Rosen MA. Air pollution re-duction via use of green energy sources for electricity and hydrogen production. Atmospheric Environment 2007;41:1777–1783.
26. Blesl M, Das A, Fahl U, Remme U. Role of energy efficiency standards in reducing CO2 emissions in Germany: An assessment with TIMES. Energy Policy 2007;35:772–785.
27. Lafforgue G, Magné B, Moreaux M. Energy substitu-tions, climate change and carbon sinks. Ecological Economics 2008;67:589–597.
28. Thornley P, Upham P, Huang Y, Rezvani S, Brammer J, Rogers J. Integrated assessment of bioelectricity technology options. Energy Policy 2009;37:890–903. 29. Daniel J, Dicorato M, Forte G, Iniyan S, Trovato M. A methodology for the electrical energy system planning of Tamil Nadu state (India). Energy Policy 2009; 37:904–914.
30. Dougherty W, Kartha S, Rajan C, Lazarus M, Bailie A, Runkle B, Fencl A. Greenhouse gas reduction ben-efits and costs of a large-scale transition to hydrogen in the USA. Energy Policy 2009;37:56–57.
31. Chau J, Sowlati T, Sokhansanj S, Preto F, Melin S, Bi X. Techno-economic analysis of wood biomass boilers for the greenhouse industry. Applied Energy 2009; 86:364–371.
32. Soimakallio S, Mäkinen T, Ekholm T, Pahkala K, Mikkola H, Paappanen T. Greenhouse gas balances of transportation biofuels, electricity and heat genera-tion in Finland-Dealing with the uncertainties. Energy Policy 2009;37:80–90.
33. Lin QG, Huang GH. Planning of energy system man-agement and GHG-emission control in the Municipality of Beijing—An inexact-dynamic stochastic program-ming model. Energy Policy 2009;37:4463–4473. 34. Li YF, Huang GH, Li YP, Xua Y, Chen WT.
Re-gional-scale electric power system planning under uncertainty—A multistage interval-stochastic integer linear programming approach. Energy Policy 2010; 38:475–490.
35. Cai YP, Huang GH, Yang ZF, Lin QG, Tan Q. Community-scale renewable energy systems plan-ning under uncertainty—An interval chance-constrained programming approach. Renewable and Sustainable Energy Reviews 2009;13:721–735. 36. Cai YP, Huang GH, Yang ZF, Tan Q. Identification of
planning under multiple uncertainties. Applied Energy 2009;86:480–495.
37. Cai YP, Huang GH, Tan Q. An inexact optimization model for regional energy systems planning in the mixed stochastic and fuzzy environment. International Journal of Energy Research 2010;33:443–468. 38. Lin QG, Huang GH. A dynamic inexact energy
sys-tems planning model for supporting greenhouse-gas emission management and sustainable renewable en-ergy development under uncertainty - A case study for the City of Waterloo, Canada. Renewable and Sustainable Energy Reviews 2009;13:1836–1853. 39. Lin QG, Huang GH, Bass B, Qin XS. IFTEM: An
in-terval-fuzzy two-stage stochastic optimization model for regional energy systems planning under uncer-tainty. Energy Policy 2009;37:868–878.
40. Bureau of Energy. The Annual Report of Energy Statistics. Ministry of Economic Affairs: Taipei, Taiwan, 2009.
41. IEA. World energy outlook. International Energy Agency: Paris, France, 2009.
42. Bureau of Energy, Ministry of Economic Affairs, Taiwan. http://www.moeaboe.gov.tw/ 2010
43. China Chemical Network. http://www.chemcp.com/ news/price/jiaotan/ 2010.
44. Marketing Business Div, CPC Corporation, Taiwan. http://www.cpc.com.tw 2010.
45. Knowledge database of finance and economics, Money DJ. http://www.funddj.com/KMDJ/Report/ ReportHome.aspx. 2010.
46. Hung MC, Ning SK. Life Cycle Assessment of Refuse Derived Fuel. The 21th Conference on Environmental Planning and Management. Taiwan, 2008.
47. Chang CY. Life Cycle Assessment of Bioenergy- An Example of Fuel Ethanol Production from Sugarcane. M.S. thesis, Institute of Natural Resources Manage-ment. 2006.
48. GHG Emissions Registry Platform, Environmental Protection Administration Executive Yuan, Taiwan. http://www.ghgregistry.tw/index.aspx. 2010
49. Taiwan Power Company. http://www.taipower.com. tw/. 2010.
