Applied Energy 95 (2012) 186–195

Contents lists available at SciVerse ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

A multi-objective optimization model for sustainable electricity generation

and CO2 mitigation (EGCM) infrastructure design considering economic
profit and financial risk
Jee-Hoon Han, Yu-Chan Ahn, In-Beum Lee ⇑
Department of Chemical Engineering, POSTECH, Pohang, Republic of Korea

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

Article history: A large number of research works were undertaken for the planning of sustainable electricity generation
Received 9 August 2011 and CO2 mitigation (EGCM) infrastructure design under uncertainty. The typical methodologies assessed
Received in revised form 10 February 2012 the performance of the problem under the variability of the uncertain parameters by optimizing the
Accepted 14 February 2012
expected value of the objective function. This approach can have large probabilities of the value opti-
Available online 9 March 2012
mized in unfavorable scenarios. In this paper, we present a mathematical programming model in plan-
ning sustainable electricity generation and CO2 mitigation (EGCM) infrastructure design, including
Keywords:
financial risk management under uncertainty. The proposed model allows us to determine available tech-
Electricity generation
CO2 mitigation
nologies to produce electricity and treat CO2 on the purpose of maximizing the expected total profit and
Infrastructure minimizing the financial risk of handling uncertain environments (i.e. CO2 mitigation operating costs,
Uncertainty carbon credit prices and electricity prices, etc.), while fulfilling electricity demands and CO2 mitigation
Multi-objective optimization standards. The multi-objective optimization problem was solved by using the weighted-sum method that
imposes a penalty for risk to the objective function. The capability of the proposed modeling framework
is illustrated and applied to a real case study based on Korea, for which valuable insights are obtained.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction benefits. To supplement the technical and economic limit of CCS,

carbon emission trading (CET) should be used for managing GHG
Greenhouse gases (GHGs) emissions are expected to cause sig- emission in energy systems as a policy based incentive.
nificant global climate change [1]. The most significant GHG is A large number of research works were undertaken for the plan-
CO2, which arises mainly from use of fossil fuels in power genera- ning of electricity generation and CO2 mitigation (EGCM) strategies
tion [2]. There has been concern about whether energy supplies such as CCS and CET under meeting the GHG mitigation standards.
can meet increasing electricity demands with the reduction of The mathematical programming models have been proposed that
GHG emissions. The most common approach to reducing CO2 emis- address the design of the EGCM infrastructure; Carbon capture
sions in electricity generation system is to replace the use of car- and storage (CCS) technologies were used to reduce GHG emitted
bon-based fossil fuels with renewable energy sources or less from power plants during electricity generation process [3–6].
GHG intensive fuels. However, we will need to rely on fossil fuels Carbon emission trading (CET) were also proposed to help attain
for several decades before alternative energy sources are fully reduction of GHG emission in a cost-effective way [7–9]. However,
developed. Thus, capture and storage (CCS) technologies will be a these studies address separately CET and CCS to mitigate CO2 emis-
promising solution to reduce CO2 emissions and keep the current sion within EGCM infrastructure. It needs to identify both the cost
energy system [1], which separate CO2 from various emission and spatial arrangement of an integrated CCS and CET system due
sources (i.e., the power plant), transport them to a storage location to the large economic benefits achieved by this process. We pro-
and isolate them from the atmosphere for a long period [3]. It posed an optimization model for EGCM infrastructure that gener-
should consider geopolitical factors such as the location and capac- ates a fully integrated, profit-maximizing CCS and CET system [4].
ity of the sequestration site. Also, it can pose significant cost The previous study addresses deterministic approaches assum-
burdens to small-scale power plant because large-scale CCS to cap- ing that all problem parameters are invariant over a given planning
ture and store large amounts of CO2 brings considerable economic horizon. However, uncertainties may exist in various impact fac-
tors of the EGCM system such as GHG emission inventory, GHG
reduction costs, electricity prices and emission reduction credits.
⇑ Corresponding author. Tel.: +82 54 279 2274; fax: +82 54 279 5528. To obtain more realistic results, these uncertainties may affect
E-mail address: [email protected] (I.-B. Lee). modeling in the design of an EGCM system.

0306-2619/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apenergy.2012.02.032
 J.-H. Han et al. / Applied Energy 95 (2012) 186–195 187

Nomenclature

Indices USCi,s,r unit sequestration cost in scenario r for CO2 in physical

e product form of electricity form i sequestered by sequestration facility type s
f facility name for electricity generation UTCPoffi,l,d,r unit transport cost in scenario r for CO2 in physical
g geographical region form i transported by pipeline with diameter d offshore
g0 geographical region (g0 – g) UTCPoni,l,d,r unit transport cost in scenario r for CO2 in physical
i physical form of CO2 form i transported by pipeline with diameter d onshore
p type of power plant EPi,c energy penalty for type of CO2 capture facility c
r scenarios
c type of capture facility Variables
d pipeline diameter E[TNP] expected total net profit
l type of transport mode TNPr total net profit in scenario r
s type of sequestration facility TNBr total net benefit in scenario r
TNCr total net cost in scenario r
Sets CCSCCi capital cost of CCS facilities for CO2
x feasibility set for first-stage decision variables CCSOCi,r operating cost of CCS facilities for CO2 in scenario r
yr feasibility set for second-stage decision variables in sce- Ge,p,f,g amount of electricity generated by electricity facility f of
nario r plant type p in region g
k set of profit targets AEPi,p,f,g CO2 emission permit reallocated to electricity facility f
of plant type p in region g
Parameters Risk (x, X) financial risk of solution x at a profit target X
probr probability of occurrence of scenario r Zr,k binary variable equal 1 if the profit of scenario r is smal-
UNBp,g,r unit net benefit of selling electricity generated from ler than the profit target Xk
type of power plant p into region g in scenario r ETCi,r emission trading cost for CO2 in scenario r
Cpricer price of carbon emission credits in scenario r CFCi,r CCS facility cost for CO2 in scenario r
X target profit BCi,c,p,f,g 1 if CO2 in physical form i is captured by capture facility
qk goal programming weight for financial risk formations type c in electricity facility f of plant type p in region g, 0
CCR capital charge rate – payback period of capital investment otherwise
LR learning rate-cost reduction as technology manufactur- NSi,s,g number of well or injection facilities of type s sequester-
ers accumulate experience ing CO2 in physical form i in region g
CCCi,c,p,f,g capital cost of building capture facility type c capturing NTPoffi,l,g,g0 ,d number of pipeline with diameter d for transporting
CO2 in physical form i in electricity facility f of plant CO2 in physical form i between regions g and g0 offshore
type p in region g NTPoni,l,g,g0 ,d number of pipeline with diameter d for transporting
SCCi,s capital cost of establishing sequestration facility type s CO2 in physical form i between regions g and g0 onshore
sequestering CO2 in physical form i Ci,c,p,f,g amount of CO2 in physical form i captured by capture
TCCPoffi,l,g,g0 ,d capital cost of establishing pipeline with diameter d facility type c in electricity facility f of plant type p in re-
offshore to transport CO2 in physical form i from harbor gion g
region g onshore to sequestration region g0 offshore Si,s,g amount of CO2 in physical form i sequestered by seques-
TCCPoni,l,g,g0 ,d capital cost of establishing pipeline with diameter d tration facility type s in region g
to transport CO2 in physical form i between regions g Qi,l,g,g,d flow rate of CO2 in physical form i transported by trans-
and g0 onshore port mode l (pipeline) with diameter d between regions
UCCi,c,p,r unit capture cost in scenario r for CO2 in physical form i g and g0
captured by capture facility type c in power plant p

Several research efforts were conducted for dealing with vari- Therefore, this study aims to address the financial risk manage-
ous uncertainties in the EGCM infrastructure. For example, interval ment associated with the planning of the EGCM infrastructure un-
mathematical programming (IMP) and stochastic mathematical der uncertainty in prices (i.e. the electricity price and carbon credit
programming (SMP) models were proposed to the design of the price) and operating costs (i.e. the carbon capture and sequestra-
EGCM infrastructure under uncertainties [10–16]. These models al- tion cost). A multi-objective optimization problem which consists
low assessing the performance of the problem under the variability of maximizing the expected total profit of the infrastructure and
of the uncertain parameters typically by optimizing the expected minimizing the financial risk is generated to consider this problem.
value of the objective function. These approaches can lead to solu- Hence, the weighted-sum method is also presented to expedite the
tions that perform well on average but have large probabilities of search for the Pareto solutions of the model. Finally, the capability
unfavorable scenarios. Hence, the introduction of a financial risk of the proposed model is illustrated through its application to a
management enables to control the variability of the objective real case study based on Korea.
function in the space of uncertain parameters [17]. The financial
risk management is to incorporate the trade-off between financial
risk and expected profit into the decision-making procedure, which 2. Problem statement
rises to a multi-objective optimization problem. This offers the
opportunity of reducing the impact of unfavorable scenarios. The design problem addressed in this study has as objective to
Moreover, few research works have adopted financial risk manage- determine the optimal configuration of an EGCM infrastructure
ment techniques in designing the EGCM infrastructure under with the goal of maximizing the expected total profit and minimiz-
uncertainties. ing financial risk. This infrastructure model includes three main
188 J.-H. Han et al. / Applied Energy 95 (2012) 186–195

components such as an environmental management system, the The presented model considers the uncertainty of the coeffi-
power plants themselves, and an energy management system. In cients of the objective function via a multi-scenario stochastic pro-
this study, three power plants (i.e. gas-fired power plant, petro- gramming approach. The prices and operating costs associated
leum-fired power plant, and coal-fired power plant) are considered with the design and operation of the EGCM infrastructure are
as major CO2 emission sources. In each power plant, two mitiga- uncertain and stochastic variables that follow a set of scenarios
tions methods are used to reduce the amount of CO2 emission: with given probability of occurrence. In fact, it aims to simulta-
(i) carbon capture and storage (CCS), and (ii) carbon emission trad- neously maximize the expected value of the profit distribution as
ing (CET). well as minimize its financial risk level.
The decision-making problem of the EGCM infrastructure mod-
el is to determine where and how to generate electricity and treat 3.1. Expected profit
CO2 under the given conditions, which include electricity demand,
CO2 reduction target, capacity limitations of electricity generation The expected total profit is given by the mean value of the profit
technologies and CO2 mitigation technologies, uncertain parame- discrete distribution:
ters (i.e. prices and operating costs); in order to simultaneously X
E½TNP ¼ probr ; TNPr ð1Þ
maximize the expected total profit and minimize the associated
r
financial risk of the EGCM infrastructure.
r is a subscript that represents a particular scenario and probr is the
probability of occurrence of this scenario. The total net profit
3. Model formulation attained in each particular scenario TNPr is calculated by the differ-
ence between total net benefit TNBr and total net cost TNCr:
In the previous work [4], a deterministic model is used to deter-
mine the optimal electricity generation and CO2 mitigation policy TNPr ¼ TNBr  TNC r 8r: ð2Þ
for a given planning horizon. The deterministic model assumes
that we know with certainty all the parameters present in the 3.1.1. Total net benefit
model. This assumption is very restrictive, thus it will serve as The total net benefit TNBr in each particular scenario r is the in-
the basis for a more complex extension, which is described in the come from selling electricity generated by power plants:
next sub-section. This extension replaces some of the uncertain XXXX
parameters through a finite set of scenarios in the model TNBr ¼ UNBp;g;r Ge;p;f ;g 8r; ð3Þ
formulation. e p f g

Table 1
Analysis conditions selected for the case study.

Case Objective Model Description

1 Single objective {Max TNP} Deterministic  Generation technology: Gas-fired, petroleum-fired and coal-fired electricity generation
 Capture technology : The absorption and desorption of carbondioxide in aqueous
monoethanolamine (MEA)
 Sequestration method :Depleted gas reservoir (DGR) and saline aquifer storage (SAS)
 Transportation mode: Liquid CO2, or LCO2 via pipeline
2 Single objective {Max E[TNP]} Stochastic The same as above
3 Two objectives {Max E[TNP]; Min Risk (x, X)} Stochastic The same as above
 Case 1 was described in [4].
 Case 2: The two-stage stochastic programming is employed to address optimization under uncertainty relies on formulating a single-objective optimization problem
where the expected performance of the system is the objective to be optimized.
 Case 3: The weighed-sum method is employed to address a multi-objective optimization problem in which the expected economic performance and a specific risk
measureare the objectives considered.

Table 2
Parameters used for estimating uncertain prices and operating costs.

Parameters Activity Type Mean value Variance

(%)
Prices Generation (UNBp,g,r)a Gas-fired Average value of each uncertain parameter is offered from [4] 10.6
Petroleum-fired 41.9
Coal-fired 8.6
Carbon emission trading 50
(Cpricer)b
Operating costs Capture (UCCi,c,p,r) Gas-fired 26.6
b

Petroleum-fired 48.8
Coal-fired 20.7
Sequestration (USCi,s,r) DGR 36.1
SAS 43.7
Transportation Onshore 25.0
(UTCPoni,l,d,r)
Offshore 12.0
(UTCPoffi,l,d,r)
a
Variance of each uncertain price was considered based on historic data provided by [19].
b
Variance of each uncertain operating cost and CETwas based on the economic analysis of [1].
 J.-H. Han et al. / Applied Energy 95 (2012) 186–195 189

Table 3
Summary of computational results for the examined model.

Case
Quantity 1 2 3a
Number of constraints 31,722 108,849 109,000
Number of integer variables 1284 1284 1334
Number of continuous variables 32,869 109,996 110,047
Optimality gap (%) 0.0 0.0 0.0
CPU time (s) 11.30 18.10 51.78
a
Xk: 1.2450  1010, qk: 5.0  107.

Fig. 2. CO2 treated by CCS facilities in each plant type associated with the extreme
where UNBp,g,r is the unit net benefit selling electricity generated solutions.

from type of power plant p into region g in scenario r and Ge,p,f,g

denotes the amount of electricity generated by electricity facility f
of plant type p in region g.

3.1.2. Total net cost

TNCr is the sum of emission trading cost ETCi,r and CCS facility
cost CFCi,r:
X
TNC r ¼ ðETC i;r þ CFC i;r Þ 8r: ð4Þ
i

3.1.3. Emission trading cost

ETCi,r is associated with the total cost of CET (Carbon Emission
Trading) in all power plants. It is calculated by multiplying the
price of carbon emission credit Cpricer in each particular scenario Fig. 3. Carbon emission credit traded in each plant type associated with the
r by the CO2 emission permit reallocated to electricity facility f of extreme solutions.

plant type p in region g AEPi,p,f,g:

XXX
ETC i;r ¼ Cpricer AEP i;p;f ;g 8i; r: ð5Þ 0 0P P P P
CCC i;c;p;f ;g BC i;c;p;f ;g þ SCC i;s NSi;s;g
11
p f g CCR BX B c p f
C C
s
CCSCC i ¼ @ @P P P AA 8i:
LR g TCCPoni;l;g;g0 d NTPoni;l;g;g0 ;d þ TCCPoffi;l;g;g0 ;d NTPoffi;l;g;g0 ;d
g0 l d

3.1.4. CCS facility cost ð7Þ

CFCi,r consists of CCS facility capital cost CCSCCi and CCS facility
operating cost CCSOCi,r: where CCR denotes capital charge rate, or depreciated present value
per year over the lifetime of the system, and the learning rate LR
CFC i;r ¼ CCSCC i þ CCSOC i;r 8i; r; ð6Þ
denotes the reduction in the cost of CCS technologies as experience
where the capital cost (e.g., CCSCCi) can be regarded as non-scenario- accumulates over time, or the factor to convert present costs into
dependent variables, whereas the operating cost (e.g., CCSOCi,r) will future costs [4].
in general depend on the specific scenario realization.CCSCCi is the CCSOCi,r is obtained by multiplying the unit capture, sequestra-
total cost of establishing capture facilities, sequestration facilities tion and transport costs (UCCi,c,p,r, USCi,s,r, UTCPoni,l,d,r and
and transport modes. It is calculated by multiplying the required UTCPoffi,l,d,r, respectively), which are regarded as uncertain param-
number of CCS facilities (BCi,c,p,f,g, NSi,s,g, NTPoni,l,g,g0 ,d, NTPoffi,l,g,g0 ,d) eters, with the corresponding amounts of capture, sequestration
and their capital costs: and transport:

Fig. 1. Pareto optimal solution curve.

190 J.-H. Han et al. / Applied Energy 95 (2012) 186–195

Fig. 4. CO2 capture, sequestration and transport activities of CCS for the minimum risk solution.

Fig. 5. CO2 capture, sequestration and transport activities of CCS for the maximum profit solution.
 J.-H. Han et al. / Applied Energy 95 (2012) 186–195 191

Fig. 6. Solutions for the two-stage stochastic model (case 2) with different financial risk levels.

Fig. 7. Different kinds of financial risk curves obtained using the multi-objective optimization model (case 3).

0 ! 1
P PPP P In Eq. (9) TNP(x) is the total annual profit associated to an EGCM
B UCC i;c;p;r C i;c;p;f ;g þ USC i;s;r Si;s;g C
B g p f c s C infrastructure design x, that is, the profit after the uncertainty has
XBB P P P P  C
C been unveiled and a scenario has been realized.
CCSOC i;r ¼ Bþ UTCPoni;l;d;r þ UTCPoffi;l;d;r Q i;l;g;g0;d C 8i;r;
B l g g0 d
i B
C
C
@ PPPP A
þ UNBp;g;r EP i;c C i;c;p;f ;g 3.3. Multiobjective problem
c p f g

ð8Þ The EGCM infrastructure design in this study is mathematically

where the last part of the right-hand-side denotes the energy pen- formulated as follows:
alty cost of the capture facility. Maximize E½TNPðx; yr Þ; Minimize Riskðx; Xk Þ
The detailed explanations for the first objective and its con-
Subject to :
straints were described by Han and Lee [4].
hðx; yr Þ ¼ 0 fOverall mass balanceg ð11Þ
3.2. Financial risk gðx; yr Þ 6 0 fCapacity limitationsg
x 2 N; yr 2 R:
Financial risk associated with a design project can be defined as
the probability of not meeting a certain target profit (maximiza- The objective of this formulation is to find values of the strate-
tion) or cost (minimization) level referred to as X [17]. The finan- gic (xeN) and operational (yr eR) decision variables, subject to the
cial risk associated with a design x and target profit Xk can be set of equality (h (x, y_r = 0)) and inequality constraints
expressed as the following probability: (gðx; yr Þ  0). The constraints were described in [4]. In the above
X model, the continuous operational variables (yr) corresponding to
Riskðx; Xk Þ ¼ P½TNPðxÞ < Xk  ¼ probr zr;k ðx; Xk Þ 8k ð9Þ scenario s indicate decisions concerned with electricity generation,
r
CET utilization and CCS utilization rate, whereas the discrete stra-
A reasonable assumption is to be neutral (unbiased) about the tegic variables (x) represents the investment decisions such as
expected risk. This assumption means that these scenarios have selection of the activity types. On the other hand, we can expect
an equal probability (probr = 1/the number of scenarios) [17]. Also, that there will be a conflict between the expected profit and the
zr,k is a binary variable defined for each scenario, as follows: financial risk because the highest-profit infrastructure does not
  guarantee the least risk one [17]. Because of this trade-off for a
1 if TNPðxÞ < Xk
zr;k ðx; Xk Þ ¼ 8 k; r ð10Þ continuous range of profit target Xk, there is no single solution
0 otherwise to this type of problem. Instead, the solution consists of a set of
192 J.-H. Han et al. / Applied Energy 95 (2012) 186–195

Fig. 8. (A) Electricity generation in each plant type according tothree different case studies. (B) Power plants which show the variation of electricity generation in specific
regions according to three different case studies.

Pareto optimal configurations for the EGCM infrastructure. The reduction target is about 30% of total CO2 emissions; (3) gas-fired,
Pareto solutions are obtained by applying the weighed-sum meth- petroleum-fired and coal-fired power plants as major sources of
od [18]. In this work to manage risk, it is done by imposing a pen- electricity generation and CO2 emission which consist of 39 power
alty for risk at different target profits (Xk). The multiobjective plants (15 gas-fired, 11 petroleum-fired, 13 coal-fired) in 16 re-
model includes a goal programming weight qk  0 in the objective gions; (4) several electricity generation, CO2 capture and seques-
function to obtain solutions where the relative importance of tration technologies and transport modes (Table 1). Also, upper/
expectation and risk is progressively changed, as presented next. lower boundary capacity, capital cost and unit operation cost of
P P each component for CCS such as capture facility, sequestration
Maximize E½TAPðx; yr Þ  probr qk ð1  zr;k Þ facility and transport mode are offered from the study of Han
r2R k2K
and Lee [4].
Subject to :
Specially, this study examines significantly the prices (UNBp,g,r
hðx; yr Þ ¼ 0 fOverall mass balanceg and Cpricer) and the operating costs (UCCi,c,p,r, USCi,s,r, UTCPoni,l,d,r
gðx; yr Þ 6 0 fCapacity limitationsg and UTCPoffi,l,d,r). The versatility of the proposed model derived
x 2 N; yr 2 R (Section 3) is compared with case studies that vary according to
probr TNPr 6 Xk þ U r ð1  zr;k Þ 8k ¼ 1; . . . ; NK; r ¼ 1; . . . ; NR; the objective and model types (Table 1).
For the model, this study used data and parameters from [4].
probr TNPr P Xk  U r zr;k 8k ¼ 1; . . . ; NK; r ¼ 1; . . . ; NR:
However, the addition of two extra features, i.e. uncertainty in
ð12Þ prices or operating costs, to the model presented in paper [4]
4. Case study makes the process of collecting data even more complicated. The
planning horizon for this case study is the year 2020. During the
To illustrate the capabilities of the proposed model, the case period, we devote our attention to the uncertainty associated with
study in paper [4] is adopted as a benchmark. This case study con- the prices and operating costs, and assume that the other operating
siders (1) the planning of EGCM infrastructure in 2020; (2) CO2 parameters can be perfectly known in advance. Also, since the
 J.-H. Han et al. / Applied Energy 95 (2012) 186–195 193

Fig. 9. CO2 mitigation amounts in each plant type according to three different case studies. (A) Allocated CO2 emission permit. (B) Carbon emission credits traded. (C) CO2
treated by CCS.

parameters of price and operating cost coefficients are inexact, the 3.16 GHz. The short times required for solving the three different
uncertainties can be expressed as probability density functions case studies and the low optimality gaps (Table 3) are satisfactory.
(PDFs) and discrete interval values. Thus, they are assumed to Having proved the application and computational effectiveness
follow normal distributions with a mean [4] and variance [1,19] of the proposed model, we next use it to determine the Pareto set
(Table 2). The uncertain parameters were described by 50 scenar- of the multi-objective programming problem. As shown in the
ios using Monte Carlo sampling. Particularly in our case study Fig. 1, there is a clear trade-off between the expected profit and
using the statistic-based methodology [20], with a sample com- financial risk. Each point of the Pareto set corresponds to a specific
posed by 50 scenarios we can ensure a relative error of 0.5% and EGCM infrastructure design. The maximum profit solution (point B)
a confidence interval of 99.9% for the value of the E[TNP] obtained. corresponds to the single-objective problem that would arise from
maximizing exclusively the expected total profit, as introduced in
Section 3. On the other hand, the minimum risk solution corre-
5. Results and discussions sponds to point A of Fig. 1. The specific EGCM infrastructure designs
of these solutions are also shown in Figs. 1–5. Specifically, in these
The proposed model was computed using CPLEX 9.0 (GAMS) in particular examples we aim to analyze the impact of the large var-
on a computer equipped with a Pentium 4 chip, operating at iability in the carbon credit price and petroleum price on the EGCM
194 J.-H. Han et al. / Applied Energy 95 (2012) 186–195

infrastructure design. Fig. 1 shows that in order to minimize the emissions trading (CET) and carbon capture and storage (CCS). The
financial risk, the model resolves to reduce a petroleum-fired gen- model also determines where and how much electricity to be gen-
eration. This is because the petroleum price shows higher variabil- erated and distributed. It is a grandiose attempt to include a large
ity than the coal and natural gas. Let us note that electricity set of integrated decisions within a single mathematical modeling
generation plants utilize CCS or CET for reducing the GHG emissions framework considering uncertainties.
(Figs. 2 and 3). The minimum financial risk solution, which corre- The problem has been considered as a multi-objective stochas-
sponds to point A of Fig. 1 and to results depicted in Figs. 2 and 3, tic MILP that simultaneously accounts for the maximization of the
entails that CCS is utilized and CET is reduced overall in the entire expected total profit and minimization of the financial risk. The
power system. This is because the price variability of carbon credit weighed-sum method has also been employed in order to expedite
had direct effect on the configuration of CO2 mitigation. Note that the solution of such model. Thus, the model explicitly incorporates
Fig. 4 shows a more decentralized EGCM infrastructure, while in the trade-off between risk and profit at the decision-making level.
Fig. 5 a considerable increment on transportation amounts, despite The effectiveness of the proposed approach as a decision-making
a considerable decrement on transportation links, between regions tool capable of providing valuable insights into the EGCM design
is observed. This is mainly due to the less variability of coal-fired problem has been also highlighted. Simulation results have shown
power plants, which shows the risk advantages of building larger that the petroleum-fired electricity generation should be replaced
plants that implement this technology versus transporting the CO2. by coal-fired one to reduce the variability of the profit distribution.
Also, consider the profit histogram of two generic feasible solu- We can propose the optimal operation of each power plant under
tions for case ‘2’ shown in Fig. 6. The first design has a higher uncertainty. These characteristics of the model can help policy
expected profit ($1.2419  1010) than the second one (($1.2405  managers to enhance our knowledge on how to establish optimal
1010); however, Design I is riskier than Design II because financial EGCM infrastructures capable of dealing with uncertainties in real-
loss can occur under several scenarios. To manage financial risk istic problems.
for the above-described problem, the multi-objective programming Finally, nuclear power generation plays an important role in
model was solved, obtaining the solution that maximizes the electricity generation in Korea. Besides CET and CCS, replacing
expected net present value and minimizes the financial risk. Fig. 7 existing fossil fuel plants with nuclear plants can provide more
illustrates a hypothetical example with three risk curves responding valuable insights on making investment strategies for planning
to these different profit targets. The figure reveals that for high optimal EGCM infrastructures. Thereby it would be worthwhile
target profits (higher than 1.240  1010 $), the solution shows a to extend the proposed modeling approach for developing a new
level of risk lower than the one for low target profits. A risk-averse objective function including building a new nuclear power plant
decision maker will prefer to lower the target profit, whereas a risk in terms of addressing another option of CO2 mitigation method
taker will prefer to higher the target profit. The trade-off lies in as well as evaluating the effects of CET and CCS in existing fossil
between these extreme solutions, which reflect different possible fuel plants as a future work.
attitudes towards.
The optimized power system also shows the annual electricity Acknowledgment
generation and CO2 mitigation in each power plant according to
the case studies (Figs. 8 and 9). The electricity generation in petro- This paper was supported by the Korea Research Foundation
leum power plants is lower in case ‘3’ than in other cases ‘1’ and ‘2’ Grant funded by the Korea Government (MOEHRD, Basic Research
due to controlling the variability of the petroleum price (Fig. 8). Promotion Fund) (KRF-2008-313-D00178).
Specially, in case ‘2’ electricity generation ceased in KOMIPO5
coal-fired power plant in region ‘g11’, while in case ‘3’ one ap- References
peared. This implies that the case 3 considering the risk manage-
ment will yield better results at a certain level considering the [1] Metz B, Davidson O, De Coninck H, Loos M, Meyer L. IPCC Special Report on
Carbon Dioxide Capture and Storage; 2005.
whole profit distribution than the case ‘2’ only optimizing the ex- [2] InternationalEnergyAgency. Energy technology perspectives; 2010.
pected economic performance in uncertainty. The optimal solutions [3] Han J-H, Lee I-B. Development of a scalable and comprehensive infrastructure
for CO2 treated by different CO2 mitigation methods in the power model for carbon dioxide utilization and disposal. Ind Eng Chem Res
2011;50:6297–315.
plants were obtained (Fig. 9). The configuration of CO2 mitigation
[4] Han J-H, Lee I-B. Development of a scalable infrastructure model for planning
is significantly variant in the case studies, unlikely that of electricity electricity generation and CO2 mitigation strategies under mandated reduction
generation. The reason is that the carbon credit price shows higher of GHG emission. Appl Energy 2011;88:5056–68.
variability than the petroleum price. Specially, coal-fired power [5] Nasiri F, Huang G. Integrated capacity planning for electricity generation: a
fuzzy environmental policy analysis approach. Energy Sources Part B
plants in regions ‘g11’ and g’15’ are vulnerable to changes in the 2008;3:259–79.
carbon credit price because the amount of CO2 treated by CET or [6] Han J-H, Lee J-U, Lee I-B. Development of a multi-period model for planning
CCS increased considerably in case ‘2’. Hence, results in case ‘3’ indi- CO2 disposal and utilization infrastructure. Ind Eng Chem Res; in press.
[7] Copeland BR, Taylor MS. Free trade and global warming: a trade theory view of
cate that lower levels of risk can be obtained by allocating more CO2 the Kyoto protocol. J Environ Econ Manage 2005;49:205–34.
emission permits than case ‘2’. Also, the amount of CO2 treated by [8] Kemfert C, Lise W, Tol RSJ. Games of climate change with international trade.
CCS facilities in the petroleum-fired power plants does not appear Environ Resource Econ 2004;28:209–32.
[9] Carraro C, Siniscalco D. Strategies for the international protection of the
in this result. We can propose switching from petroleum-fired environment. J Public Econ 1993;52:309–28.
power plants to coal-fired power plants at the design step to in- [10] Chen WT, Li YP, Huang GH, Chen X, Li YF. A two-stage inexact-stochastic
crease the total net profit for the entire power industry in respond- programming model for planning carbon dioxide emission trading under
uncertainty. Appl Energy 2010;87:1033–47.
ing to climate change under uncertainty. [11] Li YP, Huang GH, Chen X. Planning regional energy system in association with
greenhouse gas mitigation under uncertainty. Appl Energy 2011;88:599–611.
[12] Li YP, Huang GH, Nie XH, Nie SL. An inexact fuzzy-robust two-stage
programming model for managing sulfur dioxide abatement under
6. Conclusions
uncertainty. Environ Model Assess 2008;13:77–91.
[13] Huang GH, Sae-Lim N, Liu L, Chen Z. An interval-parameter fuzzy-stochastic
This work has introduced a mathematical model for risk man- programming approach for municipal solid waste management and planning.
agement in the strategic design and planning of the EGCM infra- Environ Model Assess 2001;6:271–83.
[14] Jee-Hoon Han, Jun-Hyung Ryu, Lee I-B. Developing a two-stage stochastic
structure under uncertainty in prices and operating costs. The programming model for CO2 disposal planning under uncertainty. Ind Eng
model integrates the major CO2 mitigation methods such as carbon Chem Res; in press.
 J.-H. Han et al. / Applied Energy 95 (2012) 186–195 195

[15] Han J-H, Lee I-B. Two-stage stochastic programming model for planning CO2 [18] Haimes YV, Lasdon LS, Wismer DA. On a bicriterion formation of the problems
utilization and disposal infrastructure considering the uncertainty in the CO2 of integrated system identification and system optimization. IEEE Trans Syst
emission. Ind Eng Chem Res 2011;50:13435–43. Man Cybern 1971;SMC-1:296–7.
[16] Jee-Hoon Han, Lee I-B. A dynamic stochastic optimization model for carbon [19] KoreaElectricPowerCOrporation(KEPCO). The Monthly Report on Major
capture and storage (CCS) infrastructure under uncertainty in CO2 emissions, Electric Power Statistics; 2010. p. 377.
product prices and operating costs. Ind Eng Chem Res; in press. [20] Ehrgott M, Gandibleux X. A survey and annotated bibliography of
[17] Barbaro A, Bagajewicz MJ. Managing financial risk in planning under multiobjective combinatorial optimization. OR Spektrum 2000;22:425–60.
uncertainty. AIChE J 2004;50:963–89.