Design multiperiod optimization model for the electricity sector under uncertainty – A case study of the Emirate of Abu Dhabi

Energy Conversion and Management 100 (2015) 177–190

Contents lists available at ScienceDirect

Energy Conversion and Management journal homepage: www.elsevier.com/locate/enconman

Design multiperiod optimization model for the electricity sector under uncertainty – A case study of the Emirate of Abu Dhabi Alberto Betancourt-Torcat, Ali Almansoori ⇑ Department of Chemical Engineering, The Petroleum Institute, Abu Dhabi, P.O. Box 2533, United Arab Emirates

a r t i c l e

i n f o

Article history: Received 1 February 2015 Accepted 1 May 2015 Available online 17 May 2015 Keywords: Multiperiod Uncertainty UAE Power system Nuclear power Renewables

a b s t r a c t In this study, a multiperiod model that considers uncertainty in the gas feedstock fuel price is developed for the optimal design of electric power systems. The optimization problem was formulated as a multiperiod stochastic programming model using the GAMSÒ modeling system. Previous studies have analyzed the United Arab Emirates’ (UAE) power infrastructure either using a deterministic point of view or simulation tools (e.g., MESSAGE and MARKAL). These previous research has demonstrated that natural gas will remain playing a significant role as key feedstock fuel in the UAE’s power sector. However, the present work is designed to be the first to consider different supply options for the natural gas feedstock (i.e., domestic, pipeline imports, and LNG imports) and electricity imports in the UAE power sector. Moreover, the natural gas supply and electricity import options are considered to be decision variables in the problem’s formulation. Additionally, the considered case studies assumed a realistically existing power infrastructure for the UAE, whereas previous works considered the planning of the UAE power infrastructure as a Greenfield project. Also, to the authors’ knowledge this is the first work to consider a robust optimization model for planning the UAE power infrastructure under uncertainty in the long term horizon. The model was used to study the planning of the power plant infrastructure in the UAE between 2015 and 2040 under uncertainty in the natural gas price. The optimization results show that the model is a valuable tool for planning the optimal power plant infrastructure of the country, reducing levelized electricity costs, and mitigating social and environmental damages. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction The majority of the electricity produced in the United Arab Emirates (UAE) is generated using gas-fed thermal generation plants [1,2]. Accordingly, despite holding one of the largest hydrocarbon reserves in the world, the UAE became a net importer of natural gas in the year 2007 [3]. The UAE is planning to diversify its domestic energy mix outside fossil-based electricity generation. The plans include the deployment of nuclear and renewable energy plants [2,4,5]. Additionally, an international electricity grid connecting the Gulf Cooperation Council (GCC) countries is currently under construction. Natural gas represents approximately 81% of the overall primary energy supply in the UAE. Thus, the country heavily relies on natural gas, particularly in the electricity sector where 99% of the generation is gas-based [2]. As the consumption of electricity increases at an accelerating rate, the generation capacity of the country needs to be expanded. The expansion of the UAE’s electricity sector is fundamental to ⇑ Corresponding author. E-mail address: [email protected] (A. Almansoori). http://dx.doi.org/10.1016/j.enconman.2015.05.001 0196-8904/Ó 2015 Elsevier Ltd. All rights reserved.

ensure the country’s energy security and economic growth. This process will have to be planned well in advance in order to implement the optimal strategy over a period of time that allows securing the UAE’s electricity supply at the lowest cost and mitigating environmental damages. The use of mathematical modeling approaches is a suitable tool for planning the expansion of electric power systems. Also, they can be used to study the operation of the system, and evaluate techno-economic and environmental constraints in the network. Previously, many efforts have been made to develop mathematical models for effectively addressing the planning of electric power systems. For example, Almansoori et al. [6] developed a deterministic mixed integer linear programming (MILP) model for the optimal design of the UAE’s power system. The model was used to analyze the UAE’s power system considering different gas price levels, social benefits of air emission avoidance, and CO2 emission constraints in the year 2020. However, the model did not account for different gas/electricity import supply options, and only considered a single period in the analysis. Avetisyan et al. [7] presented a model for the optimal expansion of a developing power system. The problem’s goal was to find the optimal

178

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190

Nomenclature Indices c e f p s t t0 Sets CCS Exist Gas New Nu Oper Re

g

ðsÞ

GCC country type of air emission gas supply source power plant type scenario or realization time period (years) any previous time period (years)

{gas power plants with carbon capture and storage} {existing power plants} {gas – based power plants} {new power plants} {nuclear power plants} {operating power plants} {renewable power plants} {set of decision variables}

Continuous variables ðsÞ AEe;t amount of air emission e avoided by using alternative energy sources (tonne/h) ðsÞ APp;t total installed power capacity from plants type p in the sth scenario of period t (kW) ðsÞ CC p;t amount of CO2 avoided in the pth plant by using CCS methods (tonne/h) ðsÞ CEt annual external nuclear power plants’ costs in the sth scenario of time period t ($/yr) CF problem’s objective cost function ($/yr) ðsÞ CIEt annual cost of the imported electricity to the country ($/yr) ðsÞ CNF t annual cost of the nuclear fuel in time period t ($/yr) ðsÞ CNGt cost of the natural gas consumed in the sth scenario of period t ($/yr) ðsÞ CP p;t compression power required to transport the CO2 capture (kW) ðsÞ CSC t annual captured CO2 storage cost ($/yr) ðsÞ CTC t annual captured CO2 transport cost from gas-based plants with CCS methods ($/yr) DEp;t generation capacity loss by decommissioned units in period t (kW) ðsÞ ECAPt annualized capital cost of existing plants p in time period t ($/yr) ðsÞ EEp;t amount of energy generated by existing plants p in period t (kW) ðsÞ EGp;t total amount of electricity produced by power plants p in period t (kW) ðsÞ EM e;t emission e produced by the gas-based power fleet in time period t (tonne/h) ðsÞ EOMt annual operating and maintenance cost for plant p in time period t ($/yr) ðsÞ ET c;t electricity transferred from country c to the UAE in period t via international grid (kW) ðsÞ GSf ;t gas supply from source f in time period t (Nm3/h) ðsÞ NCAPt annual capital cost for power plants p built during period t ($/yr) ðsÞ NEp;t amount of energy generated by new power plants p in period t (kW) ðsÞ NOMt annual operating and maintenance cost for new power plants p in time period t ($/yr) ðsÞ SBt social benefit related to the air emissions avoided in the sth scenario of period t ($/yr) ðsÞ TC t compression power required in CCS (kW) ðsÞ TEt total electricity generated by the power fleet in scenario s and period t (kW)

TIt

ðsÞ

TFC i;t

ðsÞ

TNGt

total electricity imported from the GCC interconnected grid in period t (kW) total amount of fuel i consumed by the power fleet in period t (Nm3/h) (kg/h) total amount of natural gas consumed by the power sector in period t (Nm3/h)

Integer variables X p;t number of power plants p decommissioned in time period t ðsÞ yp;t number of new power plants p built during time period t ðsÞ zp;t0 ;t number of existing plants p built in a previous period t0 and available during period t Parameters AFe air emission e avoidance factor (tonne/kW h) ADe avoided damage of emission e ($/tonne) CAD CO2 emission avoided damage ($/tonne CO2) CAPFp,t power plant’s p capital factor for time period t ($/kW) CCFp carbon capture factor associated to the p plant (tonne/kW h) CDp nuclear unit p decommissioning cost ($/kW h) CFp power plant’s p capacity factor (%) CPF compression power required to transport the captured CO2 (kW h/(tonnekm)) CSp nuclear power plants’ system costs ($/kW h) CSF captured CO2 storage cost factor ($/tonne) CTF captured CO2 transport cost factor ($/tonnekm) ECFc;t electricity import cost factor from country c in time period t ($/kW h) EDt Country’s electricity demand in period t (kW) EFp;e air emission factor associated to the pth plant (tonne/kW h) HVp,i average heating value of fuel i used in the pth power plant (MJ/Nm3) (MJ/kg) HRp power plant’s p heat rate (MJ/kW h) ICp power plant’s p installed capacity (kW) IR annual real debt interest rate (%/yr) LTp lifetime which is assumed as the depreciation time of the pth power plant (yr) MGf,t maximum gas volume available for delivery from source f in time period t (Nm3/h) MIt maximum grid transferred capacity in the UAE in period t (kW) OMFp;t operating and maintenance cost factor for plant p in time period t ($/yr) OT annual operating time (h/yr) PLp pipeline length travelled by the CO2 captured at the pth plant (km) PTp,t installed power capacity target for plants type p in period t (kW) RFp annual capital recovery factor (%/yr) UCp uranium fuel cost per type of power plant p ($/kW h) ðsÞ wt weight or probability of scenario s at the time period t (%) WDp nuclear waste disposal cost ($/kW h) Xp;t bound for the new power plants p built in period t Up;t bound for the decommissioned power capacity of plant p in period t et power generation losses (%) rðsÞ price of natural gas from source f in the sth scenario of f ;t period t ($/Nm3)

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190

development plan for the power system. Nevertheless, the model fails to properly account for the stochastic nature of initial information such as forecasted fuel prices. Nikolova et al. [8] presented an approach for solving the generation scheduling problem of an electric power system consisting of conventional (thermal) and renewable energy sources. The renewable plants are integrated to reduce the system’s fuel consumption. However, the model falls short to account for the system’s total cost since the production costs used in the analysis are mainly fuel costs. Additionally, the model does not take into account the network losses. Additionally, Ji et al. [9] developed a two-stage stochastic programming model for multiperiod electric system planning. The formulation accounts for demand growth, technology development and environmental policy changes. Li et al. [10] proposed a multistage stochastic programming model for planning electric power systems and managing greenhouse gas (GHG) emissions under uncertainty over a long-time planning horizon. The aim is to determine the most suitable electricity-generation schemes and capacity expansion plans. Piao et al. [11] developed a stochastic simulation optimization model for planning electric power system under uncertainty. The model allows for uncertainties considered as interval values and probability distributions. None of those previous studies considered uncertainty in the power plants’ feedstock fuel price (e.g., natural gas) or diverse fuel supply options. Furthermore, many pathways have been previously proposed specifically for the UAE’s power sector. For example, AlFarra et al. [12] using the MESSAGE simulation model proposed different pathways for the UAE’s power sector to the year 2050. The analysis is based on the CO2 emissions and economic viability of the pathways. Sgouridis et al. [13] used an Integrated Energy Model (a bottom-up energy model in the same family as LEAP [14]) to study the impact of sustainable energy transition options based on CO2 emissions and energy system costs in the UAE to the year 2030. Mondal et al. [15] evaluated the future energy-supply strategies for the UAE power sector, the analyses were done using the MARKAL simulation model. However, those previous studies are limited to represent the UAE power sector since they are based on generic energy systems’ simulation tools. Those tools do not represent the model of a particular power system, but rather they can be used to create models of different energy systems. As a result, they lack the specifics that only can be captured in a power system model. Furthermore, those approaches cannot be considered robust optimization methods instead they are accounted as calculation tools. Moreover, although previous studies considering uncertain natural gas prices in the power sector have been done for other countries [16,17]; to the authors’ knowledge none of the works available in the literature on the UAE’s power sector considered uncertainty in the gas price. As a result, there is a lack of a comprehensive mathematical model for multiperiod planning of the UAE electric system expansion under uncertainty. Accordingly, in this work a novel MILP model is presented for the optimal planning of the UAE electricity sector for multiperiod operation under uncertainty. The multiscenario approach is considered to account for uncertainty in the natural gas price. The optimization model was developed using the General Algebraic Modeling System (GAMSÒ) modeling system. The proposed optimization model can be used to study the UAE’s power sector expansion and operation under uncertain feedstock fuel (natural gas) prices, techno-economic constraints, and environmental regulations (e.g., CO2 reductions and renewable energy targets) in the planning horizon. The paper is presented as follows. Section 2 presents the definition of the problem. Section 3 presents the formulation of the multiperiod stochastic model. Section 4 shows two case studies: the first one presents the optimal deterministic planning of the UAE’s

179

power infrastructure over the time period of 2015–2040; whereas the second case showcases the power infrastructure planning under uncertain natural gas prices. Concluding remarks are presented at the end of this work.

2. Problem definition The current section presents the main features of the proposed multiperiod stochastic design optimization model for planning the UAE electricity sector.

2.1. Problem statement Given are a number of different power plant technologies with their corresponding operating capacities and air emissions. A multiperiod picture is considered where the electricity demand is changing over each time period t, and there are key process parameters (e.g., gas price) subject to uncertainty in each analyzed scenario s of a given time period t. Also, given are the capital, fuel and operating costs for each type of power plant. The operational planning problem consists of determining the choice of operating units for each time period that minimizes cost, and that is subject to meeting the electricity demand for each time period over the entire planning horizon under uncertain gas prices.

2.2. Model structure The illustrative structure of the mathematical model is presented in Fig. 1. It includes the data input, the model constraints and objective function, and the model outputs. Three main groups of power generation technologies are considered in the present model: gas-fired [18–23], renewable [24–29], and nuclear [30– 33] power plants. The detailed list of power generation technologies and their corresponding techno-economic specifications are shown in Tables 1 and 2, respectively. In the present analysis, the alternative power plants (i.e., renewable and nuclear) are considered to be not emitting technologies. This is, their carbon life cycle assessment (LCA) is assumed to comprise only their power generation stage. There are two types of fuels available in this model: natural gas (for fossil-based plants) and natural uranium (for nuclear plants).

2.3. Model key assumptions In order to understand the programming approach and scenarios presented in the current work; the following points need to be taken into consideration: (1) Each power plant has an installed and operating capacity. Additionally, the UAE’s electricity system can receive purchased power through the GCC interconnected grid. (2) The learning curves of the power technologies are considered for the key techno-economic parameters of the units. This allows reflecting their projected improvements in the planning horizon. Also, the model considers phasing out power plants that have reached the end of their operating lifetime. (3) The model is based on the estimation of the power infrastructure required to meet the annual baseline electricity gross demand in the UAE. Nevertheless, the model also estimates the type and number of power units required to satisfy the electricity demand during the seasonal on-peak times (i.e., summer). (4) Given the extensive time period under analysis in the present work, key parameters associated to the operation of the country’s power infrastructure are subject to uncertainty. One of the most significant parameter under uncertainty is the international market price of natural gas.

180

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190

MODEL CONSTRAINTS AND

DATA INPUT ELECTRICITY

OUTPUTS

DEMAND MEET POWER DEMAND AND INSTALL

PER TIME PERIOD t. NATURAL

GAS

CAPACITY IN EACH PERIOD t.

PRICE

DISTRIBUTION.

SATISFY

URANIUM PRICE.

TARGETS IN EACH PERIOD t.

TECHNO-ECONOMIC AND ENVIRONMENTAL

DATA

OF THE PLANTS. ALTERNATIVE

MODEL

OBJECTIVE FUNCTION

ENERGY

INPUT TO THE MODEL

ALTERNATIVE

ENERGY

TOTAL POWER OPTIMIZATION OBJECTIVE:

CAPACITY

CONSTRUCTION AND DECOMMISSION MINIMIZE

OF PLANTS ARE PRESENTED. MAXIMUM

ELECTRICITY

IMPORT

CAPACTIES BY PERIOD t.

TARGETS.

MAXIMUM

NATURAL

INITIAL POWER FLEET AT

FROM

TIME t=0.

DEPENDING ON TIME PERIOD t.

NUMBER OF NEW POWER

THE POWER GENERATION COSTS,

PLANTS AVAILABLE PER

INCLUDING

PERIOD t.

AND MAINTENANCE, FUELS, CARBON

GCC GRID CAPACITY.

CAPTURE

CCS COSTS.

EXTERNAL ARE ACCOUNTED IN THE

DIFFERENT

CAPITAL, AND

GAS

SUPPLY

SOURCES

THE

ELECTRICITY UNDER

PLANTS

UNCERTAINTY

CONSTRUCTION

IN THE STUDIED

AND

PLANNING

DECOMMISSION.

HORIZON.

CONSUMPTION OF POWER FUELS.

OPERATING

STORAGE,

COMPOSITION. DECISIONS ON

CUMULATIVE COSTS

GENERATION AND

AIR EMISSIONS. COSTS AND

AND

COMPOSITION.

POWER MODEL. Fig. 1. Illustrative structure of the design optimization model.

Table 1 List of power plant technologies. Type of power plant Natural gas Natural gas combined cycle (NGCC)- class 7FA (PG1) Natural gas combined cycle (NGCC)- class 7FB (PG2) Natural gas combined cycle (NGCC)- class 7FB- with 90% CO2 capture using MEA (PG3) Natural gas oxyfuel with CO2 capture (PG4) Steam turbine (ST) (PG5) Gas turbine (GT) (PG6) Wind Nordex N43/600 (PW1) Nordtank 500/41 (PW2) Sonkyo 3.5 kW (PW3) Gaia–Wind 133–11 kW (PW4) Solar Sanyo single crystalline silicon solar cells (PS1) Mono-silicon BP solar 90 W modules (PS2) Concentrating solar power (CSP) (PS3) Ocean thermal energy conversion (OTEC) (PS4) Solar land pond (SLP) (PS5) Hybrid of ocean thermal energy conversion with an offshore solar pond (OTEC–OSP) (PS6) Nuclear APR-1400 (PN1) AP-1000 (PN2) EPR-1650 (PN3)

Source [15,18] [15,19,20] [15,19,20] [15,21] [15,22] [15,23] [15,24] [15,25] [15,25] [15,26] [15,27] [15,28] [15,29] [15,29] [15,29] [15,29]

Projections of the natural gas requirements and current techno-economic limitations (to further develop domestic gas resources) raise concerns on the availability of this primary fuel to meet the country’s demands; which suggests an increased reliance on gas imports [34]. Under this uncertain scenario, the natural gas price represents a key economic parameter for planning the future power infrastructure of the country. The price could potentially determine the natural gas supply sources (which will have a direct impact on the country’s energy security), the path to follow regarding renewable and nuclear energy developments, power imports from neighboring countries (i.e., GCC interconnected grid), GHG emissions, etc. Accordingly, the price of the natural gas use as feedstock fuel for power plants has been considered the key parameter under uncertainty in the present work. The use of stochastic-based optimization models can be a practical tool in the decision making process of the UAE’s power sector. Given the overall size of the proposed problem including a significant number of integer variables, the present work adopted the multiscenario approach to make the problem tractable under uncertainty. Note that this approach involves solving the process model at each discrete realization s considered for the uncertain ðsÞ

parameter (e.g., gas price bf ;t ) weighted with the corresponding [15,30,31] [15,32] [15,30,33]

2.4. Uncertain parameters Despite the country’s efforts to introduce alternative power technologies to increase the domestic energy mix; natural gas is currently in practical terms the single primary energy source used in the UAE’s power sector. Moreover, natural gas is expected to remain as the main feedstock fuel used in the power sector in the foreseeable future. Accordingly, the natural gas represents an essential factor when planning the future power infrastructure of the country.

ðsÞ wt .

Accordingly, the model’s objective funcdiscrete probability tion is denoted by a weighted cost function that reflects the costs over the expected range of operation. The use of the multiscenario approach considered in the present work has been extensively used to address the optimal operation of systems under uncertainty [35–43]. However, the implementation of such approach to address the optimal planning of the UAE’s power system operations under uncertainty remains unexplored in the literature. In this work, the fluctuation in the natural gas was assumed to follow a normal probability distribution. This type of distribution is the most widely known and the standard for many probability problems [44,45]. The associated uncertainty is considered in the present model using a discrete distribution of the random parameter (i.e., natural gas price) with a finite number ‘‘S’’ of possible

181

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190 Table 2 Power plants’ key techno-economic parameters for 2015–2040. Type of power plant

Installed capacity (kW)

Capital cost 2015–2040 ($/kW)

Operating & maintenance cost factor 2015–2040

Heat rate

Natural gas PG1 PG2 PG3 PG4 PG5 PG6

580,000 507,000 432,000 440,000 60,000 85,000

1250–1080 595–514 978–825 1308–1103 681–572 973–817

0.0045–0.0039 $/kW h 1.80–1.55%b 3.70–1.52%b 8.6–7.25%b 0.005–0.0042 $/kW h 0.0163–0.0137 $/kW h

7.07 MJ/kW h 7.17 MJ/kW h 8.41 MJ/kW h 7.70 MJ/kW h 8.37 MJ/kW h 11.45 MJ/kW h

Wind PW1 PW2 PW3 PW4

600 (25)a 500 (30)a 3.5 (100)a 11 (45)a

1620–1242 547–419 6084–4668 12,955–9940

0.011–0.0084 $/kW h 0.039–0.030 $/kW h 0.047–0.036 $/kW h 0.238–0.183 $/kW h

N/A N/A N/A N/A

Solar PS1 PS2 PS3 PS4 PS5 PS6

10,000 10,000 16,700 50,000 50,000 50,000

17,060–9030 9200–4878 14,228–11,026 13,500–10,463 7938–6152 2970–2302

6–3.2%b 0.8–0.42%b 5,400,000–4,185,000 $/yr 9,450,000–7,324,000 $/yr 6,750,000–5,230,000 $/yr 4,050,000–3,140,000 $/yr

N/A N/A N/A N/A N/A N/A

Nuclear PN1 PN2 PN3

1,400,000 1,100,000 1,650,000

3643–2807 3582–2760 4100–3160

0.002–0.0015 $/kW h 0.0054–0.0042 $/kW h 0.002–0.0015 $/kW h

2.77E6 kg/kW h 7.48E6 kg/kW h 2.77E6 kg/kW h

N/A Power plants that do not consider Heat Rate related to fuel consumption. a Number enclosed by parenthesis indicates the total array number of turbines in the wind farm. b Operating and Maintenance cost (including repair and replacement cost for PV) given as a percentage (%) of the plant’s total capital cost.

realizations (scenarios). The discrete probabilistic scenarios were obtained using a MatlabÒ pseudo-random number generator.

The total number of new power plants type p that can be built by time period is constrained in the model as follows:

3. Model formulation This section presents the mass and energy balances as well as constraints used to mathematically represent the multiperiod operation of a power system under fuel price uncertainty. 3.1. Model inputs In this section, the key inputs are presented as part of the model’s algebraic equations. Accordingly, the total electricity demand in the country can be formulated as follows: ðsÞ ðsÞ TEðsÞ t ð1  et Þ þ TI t  TC t P EDt ;

8t

ð1Þ

where the subindex t indicates the time period and s represents a given scenario, TEtðsÞ represents the total electricity generated by the power infrastructure in a scenario s of a given period t, et represents the power generation losses in the system by time period t, TItðsÞ is the total electricity imported to the UAE from the GCC interconnected grid (see Eq. (A.2) in Appendix A for details), TC ðsÞ t is the compression power required in carbon capture and storage (CCS) processes (see Eq. (A.8) in Appendix A for details), and EDt is the forecasted country’s electricity demand per time period t. The amount of energy generated in a particular time period includes the supply from existing power units and new built plants. The amount of alternative power required to be installed at a given time period according to governmental targets [46,47] or environmental regulations is given in the model as follows: ðsÞ APp;t

P PTp;t ;

8t; 8p 2 Re;Nu

3.2. Model constraints

ð2Þ

ðsÞ where APp;t is the installed power capacity of plants type p in the sth scenario of time period t, and PTp,t represents the aimed target from p power plants for the time period.

U XLp;t 6 yðsÞ p;t 6 Xp;t ;

8t; 8p 2 New

ð3Þ

ðsÞ

where yp;t is an integer variable representing the number of new power plants type p that can be built in the sth scenario of time period t, whereas XLp;t and XUp;t are valid lower and upper bounds of the variable, respectively. Similarly, the type and number of existing power plants in a time period is constrained as follows: ðsÞ

zp;t0 ;t P 0;

8t; 8p 2 Exist

ð4Þ

ðsÞ

where zp;t0 ;t is an integer variable denoting the type and number of existing power plants p built in a previous period t0 and available during the time period t. Additionally, the model considers the decommissioning of old power plants once their operating lifetime has been reached (see Eq. (A.1) for specifics). The type and number of power plants decommissioned by time period is limited in the model as follows:

ULp;t 6 xp;t 6 UUp;t ;

8t; 8p 2 Exist

ð5Þ

where xp;t is an integer variable denoting the number of power plants p decommissioned during time period t, ULp;t and UUp;t are the lower and upper bounds of the variable, respectively. The total amount of electricity that can be transferred from the neighboring GCC countries through the regional interconnected grid is limited as follows:

TIðsÞ t 6 MIt ;

8t

ð6Þ

where MIt denotes the maximum grid transferred capacity in the UAE for period t. The natural gas supply constraints from different sources f (e.g., domestic supply, pipeline imports, and LNG cargoes) can be defined as follows:

182

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190

ðsÞ

GSf ;t 6 MGf ;t ;

8f ; t

ð7Þ

X

NCAPðsÞ t ¼

ðsÞ

yp;t ICp CAPFp;t RFp ;

8t

ð13Þ

p2New ðsÞ

where GSf ;t is the gas supply from source f (see Eq. (A.3) in Appendix A for details), and MGf,t represents the maximum volume of gas available for delivery from source f in time period t.

where NCAPtðsÞ is the total annual capital cost for new power plants built in period t. Moreover, the operating and maintenance cost of new power plants can be calculated as follows:

3.3. Electricity generation

NOMðsÞ t ¼

X

ðsÞ

yp;t OMFp;t ;

8t

ð14Þ

p2New

The electricity in the system is generated by existing and new constructed power plants in each time period. The power generated by the country’s existing power plants in a time period t can be formulated as follows:

EEðsÞ p;t ¼

X

ðsÞ

zp;t0 ;t ICp CFp ;

8t; 8p 2 Oper

ð8Þ

t0

where NOMtðsÞ is the total annual operating and maintenance cost of new power plants in time period t. The fuel cost is dominated by the natural gas since it constitutes the principal feedstock for the power system. The natural gas cost can be calculated as follows:

CNGðsÞ t ¼

ðsÞ where EEp;t represents the total electricity generated by the p plants of the existing power fleet in scenario s of time period t. It accounts for losses of power generation capacity due to units decommissioning (see Eq. (A.1) in Appendix A for specifics). ICp represents the plant’s installed capacity, and CFp denotes the plant’s capacity factor. Similarly, the power generated by the new operational power units in period t can be defined as follows:

ðsÞ

ðsÞ NEp;t ¼ yp;t ICp CFp ;

8t; 8p 2 Oper

ð9Þ

ðsÞ NEp;t

where represents the power generated by the new operational power plants type p in period t. Accordingly, the amount of power generated by power technology and time period can be defined as follows:

X ðsÞ ðsÞ ðsÞ wt bf ;t GSf ;t OT; 8t

ð15Þ

f

where CNGtðsÞ is the cost of the natural gas consumed in the sth sceðsÞ

nario of period t, wt is the weight or probability of occurrence of ðsÞ

scenario s at time period t, bf ;t represents the price of natural gas from source f in the sth scenario of period t, and OT is the annual operating time. On the other hand, the cost of the nuclear fuel (i.e., enriched uranium) consumed by the power reactors can be calculated as follows:

CNF ðsÞ t ¼

 X X  ðsÞ ðsÞ zp;t0 ;t þ yp;t ICp CFp UCp OT; 8t

ð16Þ

t0 p2Nu

ðsÞ where EGp;t is the total electricity generated by plants type p in the sth scenario of period t. This quantity is related to the fuel consumption by Eq. (A.4) in Appendix A.

where CNF tðsÞ is the annual cost of the nuclear fuel, and UCp is the uranium fuel cost per type of nuclear power reactor p. Additionally, nuclear power includes some associated external costs that must be taken into consideration in the total cost of the power infrastructure such as: nuclear waste disposal cost (WDp), nuclear reactors decommissioning costs (CDp), and the plants’ system costs (CSp). Accordingly, the total external cost associated to the power reactors is given as:

3.4. Electricity system costs

CEðsÞ t ¼

ðsÞ ðsÞ EGp;t ¼ EEðsÞ p;t þ NEp;t ;

8t; 8p 2 Gas; Re;Nu

ð10Þ

 X X  ðsÞ ðsÞ zp;t0 ;t þ yp;t ICp CFp ðWDp þ CDp þ CSp ÞOT; 8t

ð17Þ

t0 p2Nu

The electricity system costs consist of the plants’ capital, operating and maintenance, fuel, and associated external costs. Accordingly, the existing plant’s fleet capital cost can be calculated as:

ECAP ðsÞ t

¼

XX

ðsÞ zp;t0 ;t ICp CAPFp;t0 RFp ;

8t

ð11Þ

t 0 p2Exist

where represents the annualized capital cost of the existing plant’s fleet in time period t, CAPFp,t0 denotes the plant’s p capital cost factor in period t0 , and RFp is the annual capital recovery factor (see Eq. (A.10) in Appendix A for specifics). That is, the formulation considers costs learning curves; thus, the capital cost for building power plants decreases as the number of periods increase. Accordingly, the annualized capital cost of an existing power plant p depends on the initial capital investment during period t0 . The power plants’ capital costs are considered to be amortized until the end of their operating lifetimes. The operating and maintenance (O&M) costs of the existing power plants can be calculated as:

¼

XX

ðsÞ zp;t0 ;t OMFp;t ;

8t

ð12Þ

t 0 p2Exist

EOMtðsÞ

CTC tðsÞ ¼

X

CC ðsÞ p;t ðCTFÞPLp ðOTÞ;

8t

ð18Þ

p2CCS

where CTC tðsÞ is the captured CO2 transport cost in the sth scenario of

ECAPtðsÞ

EOM ðsÞ t

where CEtðsÞ is the annual external nuclear power plants’ cost in the sth scenario of time period t. The carbon capture transport cost can be calculated as follows:

where is the annualized operating and maintenance cost of the existing plant’s fleet in period t, and OMFp,t is the operating and maintenance cost factor for plant p during the tth time period. On the other hand, the capital cost for the new plants built in period t can be estimated as:

ðsÞ represents the amount of CO2 avoided in the pth time period t, CC p;t plant by using CCS methods in period t (see Eq. (A.7) in Appendix A for details), and CTF denotes the carbon transport cost factor. Furthermore, the CO2 storage cost is given as:

CSC ðsÞ t ¼

X

CC ðsÞ 8t p;t ðCSFÞOT;

ð19Þ

p2CCS

where CSC tðsÞ is the annual captured CO2 storage cost and CSF is the carbon storage cost factor. The import electricity cost from the interconnected GCC regional grid to the country is given as:

CIEtðsÞ ¼

X ECFc;t ET ðsÞ 8t c;t OT;

ð20Þ

c

where CIEtðsÞ is the annual cost of the imported electricity to the country, ECFc;t represents the electricity cost factor from the GCC ðsÞ is the electricity transcountry c to the UAE in period t, and ET c;t ferred from country c to the UAE in period t (see Eq. (A.2) in Appendix A for details). All the costs listed in this work are given in US $ (2010).

183

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190

3.5. Objective cost function The conceptual formulation and objective cost function of the optimization model are given as:

minCF ¼ g

the Abu Dhabi’s Emirate power system over the time horizon of 2015–2040, considering a time span of five years between studied periods. The first case study considers a multiperiod deterministic optimization approach, whereas the second accounts for gas price

T X S   X ðsÞ ðsÞ ðsÞ ðsÞ ðsÞ ðsÞ ðsÞ ðsÞ ðsÞ ECAP ðsÞ t þ EOM t þ NCAP t þ NOM t þ CNGt þ CNF t þ CEt þ CTC t þ CSC t þ CIEt t

s

Subject to Power Demand per time period t

ð1Þ

Alternative Energy Targets per time period t Power Plant Technologies ðnatural gas; wind; solar and nuclearÞ

ð2Þ ð3Þ  ð4Þ

Generation Capacity Decommissioned per time period t

ð5Þ

Import Power Capacity from neighboring countries per time period t

ð6Þ

Natural Gas Supply Sources and Availability per time period t

ð7Þ

Feedstock fuel prices per scenario s and time period t

ð15Þ  ð16Þ

where the problem’s set of decision variables g includes: the types of power plants (p), number of power plants and their corre  ðsÞ ðsÞ sponding operating capacities zp;t0 ;t ; yp;t ; xp;t , natural gas supply ðsÞ

sources ðGSf ;t Þ, and the amount and source of imported power ðET ðsÞ c;t Þ. The objective is to minimize the cumulative annualized electricity generation costs associated with the UAE’s power system operation over the planning horizon. Accordingly, the aim is to determine the optimal infrastructure of power plants and the amount/source of imported electricity/natural gas for each time period under operational constraints. In this work, uncertainty is considered in the natural gas price using a discrete normal distribution of the random parameter with a finite number of possible ðsÞ

scenarios s and corresponding probability or weight wt . The resulting power system framework (1)–(21) and (A.1)–(A.10) (see details for the latter equations in Appendix A) is a multiperiod stochastic optimization model. The optimization framework was developed in GAMS [48] as a MILP model. The mathematical model consists of 1993 equations, 1891 continuous variables, 223 discrete variables, and 774 parameters. The MILP problem was solved using CPLEX [49] as solver. CPLEX uses a branch and cut algorithm that solves a series of LP, subproblems. The algorithm can be summarized as follows: (1) Pre-processing identifies infeasibility and redundancy, improve bounds and rounding. (2) The linear program relaxation problem is solved. (3) Add cuts to reduce the feasible region based on information from the relaxation solution. (4) Apply heuristics that may provide good solutions quickly and help to prove optimality. (5) Create two new subproblems by choosing integer variables. (6) Select a subproblem and solve. (7) Repeat steps 1–6 until termination criteria is met (tolerance). Because a single mixed integer problem generates many subproblems, even small mixed integer problems can be very compute intensive. Additionally, given the multiperiod and stochastic nature of the present optimization model, the multiscenario approach was used to solve the proposed problem. The present optimization model can be used as a practical tool for planning the UAE’s power sector in a given time horizon. 4. Case study 2015–2040 The modeling and optimization framework presented in the previous section has been applied in two case studies for planning

ð21Þ

uncertainty in the multiperiod formulation. This allows comparing the influence of uncertainty in key economic parameters for planning the Emirates power infrastructure throughout the time horizon. Key data used in the case studies, results and analysis, and some policy suggestions are presented in this section. The following factors are required for analyzing the current case studies: (1) The initial or existing (t = 0) power infrastructure of Abu Dhabi was determined through a detailed survey of the Emirate’s power generation capacity as of the end of 2013 [1]. (2) The year 2015 is considered to be the first studied period, and a 25 year span is considered for evaluation. The 25-year span is divided among 6 periods that are separated 5 years from each other (i.e., 2015, 2020, 2025, 2030, 2035, and 2040). (3) The projected power demand data from 2015 to 2040 is obtained from the Abu Dhabi Water and Electricity Company (ADWEC) [50] for each time period. (4) Upper bounds for the installed capacity of each type of renewable and nuclear power plant (XUp;t ) [2,5,12,51], as well as annual electricity and natural gas supply capacities from different sources ðsÞ

(ET ðsÞ c;t and GSf ;t ) are considered in the model [52–54]. The timeframe 2015–2040 was selected given that Abu Dhabi (as part of its socio-economic development vision) has set specific goals for the deployment of renewable and nuclear power capacities [4,5,46,47] for the given time periods. The aim is to diversify the Emirate’s power mix and increase the country’s energy security in the future. Also, the projected construction costs of new power plants as well as the operating and maintenance costs of the units are available for the studied timeframe [15,23,30]. Furthermore, data regarding the expected gas supply and electricity import sources, maximum generation capacities, and projected fuels’ unit prices are available in the literature for the studied periods [2,34,54–59]. The key techno-economic parameters used in the present case studies are shown in Table 3. 4.1. Case study 1: Multiperiod deterministic planning of the UEA’s power infrastructure The present case study shows the optimal planning of Abu Dhabi’s power infrastructure throughout the years 2015–2040 following a multiperiod deterministic approach. Accordingly, the number of scenarios under analysis was considered to be s = 1; which corresponds to a single realization. Accordingly, natural gas average prices were considered as model’s inputs to solve the optimization problem [15,34,55,58,60–63]. According to

184

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190

Table 3 Key techno-economic inputs and assumptions for case study 2015–2040. Input type

Time periods 2020

2025

2030

2035

2040

Natural gas prices ($/MM Btu) [15,34,55,58,60–63] Domestic 3.20 Contracted Qatari 1.50 Non-contract Qatari 7.0 LNG from Fujairah 14.0 LNG from future sources N/A Power generation losses (%) 9.23

2015

3.50 2.50 9.0 15.0 N/A 8.62

3.90 3.50 10.0 16.0 N/A 8.02

4.30 4.50 11.0 17.0 17.0 7.42

4.80 12.0 12.0 18.0 18.0 6.81

5.30 13.0 13.0 19.0 19.0 6.21

Gas supply constraints for power sector (MM scf/d) [52–55,63] Domestic 390 Contracted Qatari 593 Non-contract Qatari 1334 LNG from Fujairah N/A LNG from future sources N/A

407 593 1334 1200 N/A

407 593 1334 1200 N/A

441 593 1334 1200 2000

771 N/A 1927 1200 2500

771 N/A 1927 1200 3000

Capacity constraints (MW) Installed renewable targets [2] GCC grid transfer capacity [56,57]

1200 900

500 1200

500 1500

500 1800

500 2100

450/NS 0/0 100/200 2/3

0/NS 0/200 0/100 0/1

0/NS 0/300 0/100 0/1

0/NS 0/300 0/100 0/1

0/NS 0/300 0/100 0/1

300 900 Minimum/maximum

Alternative power install capacity constraints (MW) [2,5,12,51] PV power plants (PS1–PS2) 50/NS CSP power plants (PS3) 100/200 Wind power plants (PW1–PW4) 30/60 Nuclear power plants (N° of units) 0/0 NS stands for not specified value.

Fig. 2. Abu Dhabi’s annual net power generation capacity by technology in the timeframe 2015–2040.

information reported in the literature, the current or initial (t = 0) generation capacity of the power system was assumed to be gas-based: natural gas combined cycle power plants (83%), steam turbines (11%), and gas turbines (5%) [1,47]. Also, minor contributions of renewable energy plants corresponding to wind, CSP and PV solar units (<1%) were considered for the remaining capacity. The optimization problem converged for the present case study in less than 1 CPU s on an Intel(R) Core(TM) i5M-560 with 2.67 GHz CPU processor and 4.00 GB of RAM memory machine. The solution was found after 125 iterations of the branch and cut algorithm. For the first period corresponding to the year 2015, the energy picture is very similar to the initial power fleet in terms of generation share; since no major changes in the power infrastructure are expected to take place in this time period (see Figs. 2 and 3). The

total installed power capacity reaches 17.9 GW whereas the annual gross electricity generation of the power fleet is 96,850 GW h. The gross electricity generation refers to the amount of power before losses. Additionally, although the model estimates the overall installed power capacity required by time period (including the spare capacity required to meet the peak electricity demand forecasts); the annual power generation is calculated to meet the annual baseline or average electricity demand of the Emirate per time period. The total renewable installed capacity reaches 300 MW with an annual net generation capacity of 132 MW (i.e., annual capacity factor is 44%). Regarding the annual costs, gas-based power represents around 93% of the total cost; whereas renewables rank second with 6.5%, and the remaining share (0.5%) corresponds to electricity imports from Qatar (see Table 4). Moreover, as shown

185

Power GCC interconnected grid Gas Power Producon

Nuclear Power Renewable Power Producon

Total GHG (CO2 eq.) Emissions SO2 Emissions Avoided CO2 Emissions

300 250 200 150 100 50 0 2015

NOX Emissions PM Emissions

70

Annual Air Emissions (MT/yr)

Annual Gross Electricity Generaon (GWh/yr)

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190

2020

2025

2030

2035

60 50 40 30 20 10 0 2015

2040

2020

2025

2030

2035

2040

Periods (yr)

Period (yr) Fig. 3. Abu Dhabi’s gross electricity generation share in the planning horizon 2015– 2040.

in Fig. 4 the annual GHG emissions of the power sector in the year 2015 account for 35 MT of CO2 eq.; whereas the avoided GHG emissions (due to the use of renewable energies) is 0.43 MT CO2 eq./yr. As shown in Table 4, gas-based power represents the lowest generation cost option for this year given the low price of the majority of the gas supply (i.e., domestic and contracted Qatari volumes). On the other hand, as shown in Figs. 2 and 3 the energy mix increases from the year 2020 onward. This is the result of Abu

Fig. 4. Abu Dhabi’s power sector air emissions for the timeframe 2015–2040.

Dhabi’s energy diversification engagement. Accordingly, for the second period (year 2020) gas generated power considerably decreases to 67% of the total supply share (see Fig. 3). This reflects the direct effect of introducing nuclear power in the Emirate’s electricity grid for this time period. Accordingly, nuclear power accounts for approximately 30% of the gross power generation. On the other hand, renewable power contributes with 3.1% given its low annual capacity factor. Nevertheless, the renewable energy share almost tripled due to the Emirate’s target. Furthermore, the total installed capacity reaches 23.9 GW and the annual gross

Table 4 Case study 1: Optimization results for the multiperiod deterministic planning of UEA’s power infrastructure.

a

Variable

Time periods (year)

Net power generation capacity by plant type (MW)

2015

2020

2025

2030

2035

2040

2040a (on-peak)

PG1 PG2 PG5 PG6 Total gas-based PS2 PS3 PS5 PW2 PW3 Total renewables PN3 Total net power capacity

3132 5932 1200 612 10,876 39 86 0 7 0 132 0 11,008

3132 6160 672 0 9964 305 86 43 31 0.6 465.6 4554 14,984

3132 8898 0 0 12,030 404 86 86 42 1.7 619.7 6072 18,722

2610 12,320 0 0 14,930 504 86 128 53 1.7 772.7 7590 23,293

1827 15,060 0 0 16,887 605 86 171 64 1.7 927.7 9108 26,923

261 18,708 0 0 18,969 706 86 214 74 1.7 1081.7 10,626 30,677

261 33,994 0 0 34,255 706 86 214 74 1.7 1081.7 10,626 45,963

Total annual power cost (MM $/yr) Gas-based Nuclear Renewables GCC interconnected grid imports

4790 0 335 37

4850 2570 1230 8

6790 3400 1570 62

9560 4200 1880 90

12,500 4980 2150 36

15,800 5740 2400 200

N/A N/A N/A N/A

Levelized electricity cost ($/kW h) Gas-based Nuclear Renewables GCC interconnected grid imports Total levelized cost

0.050 N/A 0.289 0.088 0.05

0.056 0.064 0.303 0.097 0.07

0.064 0.064 0.290 0.107 0.07

0.073 0.063 0.277 0.118 0.08

0.084 0.062 0.265 0.131 0.08

0.094 0.062 0.253 0.144 0.09

N/A N/A N/A N/A N/A

Natural gas consumption (MM scf/d) Domestic gas Contracted Qatari gas Non-contract Qatari gas LNG from Fujairah LNG from future sources Total gas consumption Natural gas cost (MM $/year) Social damages avoided (MM $/year) Cumulative decommissioned generation capacity (MW)

390 593 837 0 0 1820 2980 30.6 0

407 593 602 0 0 1602 3103 1161 1140

407 593 914 0 0 1914 4771 1547 1812

441 593 1334 10 0 2378 7233 1934 3019

771 N/A 1920 0 0 2692 9967 2321 5399

771 N/A 1927 329 0 3028 13,190 2708 9931

771 N/A 1927 1200 1570 5468 30,470 2708 9931

This scenario corresponds to the worst case scenario or the highest expected electricity demand peak for the summer of 2040.

186

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190

generation output approximates to 131,000 GW h (see Fig. 3). Over 1.1 GW of gas-based installed capacity is decommissioned during this time period. This capacity is replaced by nuclear power; as a result the average gas consumption decreases to 1600 MM scf/d; which is consistent with the government plans to increase energy security by curbing gas demand. The power costs are dominated by gas-based generation (56%) and nuclear power (30%); whereas the remaining corresponds to renewable power (see Table 4). Furthermore, as shown in Fig. 4, the annual GHG emissions are reduced to 32 MT of CO2 eq. (despite the increase in the annual power generation). On the other hand, the amount of avoided emissions grows to 16 MT of CO2 eq./yr; which represents a very significant increase compared with the previous period. These outcomes are the result of introducing cleaner energy sources in the power infrastructure. In the year 2025, the gross gas-based generation slightly decreases to 64% whereas that of nuclear increases to 32%. On the other hand, the renewable and import power shares practically keep unchanged. This means that the generation share loss by gas plants is gained by the nuclear power reactors given the latter slightly lower generation costs for this period (see Table 4). Accordingly, from this period onward the cost of nuclear power remains below that of gas-based generation since the average cost of the gas consumed in the power sector increases with time. Also, the generation shares up to the year 2040 remain practically unchanged given that the proportions between types of new installed power capacities (e.g., gas, nuclear, renewable) keep nearly constant. The stabilization in the generation shares after the year 2020 has to do with the geographical constraints for the location of nuclear reactors in the emirate. Thus, a maximum of one new installed unit is considered for the time periods subsequent to 2020; which does not allow adding enough generation capacity in the nuclear fleet to keep increasing its share in the total electricity supply. Moreover, the total installed capacity increases to 30 GW and the annual gross generation raises to approximately 165,000 GW h for this period. The renewable power installed capacity increases by 500 MW, which corresponds to the lower bound set for the deployment of renewable energy in this period. A similar trend is observed for the rest of the periods comprised in the planning horizon. The natural gas consumption grows to 1,910 MM scf/d; which is met by increasing the gas volume imported from the Dolphin pipeline’s uncontract capacity. For the year 2030, the installed generation capacity reaches 37.5 GW whereas the annual electricity generation grows to 205,000 GW h. The available domestic gas increases due to the development of new projects. As a result, the domestic gas volume consumed by the Emirate’s power sector increases to a new maximum level (see Tables 3 and 4). This is exacerbated by expected underpriced domestic gas. Furthermore, Dolphin’s uncontract gas supply reaches its upper limit due to its relatively lower prices compared with LNG cargoes. Additionally, LNG gas consumption takes place for the first time in the power sector, but at a small scale (0.4% of the total supply). Consequently, the total gas demand grows to 2,380 MM scf/d; which represents a 24.6% increase compared with the previous period. The GHG emissions grow to 48.2 MT of CO2 eq./year due to increases in gas-based generated power. Nonetheless, both the CO2 emissions and social damages avoided increase (see Table 4) due to the increment of alternative power. Their associated benefits are given in terms of avoided public health costs that would otherwise result from air emissions to the atmosphere. Also, there are social benefits associated to keeping the environment unaffected (see Eq. (A.9) in Appendix A for specifics). For the year 2035, the total installed power capacity reaches 43.4 GW and the annual power generation increases to 236,000 GW h/yr. The decommissioned generation capacity for

this period sharply increases compared to the previous time periods (see Table 4). This is, a significant number of gas plants from the initial power fleet reach the end of their operating lifetime, and they are replaced by nuclear plants and more advanced and efficient gas-based technologies. The amount of domestic gas available and consumed by the Emirate’s power sector increases by 75% due to new gas projects expected to be operational by this time period; which boosts the domestic gas supply. On the other hand, the piped gas imports are expected to remain almost unchanged; despite the fact, that the contract of Qatari gas supply will be expired by this time period. This means that the Emirate may have to pay international price levels for all the piped gas imported from Qatar in this period. As a result, the overall gas supply increases by 13% compared with the previous period (see Table 4) to meet the demand in 2035. The increasing trend in the cost of gas-based generated power continues as a consequence of sustain increases in the average price of the gas consumed by the power sector. For the last studied period (year 2040), the total installed capacity grows to 50 GW whereas the annual generation reaches 270,000 GW h/yr. The number of power plants that are decommissioned dramatically increases (see Table 4); while their generation capacities are replaced by newer gas-based technologies and nuclear power. The average gas volumes consumed from domestic and pipeline imports remain at their maximum allowable levels, whereas the demand for LNG gas alone experiences a sudden increase of approximately 330 MM scf/d. Gas generation increases to 65.5% of the total cost; whereas nuclear power decreases to 23.7%. Nuclear power remains as the cheapest electricity generation technology whereas gas ranks second, followed by imported power, and renewable sources (see Table 4). Additionally, we took a closer look at the worst case scenario or highest peak electricity demand expected, specifically the year 2040. During peak periods in this year the used power generation capacity can reach up to 46.8 GW, which represents an increase of 52% compared with the annual average. The natural gas consumption can grow by 80%; which would require supply of gas from other external sources such as more LNG cargoes or piped gas from neighboring countries. Other option is for more domestic gas resources to be available during peak electricity periods via gas storage. Additionally, the amount of GHG emissions can increase up to 12,633 t/h. This corresponds to an increase of over 80% compared with the annual average rate for 2040. Most of the spare installed generation capacity is composed of gas-based power plants due to their operational flexibility. 4.2. Case study 2: Uncertainty in the natural gas price The present case study illustrates the optimal planning of Abu Dhabi’s power infrastructure following a multiperiod stochastic approach. As the UAE’s domestic gas requirements increased, the Dolphin pipeline uncontract capacity may play and increasingly important role for two key reasons: (i) It is already connected to some of the largest gas-based power facilities in the country (and could be easily expanded to others). (ii) It is a reliable and safe gas transport method, and cheaper compared with LNG cargoes. However, pricing the uncontract volume of the pipeline is rather difficult. It is subject to several uncertain factors that may affect its future pricing. These factors are as follows [34]: (1) It will depend on Qatar’s government perception of a fair price for its gas. (2) Qatar’s fears of subsidizing its neighbors through cheap gas. (3) Development of a regional gas market to palliate shortages in most GCC countries. As for the gas price for the remaining supply sources, although they will always attain a certain degree of uncertainty, their degree of uncertainty is considered to be significantly less. For example, the domestic gas supply is heavily subsidized by the state, and it

187

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190

ðsÞ

The natural gas price (bf ;t ) was assumed to follow a normal probability distribution with nominal (mean) value (l) of 11.0 $/MM BTU and variance (r2) (3.83)2 $/MM BTU for 2030. The mean gas price and variance were derived from information reported in the literature [62]. Accordingly, using the mean gas price value and variance 200 scenarios were generated for the uncertain parameter (i.e., S = 200 in problem (21)), each scenario corresponds to a particular realization of the gas price. The probability or weight ðsÞ

(wt ) assigned to each scenario s was estimated by evaluating the normal probability distribution function for each realization. In addition, a preprocessing analysis was performed and determined that 200 realizations provide a suitable representation of the uncertain parameter’s space, i.e., further increases do not improve the solution noticeable, but it does increase the computational time. The CPU time required by GAMS/CPLEX to solve this problem was approximately 70 CPU s. For the present multiscenario approach each of the studied scenarios converged after 86–93 iterations of the CPLEX algorithm. Additionally, the authors’ have tried to validate the proposed strategy by comparing with similar MILP solvers (i.e., GAMS/BDMLP, GAMS/CoinCbc, and GAMS/CoinGlpk). GAMS/BDMLP branch and bound algorithm is intended for small to medium sized models, it is not too powerful compared with other commercial MIP codes in GAMS (e.g., Cplex). On the other hand, GAMS/CoinCbc and GAMS/CoinGlpk are based on the branch and cut algorithm. They are not very fast in their generic form, but are flexible and can be effective with correct use of cut generators and heuristics. For instance, cut generators can be switched off, on every so often or only at root node. However, these solvers have not led to feasible solutions given that the problem is highly constrained with very scattered feasible regions. This is, the studied scenarios are very restricted by the initial power infrastructure of the Emirate, and the alternative energy targets (minimum and maximum capacities) set for the planning horizon. Although limited at some degree, the previous solvers represent open-source mixed integer programming solvers accessible to every user. Further details on optimization schemes can be found in the literature [64–68]. Fig. 5 shows the variability in the cumulative electricity cost for selected periods in the planning horizon. As shown in the figure, the cumulative cost for the studied periods fluctuate between 80 and 88 billion $. Moreover, the highest probability of occurrence (0.195) or ‘‘peak’’ is approximately located in the cost range of 85 billion $. Correspondingly, the expected value for the cumulative costs was found to be around 85.011 billion $. This characteristic could be expected since the previous value corresponds to the most likely of the events. Additionally, the expected value is equivalent to the cumulative electricity cost obtained for the

PROBABILITY DISTRIBUTION

0.25

0.2

0.15

0.1

0.05

0 80

81

82

83

84

85

86

87

88

CUMULATIVE TOTAL ELECTRICITY COSTS (BILLION $) Fig. 5. Distribution of the cumulative electricity costs for the analyzed time periods between 2015 and 2040.

multiperiod deterministic case (i.e., 85.098 billion $) presented in the previous section. However, as shown in Fig. 5, the cumulative electricity cost has a non-Gaussian (nonsymmetrical) distribution, which suggests that for the present case study there exist a nonlinear correlation between the natural gas price and the cumulative electricity cost in the Emirate’s power sector. This result can be explained by the fact that the present case study considers an initial (t = 0) power infrastructure. This is, there exists a base power infrastructure before the first time period of the planning horizon takes place. Consequently, the power infrastructure design is not considered a Greenfield project. The initial power infrastructure, which is predominantly gas-based, conditioned the amount of gas required to meet the electricity demand in the forthcoming studied periods. Additionally, the generation capacities of alternative energy sources are limited by techno-economic constraints (e.g., availability of safe locations, high costs, lack of wind resources). Accordingly, the price elasticity of demand for natural gas is limited by techno-economic factors and energy security reasons. The UAE’s government plans for gas-based generation to remain as the core of its power sector. Accordingly, as shown in Fig. 6, the cumulative gas-based generated power varies along the gas price. This implies a negative price elasticity of demand for the natural gas because the low gas-based

0.9 0.8

PROBABILITY DISTRIBUTION

is expected to remain likewise. On the other hand, the price of the contracted Qatari gas is not expected to experience abrupt changes since it is delivered under a long-term agreement between the two countries. Whereas, the LNG imports through Fujairah are also considered to be supplied under long-term contracts. Moreover, uncertainty has been considered only in the fourth period (year 2030) to evaluate potential changes in the power infrastructure in the first and second half of the planning horizon compared with the base case study (i.e., previous deterministic case). Also, 2030 is a key year for the Emirate of Abu Dhabi in terms of energy planning. The emirate is expected to meet specific energy goals [46]. Additionally, given the overall size of the problem and significant number of integer variables, uncertainty has only been considered in one of the periods to keep the problem tractable in terms of computing time. The multiscenario multiperiod approach presented in problem (21) was used to study the power costs and electricity generation over the planning horizon in the presence of gas price uncertainty.

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 721

723

725

727

729

731

733

CUMULATIVE GAS -BASED GENERATED POWER (TWh) Fig. 6. Distribution of the cumulative gas-based power for the analyzed time periods between 2015 and 2040.

188

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190

Future work on this field includes the consideration of power generation and water desalination in a single mathematical model since both are deeply interconnected in the UAE. Both resources are traditionally produced in co-generation plants. As a result, the mismatch between electricity and water requirements during the summer could be analyzed. This is, during the summer the electricity demand substantially rises due to cooling requirements. Thus, a mathematical tool considering both activities could be used to simultaneously optimize resources in those areas.

PROBABILITY DISTRIBUTION

0.3

0.25

0.2

0.15

0.1

Appendix A. Supplementary equations of the design multiperiod model

0.05

0 36

37

38

39

40

41

42

43

44

CUMULATIVE NATURAL GAS COST (BILLION $) Fig. 7. Distribution of the cumulative natural gas costs for the analyzed time periods between 2015 and 2040.

generation values (shown in the figure) correspond to high gas prices, whereas high generation values correspond to low gas prices. This is, an increase in the natural gas price reduces its demand in the power sector, and corresponding gas-based power output. Nevertheless, the highest probability (0.83), which is strongly marked, corresponds to the highest cumulative gas-based generation value. This is due to the anticipated key role planned for gas-based generation in the Emirate’s power sector, and techno-economic constraints considered for the design of the power infrastructure. The distribution of the cumulative gas costs for the studied periods is given in Fig. 7. As shown in the figure, the natural gas is the most important component of the electricity costs. It represents around 50% of the total cost in the considered periods of the planning horizon. As one could expect, the higher range of values correspond to high natural gas costs whereas lower values correspond to low gas prices. Although the range of gas prices analyzed in the present scenario is wide, the difference between the lowest and highest probable cumulative gas costs is not significantly high (around 8 billion $). This means that regardless the gas price level; the gas costs will represent an important cost burden in the Emirate’s power sector for the foreseeable future given its predominance. 5. Conclusions

The supplementary equations describing in more detailed the proposed design multiperiod optimization model are included in the present section. A.1. Power capacity decommission and imports The decommissioned generation capacity in the power fleet can be estimated as follows:

DEp;t ¼ xp;t ICp CFp ;

8t; 8p 2 Exist

ðA:1Þ

where DEp;t is the decommissioned generation capacity of power units’ type p in period t. The type and number of decommissioned power plants by time period depends on the unit’s operating lifetime. Additionally, the UAE can receive electricity transfers from the GCC countries through the international interconnected grid. The power imported by the country is given as:

TIðsÞ t ¼

X ðsÞ ET c;t ;

8t

ðA:2Þ

c ðsÞ where ET c;t represents the electricity transfers from the GCC country c to the UAE in period t.

A.2. Feedstock fuels consumption The natural gas is the key feedstock fuel in the model since gas-based electricity generation is set to remain playing an important role in the power sector for the foreseeable future. The natural gas supply for the power sector derives from different sources; this can be expressed as follows:

TNGðsÞ t ¼

X

ðsÞ

GSf ;t ;

8t

ðA:3Þ

f

In this work a multiperiod design optimization model for the power sector under gas price uncertainty is proposed. The results show that gas-based generation remains as the leading electricity production technology. Additionally, nuclear power was found to be a very attractive option for the Emirate. It will allow reducing generation costs and CO2 emissions dramatically, as well as increasing the Emirate’s energy security by diversifying the energy mix. The high concentration of gas-based generated power could be threatened in the near to medium term future by the large consumption of natural gas in the oil sector for EOR operations. If the present trend holds, the Emirate may face the dilemma to decide where to send its natural gas resources: (1) oil fields for EOR operations, or (2) the power sector. The EOR techniques helped to nearly double the proven oil reserves in Abu Dhabi over the past decade. Moreover, the oil industry is the main income source of the Emirate and the core of its economy. On the other hand, the widespread use of electricity has enabled increasing the living standards and modernized the country. Based on the above, having to decide between these two options may not be an easy task, and either choice would hurt the country’s economy.

where TNGtðsÞ denotes the total supply of natural gas in UAE’s power sector. Moreover, the total fuel consumption (i.e., natural gas or uranium) in the power sector can be estimated as follows: ðsÞ

TFC i;t ¼

ðsÞ X EGp;t HRp

HVp;i

p

;

8t; i; 8p 2 Gas; Nu

ðA:4Þ

ðsÞ

where TFC i;t represents the total consumption of fuel i (i.e., natural gas or uranium), HRp denotes the power plant’s p heat rate, and HVp,i is the average heating value associated to fuel i used in the pth power plant. The natural gas is considered to be obtained from the country’s domestic production as well as pipeline and LNG cargoes imports. On the other hand, the uranium for the nuclear power plants is imported from the international market. A.3. Air emissions generation and avoidance The optimization method estimates the GHG and CAC emissions produced by the gas-based power plants. The renewable and

189

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190

nuclear plants are considered to be non-emitting sources in this study. Accordingly, the air emissions e (e.g., CO2, CH4, N2O, NOx, SO2, and PM10) produced by the power plants can be estimated as follows:

EMðsÞ e;t ¼

X

EGðsÞ p;t EFp;e ;

8e; t

ðA:5Þ

ðsÞ where EMe;t is the amount of air emission e produced by the gas-based power fleet in time period t, and EFp;e is the air emission factor for e associated to the pth plant. On the other hand, the use of alternative energy plants allows the avoidance of air emissions compared to a business as usual (BAU) operation in the power sector. Accordingly, the amount of emissions avoided by using these energy sources can be estimated as follows:

¼ AFe

SBtðsÞ ¼

X

CC ðsÞ p;t CAD þ

p2CCS

p2Gas

AEðsÞ e;t

remediation measures that would have to be spent otherwise. These social benefits are included in the model as follows:

X

ðsÞ EGp;t ;

8e; t

ðA:6Þ

p2Re;Nu ðsÞ where AEe;t is the amount of air emissions e avoided by using alternative energy sources instead of conventional gas-based plants in period t, and AFe represents the air emission e avoidance factor. The avoidance factor is given in terms of the average emission associated to the operation of a conventional gas-dominated power system such as that currently existing in the UAE.

A.4. Carbon capture and storage (CCS) The mathematical model includes carbon capture and storage as CO2 mitigation strategy. Accordingly, the CO2 capture systems considered in the present work include: pre-combustion (i.e., for NGCC plants) and post-combustion (i.e., for Oxyfuel plants) as CCS methods. The CO2 is considered to be stripped from the unit’s power generation process using MEA as absorbent. The captured CO2 is considered to be injected into oil fields for enhanced oil recovery (EOR) projects, or permanently sequestered in suitable geological sites. In EOR applications, the captured CO2 can replace natural gas and water injection. Gas injection under oilfields can simultaneously store GHG and help boost oil output by maintaining underground pressure. Given the potential opportunity to sustain future oil production and lower carbon emissions from point sources, CCS is regarded as an attractive option for the UAE and other oil producing countries. In the UAE, it is estimated that employing CO2-enhanced oil recovery can help replace a significant share of the natural gas used in oil production (approximately 40%) operations, and be used for power generation. The CO2 that can be avoided using plants with CCS methods (CC ðsÞ p;t ) is given as follows: ðsÞ CC ðsÞ p;t ¼ EGp;t CCFp ;

8t; 8p 2 CCS

ðA:7Þ

where CCFp is the carbon capture factor associated to the pth plant. Furthermore, the compression power required to transport the captured CO2 via pipeline can be estimated as: ðsÞ CPðsÞ 8t; 8p 2 CCS p;t ¼ CC p;t PLp CPF;

ðA:8Þ

ðsÞ where CP p;t is the compression power required to transport the CO2 capture from the pth plant to its storage site, PLp is the pipeline length travelled by the CO2 captured at the pth plant, and CPF is the compression power factor that defines the power requirements to transport the CO2. The avoidance of air emissions either by using alternative power sources or CCS methods has implicit social and environmental benefits in terms of saved public health costs and ecological

X ðsÞ AEe;t ADe ;

8t

ðA:9Þ

e

where SBtðsÞ represents the total social benefit related to the air emissions avoided in the sth scenario of period t, CAD is the CO2 emission avoided damage, and ADe is the avoided damage related to the eth emission. A.5. Capital recovery factor The annual capital recovery factor for the power plants can be estimated as follows:

RFp ¼

IRð1 þ IRÞLTp ð1 þ IRÞLTp  1

8p

ðA:10Þ

where IR is the annual real debt interest rate and LTp is the depreciation time of the pth plant. The parameter IR denotes the rate of interest that an investor expects to receive after accounting for the annual inflation. It can be described more formally as the nominal interest rate minus the inflation rate. References [1] Abu Dhabi Water and Electricity Company. Statistical Report 1998–2012. Abu Dhabi, UAE: Abu Dhabi Water and Electricity Authority; 2013. [2] Matsumoto T. Abu Dhabi energy policy – energy problems plaguing Abu Dhabi and their implications for Japan. Japan: The Institute of Energy Economics Japan (IEEJ); 2013. [3] Bachellerie IJ. Renewable energy in the GCC countries: resources, potential, and prospects. Gulf Research Center; 2012. p. 230. [4] Nader S. Paths to a low-carbon economy—The Masdar example. Energy Procedia 2009;1(1):3951–8. [5] UAE Ministry of Foreign Affairs. UAE energy outlook: factors for global renewable energy roadmaps. United Arab Emirates; 2012. p. 3. [6] Almansoori A, Betancourt-Torcat A. Design optimization model for the integration of renewable and nuclear energy in the United Arab Emirates’ power system. Appl Energy 2015;148:234–51. [7] Avetisyan M, Bayless D, Gnuni T. Optimal expansion of a developing power system under the conditions of market economy and environmental constraints. Energy Econ 2006;28(4):455–66. [8] Nikolova S, Causevski A, Al-Salaymeh A. Optimal operation of conventional power plants in power system with integrated renewable energy sources. Energy Convers Manage 2013;65:697–703. [9] Ji L et al. Development of an inexact risk-aversion optimization model for regional carbon constrained electricity system planning under uncertainty. Energy Convers Manage 2015;94:353–64. [10] Li YP, Huang GH. Electric-power systems planning and greenhouse-gas emission management under uncertainty. Energy Convers Manage 2012;57:173–82. [11] Piao MJ, Li YP, Huang GH. Development of a stochastic simulation– optimization model for planning electric power systems – a case study of Shanghai, China. Energy Convers Manage 2014;86:111–24. [12] AlFarra HJ, Abu-Hijleh B. The potential role of nuclear energy in mitigating CO2 emissions in the United Arab Emirates. Energy Policy 2012;42:272–85. [13] Sgouridis S et al. A sustainable energy transition strategy for the United Arab Emirates: evaluation of options using an Integrated Energy Model. Energy Strat Rev 2013;2(1):8–18. [14] Lazarus M, Heaps C, Raskin P. LEAP long-range energy alternatives planning system. Boston, Massachusetts: Tellus Institute; 1997. [15] Mondal MAH, Kennedy S, Mezher T. Long-term optimization of United Arab Emirates energy future: policy implications. Appl Energy 2014;114:466–74. [16] Bistline JE. Natural gas, uncertainty, and climate policy in the US electric power sector. Energy Policy 2014;74:433–42. [17] Vithayasrichareon P, Riesz J, MacGill IF. Using renewables to hedge against future electricity industry uncertainties—an Australian case study. Energy Policy 2015;76:43–56. [18] Black & Veatch Holding Company. Cost and performance data for power generation technologies. National Renewable Energy Laboratory; 2012. [19] Rubin ES, Rao AB, Chen C. Comparative assessments of fossil fuel power plants with CO2 capture and storage. In: Rubin ES et al., editors. Greenhouse gas control technologies, vol. 7. Oxford: Elsevier Science Ltd.; 2005. p. 285–93. [20] Rubin ES, Chen C, Rao AB. Cost and performance of fossil fuel power plants with CO2 capture and storage. Energy Policy 2007;35(9):4444–54. [21] Davison J. Performance and costs of power plants with capture and storage of CO2. Energy 2007;32(7):1163–76.

190

A. Betancourt-Torcat, A. Almansoori / Energy Conversion and Management 100 (2015) 177–190

[22] ICF International Company. Technology characterization: steam turbines. Washington, DC: Environmental Protection Agency; 2008. [23] U.S. Energy Information Administration. Updated capital cost estimates for utility scale electricity generating plants. Washington, DC: U.S. Department of Energy; 2013. [24] Harijan K et al. Assessment of centralized grid connected wind power cost in coastal area of Pakistan. Renewable Energy 2009;34(2):369–73. [25] Janajreh I, Su L, Alan F. Wind energy assessment: Masdar City case study. Renewable Energy 2013;52:8–15. [26] Shawon MJ, El Chaar L, Lamont LA. Overview of wind energy and its cost in the Middle East. Sustain Energy Technol Assess 2013;2:1–11. [27] Radhi H. Trade-off between environmental and economic implications of PV systems integrated into the UAE residential sector. Renew Sustain Energy Rev 2012;16(5):2468–74. [28] Harder E, Gibson JM. The costs and benefits of large-scale solar photovoltaic power production in Abu Dhabi, United Arab Emirates. Renewable Energy 2011;36(2):789–96. [29] Straatman PJT, van Sark WGJHM. A new hybrid ocean thermal energy conversion–Offshore solar pond (OTEC–OSP) design: a cost optimization approach. Sol Energy 2008;82(6):520–7. [30] World Nuclear Association. The economics of nuclear power. ; 2014 [cited 2014 04/02]. [31] Nuclear Energy Institute. Costs: fuel, operation, waste disposal & life cycle. ; 2013. [32] Alonso G, et al. Alternatives of financing for new nuclear reactors in Mexico. I.A.E. Agency, Editor. Estado de Mexico, Mexico: Instituto Nacional de Investigaciones Nucleares; 2013. [33] Areva. Key assets of the EPR reactor: competitiveness. ; 2013 [cited 2014 25-02]. [34] Krane J. Energy policy in the Gulf Arab States: shortage and reform in the World’s storehouse of energy. Cleveland, US: USAEE; 2013. Editor. [35] Grossmann IE, Sargent RWH. Optimum design of chemical plants with uncertain parameters. AIChE J 1978;24(6):1021–8. [36] Karuppiah R, Grossmann IE. Global optimization of multiscenario mixed integer nonlinear programming models arising in the synthesis of integrated water networks under uncertainty. Comput Chem Eng 2008;32(1–2): 145–60. [37] Paules Iv GE, Floudas CA. Stochastic programming in process synthesis: a twostage model with MINLP recourse for multiperiod heat-integrated distillation sequences. Comput Chem Eng 1992;16(3):189–210. [38] Rooney WC, Biegler LT. Incorporating joint confidence regions into design under uncertainty. Comput Chem Eng 1999;23(10):1563–75. [39] Rooney WC, Biegler LT. Design for model parameter uncertainty using nonlinear confidence regions. AIChE J 2001;47(8):1794–804. [40] Rooney WC, Biegler LT. Optimal process design with model parameter uncertainty and process variability. AIChE J 2003;49(2):438–49. [41] Raspanti CG, Bandoni JA, Biegler LT. New strategies for flexibility analysis and design under uncertainty. Comput Chem Eng 2000;24(9–10): 2193–209. [42] Gomes U et al. Optimal infrastructure of the upgrading operations in the oil sands under uncertainty: a Multiscenario MINLP approach. Ind Eng Chem Res 2014. [43] Almansoori A, Shah N. Design and operation of a stochastic hydrogen supply chain network under demand uncertainty. Int J Hydrog Energy 2012;37(5):3965–77. [44] Al-Qahtani K, Elkamel A, Ponnambalam K. Robust optimization for petrochemical network design under uncertainty. Ind Eng Chem Res 2008;47(11):3912–9.

[45] Betancourt-Torcat A et al. Stochastic modeling of the Oil Sands operations under Greenhouse Gas emission restrictions and Water management. Energy Fuels 2013;27(9):5559–78. [46] Abu Dhabi Council for Economic Development. The Abu Dhabi economic vision 2030; 2008. [47] Abu Dhabi’s Regulation and Supervision Bureau. Supporting the Abu Dhabi economic vision 2030. Abu Dhabi, UAE: RSB; 2013. [48] GAMS Development Corporation. General Algebraic Modeling System (GAMS). ; 2013 [cited 2014 16-02-2014]. [49] GAMS Development Corporation. Cplex solver manual. ; 2013 [cited 2014 16-02-2014]. [50] Miller K. ADWEC Winter 2012/2013: electricity and water demand forecasts. Abu Dhabi Water and Electricity Company; 2013. p. 17. [51] Emirates Nuclear Energy Corporation. Emirates Nuclear Energy Corporation Files License Application and Environmental Assessment for UAE Civil Nuclear Power Program. ; 2010 23/ 04/2010 [cited 2014 16/10/2014]. [52] Mubadala Company. Emirates LNG: a strategic gas supply project for the UAE. ; 2014 [cited 2014 22/12]. [53] International Energy Agency. World Energy Outlook 2011. Paris: International Energy Agency; 2011. [54] Dolphin Energy. Marketing & distribution: natural gas. ; 2013 23-07-2014 [cited 2014 18-09]. [55] Energy Information Administration. Country Analysis: United Arab Emirates. Washington, DC: U.S. Department of Energy; 2013. [56] Al-Asaad HK, Ebrahim AA. The GCC power grid: benefits & beyond. Kingdom of Saudia Arabia: Gulf Cooperation Council Interconnection Authority; 2012. [57] Al-Asaad HK, Al-Mohaisen A, Sud S. GCC power grid: transforming the GCC power sector into a major energy trading market. Kingdom of Saudi Arabia: Gulf Cooperation Council Interconnection Authority; 2012. [58] Al Kindi H. Abu Dhabi’s gas dilemma. In: The economic review autumn 2010. Abu Dhabi, UAE: Abu Dhabi Council for Economic Development; 2010. [59] Occidental Petroleum Corporation. Dolphin pipeline. ; 2014 [cited 2014 18-09]. [60] Medlock III KB. Natural gas price in Asia: what to expect and what it means. Houston: Rice University; 2014. [61] PR Newswire. Gasco Energy announces 2012 Financial and Operating results and estimated proved reserves. ; 2013 [cited 2014 03-03]. [62] Natural Resources Canada. World Oil Demand and Supply to 2030. Long Term Outlook: Crude Oil Prices to 2030. ; 2010 24/11/2014 [cited 2014 27/11/2014]. [63] Ernst & Young. Global LNG: will new demand and new supply mean new pricing? United Kingdom; 2014. [64] Jordehi AR. A review on constraint handling strategies in particle swarm optimisation. Neural Comput Appl 2015:1–11. [65] Rezaee Jordehi A. Parameter selection in particle swarm optimisation: a survey. J Exp Theor Artif Int 2013;25(4):527–42. [66] Jordehi AR. Enhanced leader PSO (ELPSO): a new PSO variant for solving global optimisation problems. Appl Soft Comput 2015;26:401–17. [67] Rezaee Jordehi A. Brainstorm optimisation algorithm (BSOA): an efficient algorithm for finding optimal location and setting of FACTS devices in electric power systems. Int J Electr Power Energy Syst 2015;69:48–57. [68] Rezaee Jordehi A. Enhanced leader PSO (ELPSO): a new algorithm for allocating distributed TCSC’s in power systems. Int J Electr Power Energy Syst 2015;64:771–84.