An initiative towards an energy and environment scheme for Iran: Introducing RAISE (Richest Alternatives for Implementation to Supply Energy) model

Energy Policy 89 (2016) 36–51

Contents lists available at ScienceDirect

Energy Policy journal homepage: www.elsevier.com/locate/enpol

An initiative towards an energy and environment scheme for Iran: Introducing RAISE (Richest Alternatives for Implementation to Supply Energy) model Hadi Eshraghi a,n, Mohammad Sadegh Ahadi a a

National Climate Change Office, Department of Environment, Tehran, Iran

H I G H L I G H T S








Combined cycle power plant is the best option to meet base load requirements. There's synergy between climate change mitigation and economic affordability. Power sector reacts to an emission cap by moving towards renewable energies. Instead of being exported, condensates should be refined by condensate refineries Iran's refining sector should be advanced by shifting to RFCC-equipped refineries.

art ic l e i nf o

a b s t r a c t

Article history: Received 10 October 2014 Received in revised form 16 October 2015 Accepted 17 October 2015

Decision making in Iran's energy and environment-related issues has always been tied to complexities. Discussing these complexities and the necessity to deal with them, this paper strives to help the country with a tool by introducing Richest Alternatives for Implementation to Supply Energy (RAISE), a mixed integer linear programming model developed by the means of GNUMathprog mathematical programming language. The paper fully elaborates authors' desired modeling mentality and formulations on which RAISE is programmed to work and verifies its structure by running a widely known sample case named “UTOPIA” and comparing the results with other works including OSeMOSYS and Temoa. The model applies RAISE model to Iranian energy sector to elicit optimal policy without and with a CO2 emission cap. The results suggest promotion of energy efficiency through investment on combined cycle power plants as the key to optimal policy in power generation sector. Regarding oil refining sector, investment on condensate refineries and advanced refineries equipped with Residual Fluid Catalytic Cracking (RFCC) units are suggested. Results also undermine the prevailing supposition that climate change mitigation deteriorates economic efficiency of energy system and suggest that there is a strong synergy between them. In the case of imposing a CO2 cap that aims at maintaining CO2 emissions from electricity production activities at 2012 levels, a shift to renewable energies occurs. & 2015 Elsevier Ltd. All rights reserved.

Keywords: Sustainable development Energy modeling Climate change mitigation Linear programming CO2 cap

1. Introduction During recent decades Iran's population, experiencing a 2.1% growth rate, has reached from about 38.9 million people in 1980 to 76.4 million in 2012 (WB Webpage). As a result, country's need to an efficient energy system has increased. On the other hand, with abundant natural resources, Iran's economy has always been reliant

n Correspondence to: Pardisan Eco-park, Hakim Expressway, Tehran, Iran. Fax: þ98 2188233092. E-mail addresses: [email protected], [email protected] (H. Eshraghi), [email protected], [email protected] (M.S. Ahadi).

http://dx.doi.org/10.1016/j.enpol.2015.10.023 0301-4215/& 2015 Elsevier Ltd. All rights reserved.

on foreign revenues from export of crude oil. Accordingly, Iran's energy sector has been the focal point of researchers as to “how” and “how costly” Iran could have an efficient energy system. In addition some newly emerged issues such as environmental concerns about growing Green House Gases (GHGs) emissions have posed more difficulties to the “problem of decision” in Iran. In the following section, we explain the reasons having prompted us to develop a domesticated energy optimization model. 1.1. Backgrounds There is a crucial need to a tool capable of quantifying decision space for Iran mainly because of the following bottlenecks:

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

Lack of a roadmap for end-use technologies: for decades, Iranian governments used to pay burdensome subsidies for energy that in turn has led to country's final energy intensity to soar up. In such an environment, any energy conservation measures wouldn't make sense until after 2011 subsidy reform plan. But the current dilemma is that the government does not have a deep insight on which end-use technologies to promote. Moreover, energy conservation measures and its extent for different end-use devices, has a trade off with business-as-usual supply of energy that is reflected in an energy model. This model can be used in prioritizing high potential conservation options and allocating presently limited and unplanned resources to these options. The conundrum of natural gas: possessing 18% of world gas reserves, Iran is the top holder of this fuel (BP, 2013). Growing at a rate of 10.8% during 2001 to 2012 (Institute for International Energy Studies (IIES), 2012), natural gas is supposed to play a key role in any prospective development plans of the country but presently its utilization is limited to few traditional ways including burning to generate heat in demand side or power plants, and feedstock of petrochemical industry. Export (either through pipeline or LNG), conversion via GTL to produce petroleum products or conversion to hydrogen (which in turn could be used as the feed of many hydrogen-based technologies) are counted as other choices for natural gas planning not thought about seriously so far. Diverse potent fuels: apart from oil and natural gas for which Iran is widely renowned, there are many other fuels and renewable potentials like coal, solar, wind and geothermal. Each of these resources has its own advocates and critics but it is not possible to dismiss or promote none of these options unless a prescriptive model suggests which one would excel the other. Critical environmental situation: energy-related CO2 emissions was 624 million metric tons in 2011, ranking Iran 8th in the list of top ten CO2 emitters (EIA Webpage). Such amount of carbon emission could presumably put Iran among countries with mitigation commitment in coming years and consequently lead to more complexity in decision making, since implications of such a shift to a low carbon economy are quit vague. 1.2. Motivations Above mentioned reasons prove the need to develop an integrated energy model capable of capturing all discussed countryspecific issues. On the other hand application of widely known energy models is usually linked with the following barriers: Huge amount of resources, both human and financial, are required to have these models run and this is not what the country would be willing to afford. For some of these models (e.g. MARKAL/TIMES), even if the resource limitations were lifted, it wouldn't be possible for the country to access a full package of them due to international sanctions. Even access to powerful solvers is limited and we must rely on freely available ones such as GNU Linear Programming Kit (GLPK) (GNU Webpage). Many of these models do not allow modeler to access all modeling variables, leading to shortcomings in adding desirable constraints. Furthermore, Integration of these models with supplementary modules for further assessments (e.g. uncertainty analysis) if possible, is very difficult (Hunter et al., 2013). The idea that led to developing RAISE (Richest Alternatives for Implementation to Supply Energy) and its feasibility is inspired by the brilliant work of Howells in which OSeMOSYS as an open source code for energy modeling in developing countries is introduced (Howells et al., 2011). OSEMOSYS's outstanding feature is that the level of modeling can be developed by adding functional blocks. Despite this advantage, its application in the modeling structure on which it is built upon, may be subject to shortcomings

37

concerning the size of Mathematical Programming System (MPS) file and memory (RAM) it takes to reach the solution, what will be shown later in Section 3.2. Additionally, our restriction as to being forced to use a freely available solver rather than a commercial one brings about this necessity that our model should be as numerically-smart as possible. So we found it useful to think about the ways that could make RAISE's modeling framework more efficient. Keeping in mind the advantages in modeling frameworks of OSeMOSYS (Howells et al., 2011), Temoa (Hunter et al., 2013) and MESSAGE (IIASA (International Institute of Applied Systems Analysis), 2001), we engendered the mathematical formulation of RAISE model such that it could serve these two objectives: (1) be able to include all country-related issues explicitly and (2) be as efficient as possible. Like OSeMOSYS, the programming language used to implement the model is GNUMathprog (Makhorin, 2010), a free open source tool which is as powerful as other model generators such as GAMS (Howells et al., 2011).

2. Underlying method As is usually the case with many classic energy models, present work uses mixed integer linear programming, defined by: (1) an objective function denoting net widely discounted costs of energy system which is to be minimized and (2) a set of constraints defining feasible decision space as a proxy of different technical, physical, social and economic realities in place. The key elements of RAISE model are technologies organized to extract, process, refine, convert, transport and distribute energy from upstream resources to end users, as is shown in Fig. 1. It is regarded as the Reference Energy System (RES) and is a useful conceptual model illustrating interactions that exist between technologies. In general, two key types of variables are defined for performance of technologies. The first one is a state variable denoting activity of the technology and the second one is the capacity of the technology which is a control variable limiting the first variable. In a broad classification, fuels (services) are divided into 3 groups: demand, intermediate and resource. Since demand of some fuels (services) may be subject to seasonal or daily fluctuations, the model allows production of some fuels to be unevenly distributed throughout the year. These fuels are depicted in Fig. 1 by dotted lines. This is only the case for fuels placed into demand and intermediate categories but not resource (resource fuels do not have in-year variations). Having said that, there would be 5 types of fuels: (i) evenly-distributed demand (e.g. transport in Fig. 1), (ii) unevenly-distributed demand (e.g. lighting or heating in Fig. 1), (iii) evenly-distributed intermediate (e.g. gas oil in Fig. 1), (iv) unevenly-distributed intermediate (e.g. electricity in Fig. 1), (v) resource fuels (e.g. oil in Fig. 1). Accordingly, technologies would be classified as follows: Technologies producing fuels which are without seasonal or daily fluctuations. Since load management does not matter here, only 1 activity variable is simply assigned to each operation mode of technologies of this kind: AO t,m,y refers to annual aggregated output of technology t working on operation mode m in year y. In Fig. 1 these technologies are signified by rectangular boxes. Technologies that at least one of their output fuels has in-year fluctuations and are shown by rhombic shapes in Fig. 1. Unlike the previous group, load management does matter here and for technologies lying in this category, it's necessary to define the same number of activity variables as the number of load regions: O t,m,l,y is the output of technology t working on operation mode m, in load region l and year y. Technologies that are similar to the second group but their annual production pattern is fixed by the modeler. In fact he/she

38

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

Fig. 1. A typical RES drawn to explain distinctions between various fuels and technologies.

determines the share of each load region in annual output of the technology. This could avail us in modeling intermittent renewable technologies that we have data about the availability of their resources in various load regions. These data suffice the mathematics of the model to define just one activity variable for these technologies. In Fig. 1 hydro power plant is of this kind and is shown by circular shape. Although such kind of classification may require more deliberation when it comes to developing capacity adequacy or balance constraints, it improves the capability of the model to perform calculations much faster, what becomes more important when dealing with a real case. Key relations and constraints of the model are presented in Table 1. However those may not be limited to what listed here and any other kinds of constraints including reliability considerations of power plants, market share of a specific technology, renewable targets, technology upgrade, dynamic constraints for penetration of new technologies, bounds on flow or capacities and economic resources limitations could be added thanks to open source nature of the model. Table 2 lists the notations used in Table 1.

3. Verification We used a widely known case named “UTOPIA” to test the model (Howell et al., 2011). To be compatible with RAISE, we drew RES of UTOPIA like what illustrated in Fig. 2. In UTOPIA there are 3 services among which 2 services heating and lighting (RH and RL) have in-year variable demands whereas transport (TX) doesn't. Technologies that work to meet transport demand (i.e. TXG, TXD and TXE) are first type technologies and those meeting heating and lighting demand (i.e. RL1, RHE and RHO) are second type technologies. This distinction is shown in Fig. 2 by different shapes for technologies and fuels as was explained in Section 2. There are 4 s type technologies E01, E21, E31 and E70 plus a pumped-storage technology E51 for generation and management of electricity (ELC). Technologies E01, E21 and E70 are fossil-based power plants consuming coal (HCO), uranium (URN) and diesel (DSL) respectively while E31 is a hydro power plant. Intermediate fuels GSL and DSL not only can be imported through technologies IMPGSL1 and IMPDSL1 but also can be provided by refining oil (imported by IMPOIL) in refinery SRE and intermediate fuels HCO and URN are

imported via technologies IMPHCO1 and IMPURN1 respectively. Among all intermediate fuels, only electricity (shown by dotted lines in Fig. 3), for the purpose of transmitting daily or seasonal demand-side variations to power plants, is set to have in-year variations. The verification is carried out by running UTOPIA in RAISE and comparing results with both OSeMOSYS and Temoa. Since there was disparity between OSeMOSYS and Temoa over data input, we present the verification results separately. 3.1. Verification with OSeMOSYS OSeMOSYS source code (“OSeMOSYS_2013_05_10_short.txt”) and data set of UTOPIA (“UTOPIA_2013_05_10.dat”) were downloaded from “www.osemosysmain.yolasite.com” and it was observed that what had been cited in Table 3 of (Howells et al., 2011) as UTOPIA input data differed a little from data file we had downloaded. Thus for the sake of the maximum transparency we list these disparities in Table 3, with bold data chosen for verification. Results of running UTOPIA by RAISE and OSeMOSYS (with above revisions), are shown in Fig. 3: MPS file in OSeMOSYS contains 8075 rows and 8820 columns while RAISE has 1755 rows and 1575 columns offering considerable superiority in terms of calculation speed. This is because OSeMOSYS assigns the same number as the number of load regions defined in the model, to the number of activity variables of all technologies, even if those technologies are not subjected to seasonal or daily fluctuations. Similarly it multiplies the number of activity variable of technologies by the number of operation mode of the technology with the maximum operation mode number. For instance if there were 6 load regions in model and only one of the technologies worked with 3 operation modes, then OSeMOSYS would define 18 different activity variable for all of the technologies. Then it assigns the value of zero to irrelevant activity variables of technologies with lower operation mode. Running the model in this way makes modeling more straightforward and less complicated but results in much lower calculation speed. Besides, post solution analysis which is based on output file and lp file, are rather troublesome to conduct. Objective function value is $32.3 billion in RAISE and $28.8 billion in OSeMOSYS. Although both models are apt to generate

Table 1 Underlying formulation of model. Capacity Adequacy Constraints: these sets of constraints tend to control the activity of technologies so that it does not exceed their available capacity. Equation Remark

∑m AOt , m, y ≤

PF t , y × (RC t , y + ∑ y

M t , ψ × N C t , ψ)

∀ t ∈ 1st group of technologies ∀ y ∈ planning years

∑m Ot , m, l, y ≤

PF t , y × (RC t , y + ∑ y

M t , ψ × NCt , ψ) × LRl, y

∀ t ∈ 2nd group of technologies ∀ l ∈ load regions in each year ∀ y ∈ planning years

(

f t , l, y )MAX LRl, y

if (y − br) < Lt ⇒ ψ = br if (y − br) ≥ Lt ⇒ ψ = y − Lt

if (y − br) < Lt ⇒ ψ = br if (y − br) ≥ Lt ⇒ ψ = y − Lt

PF t , y × (RC t , y + ∑ y

× ∑m AOt , m, y ≤

if (y − br) < Lt ⇒ ψ = br if (y − br) ≥ Lt ⇒ ψ = y − Lt

M t , ψ × N C t , ψ)

Energy Balance Equations: Equation

∀ t ∈ 3rd group of technologies ∀ y ∈ planning years Note: The region yielding higher value for (f t,l,y/LR l,y) is the peak operational region of the technology.

∑t ∑m AOt , m, y × S t , m, f , y = AED f , y

Remark ∀ f ∈ Evenly distributed Demand ∀ y ∈ planning years

∑t ∈ 2ndgroup ∑m Ot , m, l, y × S t , m, f , y + ∑t ∈ 3rdgroup ∑m f t , l, y × AOt , m, y × S t , m, f , y

∀ f ∈ Unevenly distributed Demand ∀ l ∈ load regions in each year ∀ ∀ f ∈ Evenly distributed Intermediate ∀ y ∈ planning years Note: f could be gas oil in Fig. 1

∑t ∑m AOt , m, y × S t , m, f , y = ∑t ∈ 1st group ∑m AOt , m, y × FRt , m, f , y + ∑t ∈

2nd group ∑l ∑m O

t , m, l, y

× FRt , m, f , y +

∑t ∈ 3rd group ∑m

AOt , m, y

FRt , m, f , y

×

∑t ∈ 2ndgroup ∑m

Ot , m, l, y

× S t , m, f , y + ∑t ∈ 3rdgroup ∑m f t , l, y × AOt , m, y × S t , m, f , y

+ ∑t ∑k, k ≠ l SOt , f , k, l, s, y × εt , f , y =

SIt , f , l, s, y ∑t ηt , f , y

y ∈ planning years

+

LRl, y × ∑t ∈ 1st group ∑m AOt , m, y × FRt , m, f , y + ∑t ∈ + ∑t ∈ 3rd group ∑m

f t , l, y

×

AOt , m, y

×

2nd group ∑m O

t , m, l, y

× FRt , m, f , y

FRt , m, f , y ∀ f ∈ Resources ∀ y ∈ planning years

AE f , y = ∑t ∈ 1st group ∑m AOt , m, y × FRt , m, f , y + ∑t ∈

∀ f ∈ Unevenly distributed Intermediate ∀ l ∈ load regions in each year ∀ Note: f could be electricity in Fig. 1

t , m, l, y × FRt , m, f , y + 2nd group ∑l ∑m O

∑t ∈ 3rd group ∑m AOt , m, y × FRt , m, f , y Daily Storage Constraints: Energy can be stored and released in different parts of day in each season. Here, load regions (in each season) are ordered in the sequence they appear in an actual day. Equation

SIt , f , l, s, y ≥ ∑k , k ≠ l SOt , f , l, k, s, y

a

SIt , f , l, s, y + ∑k , k ≠ l SOt , f , k, l, s, y ≤

PF t , y

×

(RC t , y

+ ∑

y if (y − br) < Lt ⇒ ψ = br if (y − br) ≥ Lt ⇒ ψ = y − Lt

M t,ψ

×

NCt , ψ)

×

LRl, y

PF t , y × (RVC t , y + ∑ y

if (y − br) < Lt ⇒ ψ = br if (y − br) ≥ Lt ⇒ ψ = y − Lt

f ,ψ ≤ R f ∑endyear ψ = base year AE

This equation is written for every two consecutive load regions (l and L-1) of a season. For instance if there were 3 load regions such as morning, noon and night in a season, there would be 3 equations in the form of this equation to related morning to noun, noon to night and night to morning. Remark ∀ f ∈ Resources 39

Environmental Constraints:

This inequality ensures that available capacity of storage technology t in year y is adequate enough to provide flow of fuel f in each load region l defined in season s. k is the counter for load regions with the same seasonal index as l.

M t , ψ × NVCt , ψ)

It , l, s, y = SIt , f , l − 1, s, y − ∑k , k ≠ l − 1 SOt , f , k, l − 1, s, y + ξt , l − 1, l, s, y × It , l − 1, s, y Resource Constraints: Equation

Remark For each storage technology t linked with fuel f, inequality stands for each load region l defined in season s and year y.

In order to prevent storage content overflow, this inequality states that in each load region l, net exchange of energy plus initial content at the beginning of that load region must not exceed allowable volumetric capacity of storage system t.

SIt , f , l, s, y − ∑k , k ≠ l SOt , f , k, l, s, y + It , l, s, y 365 × LRl, y

≤

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

y ∈ planning years

= AUD f , y × DDl, f , y

40

Table 1 (continued ) Capacity Adequacy Constraints: these sets of constraints tend to control the activity of technologies so that it does not exceed their available capacity. Equation Remark Equation

∑t ∈ 1st group ∑m AOt , m, y × χ t , m, g , y + ∑t ∈

2nd group ∑m ∑l O

+ ∑t ∈ 3rd group ∑m AOt , m, y × χ t , m, g , y ≤ C

t , m, l, y

Remark ∀ y ∈ planning years

× χ t , m, g , y

g,y

Objective Function: There might be costs on production of particular resources. Moreover additional costs may be involved as to consumption of economic resources tied to both introduction of new capacities and flows of energy. Other costs concern externalities arising from emission of GHGs and air pollutants. These externalities can be internalized by taking them into account when developing objective function, provided that the unit external cost of GHG or pollutant is known. Nevertheless, the following formulation is believed to be the most hindering costs ahead of Iran's energy system.

Objective Function=Capex+Opexfix + Opexvar Capex = ∑t (∑ey ψ = br

IC t , ψ × Mt , ψ × NC t , ψ (1 + dr )ψ − br

(1 + dr )ey − ψ + 1 − 1 ) t (1 + dr )L − 1 ) (1 + dr )ey − br

IC t , ψ × Mt , ψ × NC t , ψ× (1 −

− ∑ey

ψ = ey − Lt + 1

= ∑ey ψ = by

1 (1 + dr )ψ − by + 0.5

(∑t FC t , ψ × (RC t , ψ + ∑ψ

if (ψ − br) < Lt ⇒ δ = br if (ψ − br) ≥ Lt ⇒ δ = ψ − Lt

M t , δ × N Ct , δ))

Opexvar ∑t ∈ 1stand3rdgroup ∑m VC t , m, ψ × AOt , m, ψ + ∑t ∈ 2ndgroup ∑m VC t , m, ψ × (∑l Ot , m, l, ψ) = ∑ey ψ = by

1 (+ (1 + dr )ψ − by + 0.5

) ∑t ∈ StorageTechnologies ∑s ∑l ∈ s (VCSI t , y × SIt , f , l, s, y − ∑k , k ≠ l SOt , f , k, l, s, y × VCSOt , y )

a Although letter f in both sides indicates that input and output flows are the same, it is possible to model variant input and output carriers with the same methodology and minor changes in source code. An example would be electrical heat storage systems.

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

Opexfix

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

41

Table 2 Description of notations used in Table 1. AO t,m,y O t,m,l,y NC t,ψ M t,ψ RC t,y PF t,y by LR l,y f t,l,y AED f,y AUD f,y DD l,f,y S t,m,f,y FR t,m,f,y SI t,f,l,s,y SO t,f,l,k,s,y η t,f,y ε t,f,y AE f,y ξ t,l,k,s,y NVC t, ψ RVC t,y I t,l,s,y χ t,m,g,y C g,y ey IC t,y FC t,y VC t,m,y VCSI t,y VCSO t,y

Annual output of technology t working on operation mode m and year y. Output of technology t working on operation mode m, in load region l and year y. New capacity of technology t proposed (by model) to be built and utilized in year ψ. (If capacities were to be integer, it would be numbers of units with a userspecified capacity M t,ψ.) Minimum allowable addition size for building new capacities of technology t in year ψ if integer programming is used, otherwise it's equal to 1. Residual capacity of technology t in year y (y refers to programming years not past years.) Plant factor of technology t in year y (which is the ratio of the programmable capacity of a plant to its rated capacity.) base year length of load region l as a fraction of year y. Share of load region l in output of technology t with fixed production pattern in year y. Annual evenly-distributed Demand for fuel (service) f in year y. This demand doesn't vary in different time zones of year. Annual unevenly-distributed Demand for fuel (service) f in year y. This demand fluctuates in different time zones of year. Demand distribution of fuel f in load region l and year y. Share of fuel (service) f in output stream of technology t working on operation mode m and year y. Fuel requirements of technology t for fuel f, in order to deliver 1 unit of output in operation mode m and year y. The lower the FR t,m,f,y the more efficient the technology. Input of fuel f to storage technology t in load region l and year y. s is the season that l belongs to. Output of fuel f from storage technology t in load region k and year y which was put into storage in load region l and year y. s is the season that l and k belong to. efficiency of input route to storage technology t for storing fuel f in year y. efficiency of output route from storage technology t for releasing fuel f in year y. Annual extraction of resource fuel f in year y. loss in content of the storage stock from load region l to k which are in season s and year y. New volumetric capacity supporting flows through storage technology t to be invested and utilized in year ψ. (If capacities were to be integer, it would be numbers of units with a user-specified minimum capacity.) Residual volumetric capacity supporting flows through storage technology t in year y (y refers to programming years not past years.) Initial accumulated energy content in storage technology t, at the beginning of the load region l which is in season s and year y. emission of gas g in year y resulting from 1 unit of activity of technology t in operation mode m. maximum allowable annual emission of gas g in year y. end year. Investment cost for installation of 1 unit of capacity of technology t in year y. Annual fixed cost for maintenance and operation of 1 unit of capacity of technology t in year y. Variable costs for maintenance and operation per unit of output of technology t working on operation mode m in year y. Variable cost for maintenance and operation per unit of input to storage technology t in year y. Variable cost for maintenance and operation per unit of output from storage technology t in year y.

almost same trends for technology selection, overall capacities in RAISE are higher than OSeMOSYS's. Such overinvestment is only the case for technologies producing ELC, RH and RL not those producing TX. This is because OSeMOSYS asks the modeler to distinguish between two types of technologies: technologies forced to meet in-year peak power requirements and those which are only supposed to meet annual loadings. Obviously placing a

technology whose output fuel is either used to meet an in-year variable demand directly (e.g. RHE, RHO and RL1) or is converted by downstream technologies and afterwards is used to meet an inyear variable demand (e.g. E01), in the first category might result in higher capacities for that technology in the optimal solution. OSeMOSYS developers in the UTOPIA data file have not forced three end-use devices RHE, RHO and RL1 to adjust their capacities

Fig. 2. RES of UTOPIA, dotted lines signify in-year variable fuels that are produced by either storage or second type technologies shown with

and

respectively.

42

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

Fig. 3. Comparison between UTOPIA results of RAISE (blue bars) and OSeMOSYS (red bars). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

based on peak power requirements and this is the point from which differences in capacities have arisen. To prove that, we account for this change in UTOPIA data file and this time, compare the same RAISE results with amended OSeMOSYS's in Fig. 4.,1 As we can see, results are closely reported with OSeMOSYS objective function reaching from $28.8 billion to $32.2 billion. 3.2. Verification with Temoa Temoa data input for UTOPIA can be obtained from www.te moaproject.org. In addition to some minor differences between Temoa data input for UTOPIA and what we run RIASE with,2 there exist 2 additional constraints in Temoa to make the model as close as possible to real world (Hunter et al., 2013). The first constraint is on end-use devices. Since they are not as manageable as other supply technologies, there may be an optimal solution in which an existing end-use device in a load region is completely off or is not utilized proportional to its capacity, what not favorable, because as

long as households possess a typical end-use device, they rely on it to meet their demand and model shouldn't decide on dispatching them in the way it does for the other technologies. In order to prevent this from happening, for a typical end-use technology t producing in-year variable demand f, there will be Eq. (1) for each load regionl in year y:

∑m Ot, m, l, y × S t, m, f , y AUD f , y × DDl, f , y

This change can be done by changing the value of parameter “TechWithCapacityNeededToMeetPeakTS” for these technologies from 0 to 1 in UTOPIA data file. 2 These are related to: right hand side of maximum total capacity constraint of RHE for years 1991 to 1999, right hand side of minimum total capacity constraint of E31, omitting maximum total capacity constraint for E51 and TXE, adding a new maximum total capacity constraint for TXD in which right hand sides are 0.6, 1.76 and 4.76 PJ/year for years 1990, 2000 and 2010 respectively. Furthermore, instead of fixed values, capital costs and fixed operation and maintenance costs for E01 do change in the planning period: the former decreases from 1400 to 1200 M$/GW and the later increases from 40 to 100 M$/GW/year.

(1)

The right hand side of Eq. (1) doesn't have l argument, meaning that it forces the contribution of device t for meeting demand of fuel (service) f to be equal for all load regions. The second constraint doesn't allow the rates of production of a user-defined set of power generating technologies named “baseload” to vary over the course of a season. Mathematically, for all technologies in “baseload” set, Eq. (2) is written for all load regions l defined in season s and year y:

∑m Ot, m, l, y LRl, y

1

= Constantt, y

= ConstantBaseloadRatet, s, y

(2)

Again, the right hand side doesn't have l index but rather a seasonal index s to use it for all load regions defined in season s. Adjusting RAISE to Temoa's data set for UTOPIA, as well as adding these two additional constraints with baseload set including E01, E21 and E31, we will have Fig. 5: RAISE objective function is $33.2 billion while Temoa's is $36.5 billion. This discrepancy along with minor divergence in capacity amounts coming up in last years of modeling period, are attributed

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

to difference in objective function formulation. Unlike lump sum investment procedure (modified by sinking fund depreciation method) adopted in RAISE (and also OSeMOSYS), Temoa assumes that investment costs are covered by loans which are going to be made up by annual future paybacks. As such, a somewhat different tendency towards capital investment is witnessed in RAISE and Temoa, approaching ultimate years.

4. Results and discussion In this section, we apply RAISE model for Iran's energy sector. 4.1. Overview of Iran's power generation sector The power industry in Iran, including power generation,

43

transmission and distribution facilities, is owned, operated and administrated by the Ministry of Energy (MOE) through its executive organizations that include TAVANIR (Power Generation and Transmission Management Organization) and the regional power companies. In 2012, Iran had 60,571 MW operational installed electricity generation capacities that produced 254,265 GWh of electricity most of which derived from subsidized fossil fuels (MOE, 2014). As a consequence of subsidized fuel prices, low capital costs, short construction times (often between 1 or 2 years), and country being master in manufacturing and installation of associated instruments, Open Cycle Gas Turbines (OCGT) with average efficiency of as low as 30%, have been the most in-hand option for meeting power requirements during past decade. Steam Power Plants (SPP) and Combined Cycle Power Plants (CCPP), with average efficiencies of 36% and 46% respectively, are the two other fossil

Table 3 Disparities between UTOPIA data file and Table 1 of Howells et al. (2011). Parameter source

UTOPIA data file Howells et al. (2011)

Residual capacity of TXG(PJ/year)

0 (1990:4.6,2000:1.5,2010:0)

Input to activity ratio TXG

TXE

RHO

TXD

E21

1 4.33

1 1.21

1.43 1

1 4.33

1 2.5

Capital cost of RL1

Variable cost of E70

0 100 M$/GW

0.4 M$/PJ 0

Fig. 4. Comparison between UTOPIA results of RAISE (blue bars) and amended OSeMOSYS (red bars). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

44

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

Fig. 5. Comparison between UTOPIA results of RAISE (blue bars) and Temoa (red bars). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

4.2. Iran's oil refining sector In 2012 the country owned 9 oil refineries with total refining capacity of 1.8 million barrels of oil per day. On the energy basis, 27.8% of output stream of refineries goes to fuel oil and other

60 50 Thousands MW

fuel consuming technologies dispatched for base load requirements. Boosting the capacity of SPPs in coming years is not among the MOE's policies. However, it is essential to retain part of fuel oil-fired steam generation capacity due to the limitations of natural gas supply in cold seasons. On the other hand, MOE's policy towards CCPPs is kind of an “if possible” approach meaning that they are constructed usually after OCGTs are constructed and if required budget for conversion of OCGT to CCPP is funded. Rest of the power production capacities including 9745 MW hydro and 106 MW wind, is dedicated to renewable energies (MOE, 2014). Other sources such as solar energy, hydrogen and biomass are still at the pilot scale. Fig. 6 illustrates historical evolution of power industry in the country. Within energy subsectors, highest level of CO2 emissions is from power sector. With the emission of 174,721 Gg CO2 in 2012, this sector has experienced an average annual growth rate of 6.1% between 2000 and 2012 (MOE, 2014). Grams of CO2 emitted per kWh of electricity produced in Iran are near 684 while this indicator for Organization for Economic Co-operation and Development (OECD) and world average is 433 and 565 respectively (International Energy Agency (IEA), 2013).

40 Other Hydro CCPP SCPP OCGT

30 20 10 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Fig. 6. Power generation capacities [MOE, 2014].

heavier hydrocarbons while only 16.5% of it consists of gasoline (IIES, 2012). These 9 refineries have been operating for many years and in 2006 the government launched enhancement plans aiming at increasing the share of light products of refineries. One such enhancement plan was on Arak refinery that became operational in 2012. The refinery was equipped with a RFCC unit shifting its capacity and gasoline production share form 217,000 barrel/day and 15.6% to 250,000 barrel/day and 34% respectively. Other than Arak refinery which has a RFCC unit and Abadan refinery which has a Fluid Catalytic Cracking (FCC) unit, other 7 refineries lack such relatively advanced molecular fraction technologies. Instead, they use hydrocracking and visbreaking processes to fracture heavy molecules into lighter molecules. An exception holds for

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

two small refineries of Lavan and Kermanshah which only rely on simlpe topping process for separating heavy molecules from lighter ones. During recent decade, growing demand for gasoline and country's constant refining capacities, have turned Iran into an importer of gasoline. This paradigm worked until international sanctions hindered provision of gasoline from external routes. With regard to refining sector the government is left with few choices: relying on imported products, building condensate refineries lacking units of cracking, boosting refining capacity by new refineries with conventional technologies for molecular fracture (hydrocracking and visbreaking) and finally boosting refining capacity by new refineries with expensive components for cracking.

45

Carbon Capture and Storage (CCS) technologies when a carbon cap imposed, they are also considered as possible upgrades of fossilbased power plants (red boxes in Fig. 7). Moreover all the current 9 refineries are included along with a condensate refinery and two types of refineries: “New Refinery T1” which is a typical refinery with hydrocracking and visbreaking units as a proxy for country's existing refining technology; and “New Refinery T2” which is a typical refinery equipped with advanced RFCC units and operates near the operational characteristics of Arak refinery. 4.3.1. Modeling horizon The model spans 24 years period between 2012 (as base year used for calibration) and 2035.

4.3. Reference energy system Reference energy system shown in Fig. 7 is used to provide a conceptual model of countrywide energy flow from resources down to final energy demands. The figure utilizes the shapes and conventions that were previously discussed. In order to examine whether or not it is rational to equip existing power plants with

4.3.2. Demand fuels At this stage exogenous demands are final energy demand and are imported from baseline scenario of (Eshraghi and Maleki, 2013) corresponding to the most likely development path of the country's socio-economic sectors with an average annual GDP growth of 3.5%, Table 4.

Fig. 7. Iran's reference energy system.

46

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

Table 4 Projected final energy demands (GWyr) [Eshraghi and Maleki (2013)].

Electricity Natural Gas LPG Jet fuel Gasoline Kerosene Diesel Fuel oil Coal Export

2012

2013

2015

2017

2019

2021

2023

2025

2027

2029

2031

2033

2035

22.42 157.10 3.16 2.07 23.60 5.26 33.32 7.00 2.70 176.89

23.14 161.03 3.22 2.22 24.30 5.25 33.99 7.20 2.80 177.06

24.6 169.38 3.32 2.56 26.00 5.20 35.36 7.60 3.10 177.42

26.07 178.17 3.42 2.92 27.70 5.15 36.79 8.04 3.40 177.77

27.61 187.42 3.53 3.33 29.60 5.09 38.27 8.49 3.80 178.13

29.23 197.15 3.63 3.78 31.60 5.03 39.82 8.98 4.20 178.48

30.86 207.38 3.75 4.28 33.70 4.93 41.43 9.48 4.60 178.84

32.59 218.15 3.86 4.82 36.00 4.84 43.10 10.02 5.10 179.20

34.91 229.47 3.98 5.17 38.50 4.94 44.84 10.59 5.60 179.56

37.28 241.38 4.10 5.54 41.10 4.99 46.66 11.19 6.10 179.92

39.82 253.91 4.23 5.93 43.80 5.05 48.54 11.83 6.70 180.28

42.55 267.08 4.36 6.35 46.80 5.11 50.50 12.50 7.40 180.64

45.48 281.20 4.50 6.81 50.02 5.17 52.54 13.21 7.90 181.00

Last row in Table 4 is the energy implications of the foreign revenues that are expected to come through energy exports. It is obtained by dividing expected future foreign revenues by the future unit prices of crude oil. Hence it would be the amounts of crude oil that must be taken out of the system boundaries if there were not any other alternatives for export. However since there are such other possible choices for export as natural gas and petroleum products, relative prices of these alternatives with respect to crude oil are considered to take the competition between export of any of these fuels into account. This way model decides on the export of a specific fuel based on its relative price of export, its upstream costs as well as the opportunity costs that it may get if used in other ways. 4.3.3. Load regions and load curves According to electricity hourly load curve for year 2012 (IGMC Webpage), each year were segregated into 8 season and each season were further segregated into 3 parts (for cold seasons) or 5 parts (for hot seasons when grid experiences its peak two times per day). In order to provide a more realistic picture of electricity generation, 4 types of power plants namely coal power plants, nuclear, SCPP and CCPP are supposed to serve as base load generating plants according to Eq. (2). Table 5 shows different load regions and corresponding demand distributions in each. 4.3.4. Intermediate fuels and technologies Finding reliable data for conversion technologies appears to be somewhat elusive. Nevertheless we could gather relevant data regarding technical and economic aspects of conversion technologies illustrated in Fig. 7 from various sources (MOE, 2014; Razavi and Eshraghi, 2014; TAVANIR, 2005; NIGC, 2012). After crosschecking available data from different sources we had, Table 6 was finalized for data entry. Moreover using data from annual energy balances (MOE, 2014), residual capacities of technologies for coming years were entered as in Table 7. 4.3.5. Resource constraints There are 4 fossil resources included in the model: offshore crude oil fields, onshore crude oil fields, non-associated gas fields and coal with total proven accessible stocks of 8248, 20,102, 43,207 and 278.1 GWyr (MOE, 2014). The proven potential of renewable energies for electricity production is examined to limit the development of the exploiting technologies. From studies carried out in sources (Department of Environment (DOE), 2010; Bozorgzadeh, 2012) it is estimated that total economic macro hydro-electric (larger than 100 MW) potential of the country is about 19,000 MW. Regarding wind energy recent surveys carried out in 45 suitable sites by Renewable Energy Organization of Iran (REOI) proves at least 7000 MW potent capacity for large wind farms (Alamdari et al., 2012; DOE, 2010). For other two remarkable renewable energies, solar photovoltaic and geothermal power

Table 5 Load Regions and associated demand distribution [IGMC Webpage]. Load region

January‐February‐ March April‐May

June

July

August

September

October

November‐ December

Electricity demand distribution

Name

Duration as fraction of year

Base Intermediate Peak Base Intermediate Peak Base Intermediate_day Peak_day Intermediate_night Peak_night Base Intermediate_day Peak_Day Intermediate_night Peak_Night Base Intermediate_day Peak_day Intermediate_night Peak_night Base Intermediate_day Peak_day Intermediate_night Peak_night Base Intermediate Peak Base Intermediate Peak

0.11458 0.0729 0.0625 0.0764 0.0694 0.0208 0.03819 0.0139 0.0069 0.0173 0.0069 0.03819 0.0139 0.0069 0.0173 0.0069 0.03819 0.0139 0.0069 0.0138 0.0103 0.03819 0.0139 0.0069 0.0138 0.0104 0.0382 0.0278 0.0174 0.0764 0.0486 0.0416

0.0889 0.0641 0.0594 0.0648 0.0652 0.0208 0.0417 0.0171 0.0089 0.0208 0.0087 0.0449 0.0185 0.0096 0.0224 0.0094 0.0479 0.0193 0.0100 0.0190 0.0149 0.0431 0.0177 0.0092 0.0175 0.0136 0.0352 0.0297 0.0192 0.0590 0.0424 0.0393

plants, maximum exploitation capacity are assumed to be 5000 MW and 1500 MW respectively (REOI, 2010). 4.4. Results This section presents Iran RAISE model results in two scenarios both of which were run by a 12% discount rate. 4.4.1. S1 scenario The first scenario called S1, is an indication of the most economically affordable prospect for development. In Fig. 8 recommended power technologies and their capacities are illustrated. The trend is based on the development of efficient combined cycle power plants such that in 2035 its contribution will be 65 GW generation capacity of total 112 GW. The reason lies with the efficiency and the relatively fair costs of CCPPs. This clearly defies past policies that were mainly about the least capital intensive option: gas turbine power plants. Although during 2028 to

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

47

Table 6 Techno-economic data for bottom-up modeling [MOE, 2014; Razavi and Eshraghi, 2014; TAVANIR, 2005; NIGC, 2012]. IC ($a/ kW)

FC ($a /kW)

VC ($a /kWyr)

Life (yr) Plant factor (%)

Fuel requirements parameter (RC)

Output fuel (s) Share in output (S)

Offshore oil extraction

681

6.81

15.4

40

90

1.05

Onshore oil extraction

524

5.24

10.2

40

95

1.03

Natural gas extraction Coal mining and processing Condensate refinery

570 380

5.7 3.8

9.8 4.8

40 30

90 95

1.05 1.5

Crude oil Associated gas Crude oil Associated gas Rich gas Coal

0.776 0.201 0.860 0.120 1 1

102

3.5

8.2

35

95

1

New refinery T1

142

9.5

14.2

35

95

1

New refinery T2

249

12.5

14.2

35

95

1

Natural gas refinery

42

5.3

11.9

35

95

1

Coal-fired power plant CCPP SCPP OCGT Diesel engine Nuclear power plant Hydro power plant Pump storage dam Wind turbine Solar PV power plant Geothermal power plant Elec. Trans. & Dist.

866.5 614 753 350 350 3100 1750 2100 1100 3500 b 2200 690

13.6 2.6 5.7 1.6 3.5 80 4.4 6.7 5.5 9 40 20

3.5 2.63 2.63 4.38 6.11 6.2 2.62 5.1 39.4 35 9.6 30

30 30 30 25 10 40 40 30 25 25 30 40

75 75 75 70 80 75 40 90 35 35 80 95

3 2.17c 2.8 c 3.41c 3.03 3 – – – – – 1.19

LPG Jet fuel Gasoline Kerosene Gas oil LPG Jet fuel Gasoline Kerosene Gas oil Fuel oil LPG Jet fuel Gasoline Kerosene Gas oil Fuel oil Natural gas Condensate Electricity Electricity Electricity Electricity Electricity Electricity Electricity Electricity Electricity Electricity Electricity Electricity

0.04 0.01 0.58 0.06 0.25 0.02 0.03 0.16 0.07 0.33 0.28 0.07 0.05 0.23 0.19 0.30 0.14 0.870 0.119 1 1 1 1 1 1 1 – 1 1 1 1

a b c

2005 US $. It is assumed that there will be a decline in IC such that in 2035 it will be 2500 $/kW. Power plant performance is assumed to be the same for all the operation modes, regardless of being fueled by either natural gas or liquid fuels.

2035, the model is kind of likely to choose slight amounts of OCGT, its general tendency towards OCGT is to maintain it in a relatively constant level. For other options except coal power plant, no new investment is seen and they evolve as their envisioned residual capacities. Proposed capacity for coal power plant exceeds the already 600 MW planned plant to reach 1800 MW in 2018. With this coal-fired power capacity compounded by required coal demand, all of the limited coal reserves (278.1 GWyr) will be consumed by 2035. Given the same patterns in electricity consumption, the peak power in 2035 will soar to around 78.8 GW. Fig. 9 shows how different technologies are dispatched in August 2035 when the peak load happens. Coal plants, nuclear, SCPP and CCPP are 4 types of power plants that are dispatched at first. If there were additional unmet demand, it would be on hydro plants and OCGTs to come into action. Fig. 9 also indicates that in both day and night peaks of August, storage dam discharges up to its available capacity (1 GW). This stored energy, as shown in the figure, comes from base load region at an average rate of 0.47 GW. Such technology mix for electricity production, leads to the fuel mix for power plants as shown in Fig. 10: Substitution of natural gas with liquid fuels could be another policy implication out of Fig. 10. But it requires the upstream capacity for natural gas extraction, refining and processing to boost.

Fig. 11 suggests that the chief share of natural gas supplied, is allocated to meet demand while only a comparatively smaller portion of that is for power sector and also export in the last years. The figure above suggests that natural gas supply will hit from 1000 Million Barrel of Oil Equivalent (MBOE) in 2012 to more than 2100 MBOE by 2035, what sounds to be viable and in accordance with government's official plans to double natural gas production capacity just in South Pars gas field3 by 2016 (POGC Webgage). It may be interesting to figure out where the optimality status is likely to take CO2 emissions of power sector. Fig. 12 clears the picture: CO2 emission under optimal solution is to grow at an average annual rate of 2.2% reaching about 207,000 Gg in 2035. This is in sharp contrast to the current emission patterns that has brought the country in the club of huge emitters with a 6.1% annual growth rate. If business as usual habits and therefore its corresponding growth rate persist in coming years, almost 680,000 Gg CO2 will be released in 2035 only from electricity production activities. The 3 Discovered in 1990 and located 62 miles offshore in the Persian Gulf, South Pars is one of the largest independent gas reservoirs in the world having a 24-phase development scheme. The Iranian portion is estimated to contain some 14 TCM of gas reserves and some 18 bn bl of gas condensates. This amounts to nearly 7.5% of the world gas reserves and about half of the Iran’s gas reserves (POGC Webpage).

48

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

Table 7 Residual capacities (GW) [MOE, 2014].

Coal-fired power plant Geothermal plant Hydro power plant Wind turbine Solar PV power plant Nuclear power plant OCGT SCPP CCPP Diesel engine Coal mining and processing Natural gas extraction Offshore oil extraction Onshore oil extraction Natural gas refinery Condensate refinery Tehran refinery Isfahan refinery Abadan refinery Arak refinery Bandar abas refinery Shiraz refinery Tabriz refinery Lavan refinery Kermanshah refinery Storage dam

2012

2014

2016

2018

2020

2022

2024

2026

2028

2030

2032

2034

2035

0.0 0.0 9.7 0.1 0.0 0.0 21.2 15.3 13.0 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

0.0 0.0 9.7 0.1 0.0 0.0 19.6 14.3 13.0 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

0.6 0.0 9.7 0.1 0.0 1.0 18.1 13.5 13.0 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

0.6 0.0 9.7 0.1 0.0 1.0 16.8 12.7 13.0 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

0.6 0.0 9.7 0.1 0.0 1.0 15.5 12.0 13.0 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

0.6 0.0 9.7 0.1 0.0 1.0 14.3 11.3 13.0 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

0.6 0.0 9.7 0.1 0.0 1.0 13.2 10.7 13.0 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

0.6 0.0 9.7 0.1 0.0 1.0 12.2 10.1 13.0 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

0.6 0.0 9.7 0.1 0.0 1.0 11.3 9.5 13.0 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

0.6 0.0 9.7 0.1 0.0 1.0 10.5 8.9 12.0 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

0.6 0.0 9.7 0.1 0.0 1.0 9.7 8.4 10.5 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

0.6 0.0 9.7 0.1 0.0 1.0 8.9 7.9 9.0 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

0.6 0.0 9.7 0.1 0.0 1.0 8.6 7.7 8.0 0.3 0.0 212.5 54.3 254.4 212.5 0.0 17.1 26.2 28.0 17.8 21.6 3.8 7.9 2.7 1.5 1.0

700 120

600

Sotrage Dam Hydro Power

80

GW

OCGT 60 CCPP SCPP

40

Coal Power Plant 20

Nuclear Power Plant

Million Barrel of Oil Equivalent

Other 100

500 Coal

400

Gas oil 300

Fuel oil Natural gas

200 100

0 2012

2015

2018

2021

2024

2027

2030

2033

2035

Fig. 8. Optimal trend for electricity production capacities.

0 2012

2015

2018

2021

2024

2027

2030

2033

Fig. 10. power plants fuel mix.

Fig. 9. Optimal activity of different technologies in August of 2035.

2035

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

Fig. 11. natural gas balance (negative values represent fuel consumption).

250

Thousands Gg

200 OCGT 150 SCPP CCPP

100

Coal Power Plant 50

0

2015

2020

2025

2030

2035

Fig. 12. CO2 emission trend from power plants.

structural change resulting from natural gas domination as power plants feedstock coupled with the policy shift to CCPPs, is the reason for decline in CO2 emission. The point here is that in case of Iran, optimality or in other words economic affordability is strongly tied to climate change mitigation. This policy implication contrasts the prevailing thought that CO2 mitigation in Iran's energy sector conflicts long term economic profitability. The rest of this section addresses optimal development of refining sector of Iran. Fig. 13 shows how different existing and potential future refineries contribute to gasoline production. The only constraint we imposed here is that for new refineries (both T1 and T2 as well as condensate refinery) the first year in which they can be operating, is set on 2016. From the figure we see that Shiraz and Isfahan refineries are

49

surprisingly off and gasoline import still stands until 2016 which is the year condensate and New Refinery T2 are allowed to be selected. This means that from an energy point of view, the output streams of Shiraz and Isfahan refineries are so inferior that selling domestic crude oil and importing gasoline is a better choice. Isfahan refinery converts only 13.5% of its energy input into gasoline and nearly 14% of energy content of output streams in these refineries is something other than marketable petroleum products. Besides, it is inferred from Fig. 13 that condensate refineries converting 58% of their input into gasoline, Table 6, and have far less capital investments in comparison to conventional oil refineries are the best option to meet growing petroleum products demands. These condensates come from country's huge rich gas treatment capacity and currently are shipped for export. The capacity of condensate refinery in Fig. 7 is limited by the availability of domestic condensates. The second choice to contribute to country's refining capacity is “New Refinery T2” that, as was explained, is a RFCC equipped refinery like Arak refinery. On the other hand there is no new capacity for “New Refinery T1”. This suggests that investment on any new refining capacity with the same technical structure of existing refineries should be avoided. 4.4.2. S2 scenario This scenario called S2, imposes a CO2 emission cap on all electricity production activities. The reason for formulating such a scenario comes from the impression that Iran is not going to remain a non-committed country as it was under Kyoto Protocol. It lies with the fact that Iran has been amongst top 10 emitting countries since 2006 and it may not be allowed by the international community to maintain its current CO2 emission growth rate anymore. This supposition particularly has become more serious after eighteenth session of the climate change Conference of Parties (COP 18) held in 2012 in Doha. There, all countries agreed to extend Kyoto Protocol by 2020 as well as to work toward a universal climate change agreement covering all countries as of 2020 and to find ways beyond the existing pledges to curb emissions so that the world can stay below the agreed maximum 2 °C temperature rise. Therefore it's likely to see some previously non annex-I countries have emission targets in the post Kyoto climate regime. Meanwhile the consequences and requirements of such targets, both financially and technically, have always been mixed up with ambiguities as to which technologies should be deployed and how much the least incurred costs would be. Having said that, S2

140 Gasoline Import

120

Condensate Refinery New Refinery T2

Million Liter per day

100

New Refinery T1 Tehran Refinery

80

Isfahan Refinery Abadan Refinery

60

Arak Refinery BandarAbas Refinery

40 Shiraz Refinery

20

Tabriz Refinery Lavan Refinery

0

Kermanshah Refinery

Fig. 13. Gasoline production from refineries (Shiraz and Isfahan refineries remain off by 2035.)

50

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

120

600

Geothermal

500 Million Barrel of Oil Equivalent

100

Wind 80

Hydro Power

GW

Sotrage Dam 60

OCGT CCPP

40 SCPP Coal Power Plant

20

400 Coal Gas oil

300

Fuel oil Natural gas

200

100

Nuclear Power Plant 0

0 2012

2015

2018

2021

2024

2027

2030

2033

2012

2035

2015

Hydro Power

Wind Turbine

Geothermal

2035 2034 2033 2032 2031 2030 2029 2028 -15

-10

-5

0 GW

5

Fig. 15. S2 capacities vs. S1.

10

2024

2027

2030

2033

2035

15

20

200

150 OCGT Thousands Gg

scenario is designed to seek the implications of imposing a cap aimed at maintaining CO2 emission levels for years after 2020, in the level of 2012. In Fig. 14 power generation capacities under a cap on CO2 emission are shown: Here unlike S1 scenario, OCGTs' capacity is diminishing while hydro, geothermal and wind power plants are not outsiders anymore. These renewable energy sources are suggested to be invested such that in 2035, 18.4 GW of hydro power, 1.5 GW of geothermal and 7 GW of wind need to be operating, almost up to their maximum exploitable capacity. Fig. 15 compares power generation capacities in S2 versus S1 for the years beyond 2028: It is seen that since power factor of intermittent sources such as hydro and wind power plants are far smaller than that of OCGT, the same demand is met with higher installed capacities. The fact that divergence between S1 and S2 scenarios emerges as late as 2028, suggests that S1 scenario or the so called optimal path, is by itself a mitigation scenario to some extent. As Figs. 16 and 17 show, there is a clear similarity in the trend of power plants fuel mix and CO2 emission of S2 scenario compared to S1 scenario. By developing renewable energies in S2 scenario, there will be a decline in natural gas consumption of power plants that reaches nearly 86 MBOE in 2035 and on a cumulative basis hits 260 MBOE in the period between 2012 and 2035. This amount of fuel saving (together with slight amounts of fuel oil and coal) prevents the emission of approximately 90,000 Gg of CO2 in the same period. However, investment on renewable energies in S2 scenario raises objective function by $303 M (2005 constant prices). It means that in such an environment, there must be a specified cost for one additional unit of CO2 saved, known as marginal abatement cost. These costs could be obtained by analyzing the shadow prices of the binding constraints which have exactly the same mathematical CCPP

2021

Fig. 16. Power plants fuel mix in S2 scenario.

Fig. 14. Optimal power capacities in S2 scenario.

OCGT

2018

SCPP 100 CCPP Coal Power Plant 50

0 2015

2020

2025

2030

2035

Fig. 17. CO2 emissions from thermal power plants in S2 scenario.

interpretation as the marginal abatement costs. In S2 scenario there was only nonzero shadow prices corresponding to emission cap constraint for years after 2030 (in another word CO2 emission cap constraint is nonbinding for the years preceding 2030). These costs, which were between $3.7 and $5.7 per tons of CO2, could be interpreted as a reflection of decarbonization costs in the power generation sector.

5. Conclusion and policy implications This paper was an initial effort in order to fill the gaps that exist in the field of energy policy. Traditional decision making paradigms which were mostly based on trial and error no longer work and must be put aside. These policies during decades have made Iran's energy system quite inefficient and aggravated the environmental situation. As the devastating effects of these policies became more apparent, the government was convinced that business-as-usual trends should be relinquished and instead, the country ought to step into an altered path. This was where the question “which are the bests in long-term?” was subsequently posed. As a result, arguments about development of a national master energy-environment plan whose core would be an optimization model has come into existence in academia and government in recent years. Using a mixed integer linear programming approach, RAISE model in its underlying methodology is similar to OSeMOSYS, Temoa and MESSAGE. However it is enriched with the structure and constraints that not only best suits country's specific bottlenecks, but also enjoys from enhanced numerical performance. In the first step ahead, we developed a single region model of RAISE for Iran's energy sector from 2012 to 2035. The model

H. Eshraghi, M.S. Ahadi / Energy Policy 89 (2016) 36–51

comprises existing and the most likely electricity producing technologies as well as oil refineries conceived to show up in the future. The analysis was conducted in two scenarios: S1 and S2. The former represents the least cost future path, while the latter is modified with an additional constraint that limits CO2 emissions from electricity production activities under 2012 level. Main findings are as follows:

51

sectors is subject to accessibility to a wide range of end-use data. For Iran, these data are even harder to achieve compared to supply-side data which are mainly government-controlled. Thus, to perform demand-side analysis there will be a need to field studies in the next steps.

Acknowledgment 1. Promotion of energy efficiency through investment on combined cycle power plants is the best option to meet base load in power sector. The optimal capacity for this technology is recommended to burgeon from 13 GW in 2012 to 65 GW in 2035. 2. However, the mostly relied on technology, open cycle gas turbine, is not a wise choice unless slightly for meeting intermediate and peak loads. Its optimal share in power technology mix drops from 36% in 2012 to around 20% in 2035. 3. A sharp substitution of liquid fuels with natural gas in power plants fuel mix needs to take place. This calls for a rather rapid expansion of country's upstream natural gas supply capacity. 4. Under optimal conditions (S1 scenario), CO2 emission annual growth rate in the modeling horizon, will be 2.2% which is considerably lower in comparison with 2000–2012 average figure, 6.1%. This apparently undermines the prevailing supposition that climate change mitigation deteriorates the economy and suggests that there is a strong synergy between them. 5. Despite model's tendency to invest on coal-fired power plants, they cannot turn into a game changer in the power sector due to the limitation of domestic coal reserves. 6. Power sector reacts to an imposed emission cap in S2 scenario by moving towards such renewable energies as hydro and wind power. The extra monetary burden of these investments is nearly $303 M (2005 prices). 7. Technical structure of country's refineries except for Arak and Abadan refineries is outdated. For two Shiraz and Isfahan refineries, economic value of inputs is higher than total economic value of outputs and therefore model suggests that they'd better go off. 8. Model results indicate a strong tendency towards condensate refineries. These refineries typically produce no fuel oil and convert half of their energy inputs into gasoline which is the most burdensome energy carrier to provide for country's growing demand. Thus it is important to stop exporting domestic condensate and use them as the feedstock for petroleum products production. However there is a limited amount of condensate available from gas processing units. The rest of the shortcoming in refining capacity should be met neither by imports nor by country's average refining figures, instead more capital intensive refineries equipped with RFCC units should be considered. The study however addresses only supply-side of Iran's energy system and end-users are not yet analyzed. Although the endusers have a great potential for decarbonization, modeling these

Here we want to express our deep gratitude to Dr. Amir Hossein Fakehi, for his consultations that really helped us conduct this study.

References Alamdari, P., Nematollahi, O., Mirhosseini, M., 2012. Assessment of wind energy in Iran: a review. Renew. Sustain. Energy Rev. 16 (1), 836–860. Bozorgzadeh, E., 2012. Hydropower development in Iran: vision and strategy. Compr. Renew. Energy 6, 253–263, Hydro Power, 2012. BP, 2013. Statistical Review of World Energy. Beyond Petroleum. London. Department of Environment (DOE), 2010. Iran Second National Communication of UNFCCC. United Nations Development Program (UNDP). Accessed in September via: 〈http://unfccc.int/essential_background/library/items/3599.php? rec¼j&priref¼7596#beg〉. EIA Webpage. Accessed in September via: 〈http://www.eia.gov/countries/countrydata.cfm?fips ¼ir〉. GNU Webpage. Accessed in September via: 〈http://www.gnu.org/software/glpk/〉. Howells, M., Rogner, H., Strachan, N., Heaps, C., Huntington, H., Kypreos, S., Hughes, A., Silveira, S., DeCarolis, J., Bazillian, M., Roehrl, A., 2011. OSeMOSYS: The Open Source Energy Modeling System: an introduction to its ethos. Struct. Dev. Energy Policy 39 (10), 5850–5870. K. Hunter, S. Sreepathi, J.F. DeCarolis, Nov 2013. Modeling for Insight Using Tools for Energy Model Optimization and Analysis (Temoa). Energy Economics. 40. 339– 349. ISSN 0140-9883. November 2013. IIASA (International Institute of Applied Systems Analysis), 2001. Model Message— Command Line User Manual. Accessed in September via: 〈http://www.iiasa.ac. at/Research/ECS/docs/ MESSAGE_man018.pdfS〉. International Energy Agency (IEA), 2013. CO2 Emission from Fuel Combustion Highlight. International Energy Agency, Paris, France. Institute for International Energy Studies (IIES), 2012. Iran’s Hydrocarbons Energy Balance 2011. Ministry of Petroleum. Iranian Grid Management Company (IGMC) Webpage, Access by: 〈http://www. igmc.ir/Default.aspx?tabid ¼ 477〉. Eshraghi, H., Maleki, 2013. Reshaping energy policy for sustainable development: curbing Iran’s carbon emission monster via renewable energies. Energy Sources Part B: Eco. Plan. Policy (in press). ID: 810313, /http://dx.doi.org/10.1080/ 15567249.2013.810313S. Makhorin, A., 2010. Modeling Language GNU MathProg. Access in September via: 〈http://www3.nd.edu/  jeff/mathprog/glpk-4.47/doc/gmpl.pdf〉. MOE, 2014. Energy Balance 2012, Power and Energy Planning Department. Ministry of Energy of I.R.IRAN, Tehran, Iran. NIGC, 2012. Annual Report of National Gas Company 1389, Deputy of Planning and development. National Iranian Gas Company, Tehran, In Persian. Razavi, M.S., Eshraghi, H., 2014. Modeling Iran’s Supply-side Energy System via MESSAGE V Energy Model. Institute for International Energy Studies, Tehran, Iran, In Persian. REOI, 2010. what do you know about renewable energies in Iran?. Renewable Energy Organization of Iran report collections, Tehran, Iran, In Persian. TAVANIR, 2005. Techno-economic characteristics of power plants. Planning division, Tehran, Iran, In Persian. WB (World Bank) Webpage, Accessed in September via: 〈http://data.worldbank. org/indicator/SP.POP.TOTL?page¼ 6〉.