Economic and environmental costs of replacing nuclear fission with solar and wind energy in Sweden

Energy Policy 112 (2018) 56–66

Contents lists available at ScienceDirect

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

Economic and environmental costs of replacing nuclear fission with solar and wind energy in Sweden

MARK

⁎

Sanghyun Honga, , Staffan Qvistb, Barry W. Brooka a b

Faculty of Science, Engineering & Technology, Private Bag 55, University of Tasmania, Hobart 7001, Australia Department of Physics and Astronomy, Uppsala University, Sweden

A R T I C L E I N F O

A B S T R A C T

Keywords: Nuclear power Renewable Sustainability assessment Alternative cases Optimization algorithm

Nuclear power is facing an uncertain future in Sweden due to political directives that are seeking to phase out this energy source over coming decades. Here we examine the environmental and economic costs of hypothetical future renewable-energy-focused cases compared with the current nuclear and hydroelectricity-centred mix in Sweden. We show that if wind and photovoltaics replace entire nuclear power while maintaining the current level of dispatchable backup capacity including hydroelectric power and peak gas power, 154 GW of wind power will be required and will generate 427.1 TWh (compared with the actual demand of 143.7 TWh) to reliably meet demand each hour of the year. As a consequence, the annual spending on electricity systems will be five times higher than the status quo. Increasing dispatchable power, increasing transmission capacities to other countries, and generating electricity from combined heat and power plants even when there is no heat demand, will together reduce the required capacities of wind and solar photovoltaic by half, but it will double the greenhousegas emissions during the combustion process. In conclusion, our economic and greenhouse-gas emissions analyses demonstrate that replacing nuclear power with renewables will be neither economic nor environmentallyfriendly with regards to the climate.

1. Introduction The electricity generation sector in Sweden has changed dramatically in the almost half-century since the first nuclear reactor started to operate (IEA, 2016a), with consumption more than doubling, from 65 TWh in 1971–150 TWh in 2014 (Swedish Energy Agency, 2015). However, unlike many other nations, where increasing fossil-fuel consumption has been required, Sweden is one of few countries (along with South Korea, Germany, France and Japan) that historically selected a nuclear power expansion program as the principal energy source for meeting demand growth (Cherp et al., 2016). In 2014 in Sweden, nuclear power generated 63.8 TWh and hydroelectricity 61.7 TWh, with the sum of all the other sources (wind, solar photovoltaic, biofuels and gas) amounting to 24.4 TWh (< 16.3% of the total electricity generation). Wind, in particular, provided a 7.5% share in 2014. The greenhouse-gas emissions of electricity and heat production in Sweden vary over time during the last three decades (SCB, 2016). The largest amount of greenhouse-gas emissions was 13.5 Mt CO2e in 1996, and the smallest amount was 6.4 Mt CO2e in 2015. The energy sectors including electricity, heat production and transport (excluding the emissions of industry and agriculture) emitted 29.1 MT CO2e in 2015.

⁎

In 1980, the Swedish government decided to phase out nuclear power following the nuclear referendum (WNA, 2017). However, the decision was revoked in 2010 (WNN, 2017), and the debate on the use of nuclear power is still ongoing. Recently, several Swedish political parties are arguing for a rapid transition away from nuclear in the Swedish electricity system (Qvist and Brook, 2015). Although it might seem possible to replace a unit of nuclear power generation with a unit of renewable generation, technical and economic barriers exist if the penetration of intermittent power grows large (Amoli and Sakis Meliopoulos, 2015; Huber et al., 2014). A large share of non-dispatchable renewables will require substantial backup generation capacity, as well as additional transmission capacity, if reliable power and grid frequency are to be maintained, unless Sweden attempts to rely on the remainder of the European grid to cover peak demand. We also noted that the phase-out of nuclear policies in Sweden and is not driven primarily by environmental or economic concerns, but they are based on political decisions or a lack of public acceptance (Qvist and Brook, 2015). Recently, a study evaluated the possible environmental consequences of replacing nuclear power generation with wind and lowercarbon fossil systems in Sweden (Wagner and Rachlew, 2016). This

Corresponding author. E-mail address: [email protected] (S. Hong).

http://dx.doi.org/10.1016/j.enpol.2017.10.013 Received 13 March 2017; Received in revised form 3 October 2017; Accepted 5 October 2017 0301-4215/ © 2017 Elsevier Ltd. All rights reserved.

Energy Policy 112 (2018) 56–66

S. Hong et al.

resources are not limited by geographical, legal or economic barriers in Sweden. Although there are other renewable options such as off-shore wind power and concentrating solar power, these options were excluded from this analysis on the grounds of economic competitiveness and low solar irradiation levels (Köberle et al., 2015). Although the electricity price of solar photovoltaics is also currently economically unviable in Sweden (Köberle et al., 2015), we included this source for its potential ability to complement periods when wind output is extremely low or zero, while backup power sources are fully exploited. Energy storage is likely to reduce required backup capacity and provide balancing services in many countries in the future (Palizban and Kauhaniemi, 2016). However, because of the country's extensive hydroelectric power resources, which provide system flexibility, it is unlikely to play a large role in Sweden (Huber et al., 2014). Based on the constraints listed above, the most cost-competitive mix for each energysystem condition, for a range of cases, was optimized using a simulated annealing algorithm. Currently, the Swedish electricity grid is divided into four geographic bidding areas (Swedish Energy Agency, 2015). The bidding areas, including SE1, SE2, SE3 and SE4, are connected via high-voltage transmission lines. Sweden is also linked to neighboring nations through high capacity transmission lines including Denmark, Norway, Finland, Lithuania, Poland, and Germany. The maximum capacities (GW) of the transmission lines are considered to be a physical limitation for intra-national and international transmission. We treated the international connection as primary ‘batteries’ that will supply electricity when it is required. The total amount of electricity that the other nations can supply is determined from the historical electricity inflow from neighboring nations between 2011 and 2013 (SMHI, 2015).

study examined the required capacity and operation strategies of backup power (natural gas), the role of energy storage and the operational conditions of hydroelectric power systems. The key finding of the study is that to replace 63.8 TWh of electricity generation from nuclear power (9 GW peak capacity) 22.3 GW of wind along with 8.6 GW of gas backup would be required, generating 64.8 TWh (with small spillage). The CO2 emissions of such the proposed scenario were estimated to be twice that of the status-quo generation and demand conditions. The Wagner and Rachelew (2016) paper was an important contribution that highlighted the difficulty in replacing nuclear power without increasing environmental impacts. However, the study did not consider five key aspects of system viability: (1) the transmission limitation between internal geographic bidding areas, (2) the economic costs of the transformation, (3) the seasonal water inflows for hydroelectric power, (4) the potential contribution of solar photovoltaic, if installed at a large scale, and (5) the role of distributed bioenergy if electricity and heat production processes are decoupled.1 The Carbon-Neutral Scenario (CNS) was published by the IEA in 2016 which was aiming a near carbon-neutral energy system in the Nordic region by 2050 (IEA, 2016b). The report concluded that wind could replace both fossil fuels (coal) and nuclear by increasing the wind share from 7% to 30% by 2050. Nordic countries would then export 53 TWh of electricity to the rest of Europe. Sweden would phase out all nuclear fleets by 2050 while increasing wind capacity up to 31.4 GW by 2050. The results of this report have two serious issues that should be addressed. First, the high penetration of variable renewables, mostly wind, will require technical (e.g., energy storage and backup power) and economic (e.g., demand response and market mechanisms) balancing mechanisms. Second, for the expansion of wind, the interconnection between Nordic countries and with other European countries will require substantial upgrades. In our analysis, we presented ‘alternative present’ cases based on the current electricity demand profile to reduce any uncertainties due to technological and economic changes, and assumptions for more complex modeling works. Our modeling is underpinned by transparent and objective optimization algorithms. First, we determined the electricity capacity and generation required to replace nuclear power generation with wind and solar power using public data from Sweden (SMHI, 2015; Swedish Energy Agency, 2015). We then estimated the annual electricity cost using the levelized cost of each electricity generation option, as reported in by the Swedish Government Energy Commission (Energikommissionen) (2016) and NREL (2016) reports, and the greenhouse-gas emissions of electricity based on the Intergovernmental Panel on Climate Change Guidelines for National Greenhouse Gas Inventories (IPCC, 2006). We also modeled a range of variations that could affect the electricity system in Sweden, such as the initial reservoir levels of hydroelectric power dams, the installed dispatchable backup capacity, and both the international transmission capacity and interconnector flow between bidding areas in Sweden. Our aims were to: (i) compare the greenhouse-gas emissions and electricity costs of the status quo and hypothetical renewable-energy-focused cases, and (ii) identify any alternative presents using renewables that avoid increasing annual costs or greenhouse-gas emissions.

2.2. Data We anchored our cases in actual hourly generation, consumption and weather data from each bidding area for the years 2014. We obtained the wind speed from 98 stations and solar irradiation values from 14 stations (SMHI, 2015). We then modeled the hourly wind and photovoltaic output profile of each site and averaged these to obtain the power output of each bidding area. We adjusted the hourly wind speed data measured from the height of weather stations to the wind power station's hub height (120 m) using a wind-gradient equation:

Vhub (h) = Vm *(hhub/ hm)a

(1)

where Vm = obtained wind speed, Vhub = estimated wind speed at the hug height, hhub= hub height, hm = measured height, and a = the Hellman exponent (Hong et al., 2013). The power output at hub height from the wind speed at the hub height was estimated using the wind speed and power equation. 3 P = 0.5*Vhub *πr 2*Cp *ρ

(2)

where p = wind power output, r = length of a blade (68 m), Cp = efficiency (40%), and ρ = density of air (1.23 kg m−3). For the calculation, we applied 3 m s−1 for the cut-in and 25 m s−1 for the cut-off speed of the wind turbines. The modeled wind power station reaches the maximum power output when the wind speed reaches 15 m s−1 (rated wind speed) (Vestas, 2016). The wind power output is normalized to between 0 (minimum) and 1 (maximum) to be multiplied by the installed wind capacity of each model. For solar photovoltaic, we used the power output information of the BP 280 W photovoltaic module, with normal operating assumptions (BP Solar, 2011). The hourly water inflow data for hydroelectric power plants were estimated from the weekly inflow data obtained from Svensk energi (2015a), spread equally across 168 h. The generation capacity and fuel consumptions are from Svensk energi (2015a) and IEA (2016a).

2. Methodology 2.1. Assumptions We included on-shore wind and photovoltaic solar power to replace nuclear power output. We also assumed that renewable energy 1 Due to economic reasons, combined heat and power plants in Sweden are operated during winter seasons to produce both electricity and heat. However, in our modeling, we assumed that combined heat and power plants can be operated either to produce both electricity and heat or to generate only electricity.

57

Energy Policy 112 (2018) 56–66

S. Hong et al.

Table 1 Capacity mixes of five future cases for electricity and heat supply in Sweden: (1) the Söder case, (2) the all-renewable case, (3) the 2x gas case, (4) the 4x gas case, and (5) the 2x transmission case. All cases have two different hydroelectric power variations: (1) high reservoir level (75%) or (2) low reservoir level (40%). Note that the electricity and heat production are coupled for the main cases. Capacity (MW)

Södera

Gas Dist. CHP Industrial CHP Hydro Condensingc Transmission

Optimized 1569 MW 3138 MW (2x) 6276 MW (4x) 4126 MW 3641 MW 3641 MW 3641 MW 1240 MW 1375 MW 1375 MW 1375 MW 16,150 MW installed with reservoir dam fill level at Jan 1. of either 75% (high) or 40% (low) 1498 MW 1498 MW 1498 MW 1498 MW 1x 1x 1x 1x

a b c

2x gasb

All renewables

4x gasb

2x transmissionb 1569 MW 3641 MW 1375 MW 1498 MW 2x

Although the Söder case, which sought to replace nuclear with renewables, is not peer-reviewed, they are frequently referenced in the Swedish energy systems debate. The all-renewable case is a baseline for increasing gas and transmission capacities. Condensing power uses fossil fuels such as coal, gas and oil for peak generation in Sweden.

model replicates the real generation profile closely.) The difference between the real generation data and the modeled result occurred mainly due to the market mechanisms. The model could be improved by adopting more realistic market mechanisms, but this is not the purpose of our study.

2.3. Modeled cases We generated a set of electricity-system models based on a range of conditions that represent different hypothetical energy mixes in Sweden (Table 1). First, we reproduced the most well-known study available for replacing nuclear power with renewables by Söder (2014) using our model (the “Lennart Söder” case). Although the Söder studies are not peer-reviewed, they are very frequently referenced in the Swedish energy systems debate.2 In Söder's (2014) case, 48 TWh of wind power and 12 TWh of solar photovoltaic replaced the annual electricity generation of nuclear power in Sweden (60 TWh). We modified the electricity generation of wind power to 51 TWh, and solar photovoltaic to 13 TWh to replace 64 TWh of nuclear power in 2014. As in Söders modeling, the capacity of distributed combined heat-and-power (CHP) was increased to 4126 MW, and industrial CHP to 1240 MW. We then sought to minimize the required capacity of gas power by optimizing the wind and photovoltaic capacities across the bidding regions. For the all-renewables case, we assumed that renewables, including onshore wind and solar photovoltaics, would replace the entire nuclear power generation. The all-renewable case made use of the current capacity of dispatchable power and transmission, with the capacities and locations of the wind and solar photovoltaics being then optimized to replace the nuclear generation. We also considered the impact of adding additional gas capacity (x2 and ×4 increase over current installed capacity), and additional regional and international transmission capacities (x2 current). To maximize the operational flexibility of distributed CHP, we assumed that electricity generation of distributed CHP could be either coupled (main models) or decoupled (Appendix A and B) with heat demand. For our main models, we assumed that the maximum power output of distributed CHP was limited to the maximum generation of each month from the actual generation data since 2001. We also considered different initial reservoir levels of hydroelectric power, based on wet (75% fill level) or dry (40% fill level) years. The median fill level in January (a starting month for our analysis) was slightly above 60% during 1960 and 2015. The maximum was > 80%, and the minimum was < 40% during the same period (Svensk energi, 2015b). Note that a wet year case was only of interest for comparative purposes; the actual system needed to be dimensioned for a dry year (with less available hydroelectric power). We compared each of the proposed models with the actual hourly electricity consumption and generation data in 2014 (Appendix D). For verification, we also ran a model with all the nuclear power plants currently available and 65% fill level of hydro reservoirs following the actual reservoir level in 2014. (Fig. D1 in Appendix D confirms that our

2.4. Energy system modeling We adopted an hourly generation for electricity generation modeling studies, which was widely accepted general approach (Elliston et al., 2012; Hong et al., 2013). We increased wind and photovoltaic capacities to meet the demand in 2014, while the other capacities were fixed. We, then used a simulated optimization approach to determine the most efficient wind and photovoltaic capacities across bidding areas in Sweden. We, then, used a simulated optimization approach to optimize the wind and photovoltaic capacities across bidding areas in Sweden. We used a simulated annealing algorithm, for optimization, to reproduce the previous work by Söder (2015) on “all-renewable” Swedish electricity cases and to assess the new alternative electricity cases; this approach has been shown to be well-suited for electricity network modeling (Hong et al., 2014). The simulated annealing algorithm is a global optimization process that can avoid suboptimal ‘local’ solutions (Hong et al., 2014). The process of simulated annealing involves generating a stochastic sample, modifying the sample, comparing the original and new samples, and selecting one of the samples based on selection probability to optimize the result (Fig. 1). In detail, the optimization algorithm produced a random initial capacity mix that can support hourly electricity demand in Sweden. A modified mix was then created based on the original mix, which also provided sufficient electricity for the demand each hour of the year. To obtain the modified mix random increases or decreases in the capacity of wind and solar photovoltaic of each bidding area were applied. The next step was to compare the generated sets, accepting whichever set was closer to the desired outcome. We defined a lower installed capacity of wind and solar photovoltaic as the desired direction that the optimization algorithm should follow. However, to avoid being trapped by a locally optimized answer, a less optimized solution could be selected with a certain probability, and a more optimized solution can also be rejected. The chance of accepting or rejecting sample reduces as the optimization proceeds over many iterations. This process is repeated a few times until the optimal result was found. The simulated annealing algorithm adopts the actual annealing process to decide the number of iterations. The ‘temperature’ which was the starting point of the algorithm was set to 1000 °C. The result when the temperature became 0 °C was the optimal result. As the process progressed, the temperature could either decreases or increases. For each iteration, if a more optimized solution was selected, the temperature decreased, otherwise the temperature increased. The increasing or decreasing temperature reduced as the optimization process progresses. The order of electricity generation options followed the levelized cost of

2 Frequently referenced by senior members of the ruling coalition government (Bäcklund, 2017; Nordin and Asenius, 2017; Nordin and Persson, 2017), many leading magazines (ElektronikTidningen, 2016; Ingenjören, 2016; KTH Magazine, 2015) and organizations (Svensk Vindkraftforening, 2017) used as a reference by other Swedish researchers in other, non-peer reviewed, but also widely spread studies (Siden et al., 2016), as well as mentions in innumerable high-level political debates.

58

Energy Policy 112 (2018) 56–66

S. Hong et al.

Table 3 Greenhouse-gas (CO2, CH4 and N2O) emissions during the combustion process of natural gas, biofuel and waste (IPCC, 2006) in kg TJ−1 of fuel consumption. kg TJ−1

Natural gas

Solid biofuel

Waste

CO2 CH4 N2O

64,200 3 0.6

100,000 30 4

91,700 30 4

balancing-service, additional transmission, curtailment and grid-connection costs, were not included in the calculations, meaning that the total cost of intermittent renewables in our cases is underestimated (OECD/NEA, 2012). The generation cost of the entire electricity system was calculated by dividing the total annual cost of all generation options by either the electricity consumption or generation. To calculate the annual generation cost of each option, we multiplied its annual generation with its specific levelized cost. We estimated the greenhouse-gas emissions during the combustion process based on the IPCC greenhouse-gas inventory guidelines (IPCC, 2006). We could not obtain the detailed fuel mixes of combustible power generation,3 including condensing power plants, district CHP, and industrial CHP. We, therefore, assumed that natural gas was the major source of condensing power, industrial CHP and gas power plants, and biofuels (primary solid) was the major source of distributed CHP based on fuel consumption information in IEA (2016a). Fuel consumption (TJ) of each power source required to generate a unit of electrical energy (MWh) was calculated from the primary energy supply for electricity generation and the actual electricity output (IEA, 2016a) (Table 3). Note that the values are not life-cycle greenhouse-gas emissions. We, also, estimated the life-cycle greenhouse-gas emissions for electricity generation based on the report published by Vattenfall (2012). Based on these figures for Sweden, peak-load gas is the highest greenhouse gas emitter (1269 kg CO2e MWh−1) followed by natural gas CHP and biomass CHP, while nuclear power emits the lowest amount at 5 kg CO2e MWh−1 (Table 4).

Fig. 1. A flowchart describing the optimization process for technology and capacity mix in the Swedish energy system, using a simulated annealing approach.

3. Results electricity (Energikommissionen, 2016). However, to introduce new wind and solar photovoltaic capacity into the grid, these sources were placed before hydroelectric power and other dispatchable systems.

3.1. Capacity and generation Compared with the energy-mix situation currently, all alternative cases required a higher installed generation capacity, due to the lower capacity factors of wind and solar photovoltaic compared to nuclear (Fig. 2). The electricity generation from hydroelectric power and other dispatchable backup systems is concentrated in the periods when wind and solar photovoltaic production is insufficient to meet demand. The assumed fill level in the hydroelectric power reservoirs has a strong effect on the required non-hydro renewable capacity. The total electricity capacity and generation of the Söder scenario for a dry year (63.0 GW and 166.4 TWh, respectively) is slightly higher than those of the equivalent case for a wet year (58.6 GW and 152.0 TWh, respectively). The Söder scenario for a dry year requires 10.3 GW of gas capacity that will generate 27.0 TWh. The required wind and solar photovoltaic capacities are 20.6 GW and 9.0 GW, respectively. Other backup systems, including hydroelectric power and distributed CHP, provide 61.1 TWh and 8.1 TWh, respectively. The Söder scenario in our optimization for a wet year requires 5.1 GW of gas capacity, generating 9.4 TWh, compared with 5.1 GW of gas generating 1.27 TWh in Söder's original work (2014); the capacity factor of gas generation for a wet year is lower (21%) than the one for a dry-year case (30%). The capacity factor of gas power in 2014 was <

2.5. Sustainability We estimated the levelized cost of electricity based on the values of the published information by Energikommissionen (2016) (Table 2). Since the report did not provide the levelized cost of electricity for natural-gas systems, we were forced in this case to apply the average value of European countries included in the IEA report (IEA et al., 2015). The grid-integration costs of power generation, including Table 2 Levelized cost of electricity by fuel source obtained from Energikommissionen (2016) and IEA (2015). LCOE (US$ MWh−1)a

Operational

Capital

Tax

Fees

Subsidies

Total

22.6 19.7 30.1 29.9 8.6

15.4 60.3 8.7 23.5 200.5

9.8 0.5 9.8 0.9 0.0

0.2 0.0 4.9 −1.4 0.0

0.0 18.7 0.0 18.7 98.4

99.4 48.1 99.1 53.6 71.6 307.5

b

Gas Hydro Wind Nuclear Biofuel Photovoltaic a

Exchange rate of 0.123 is applied. 1 SEK (Swedish Krona) = 0.123 USD (US Dollars). Levelized cost of electricity of gas is obtained from the average value of European countries (5% discount rate) from IEA (2015) due to the lack of a local source of data. b

3 Condensing power uses fossil fuels such as coal, gas and oil for peak generation in Sweden (Fortum, 2017).

59

Energy Policy 112 (2018) 56–66

S. Hong et al.

and 21.8% for a wet year under the all-renewables case compared with the capacity factor of hydroelectric power of the current status quo (2014) is 43.5%. If installed gas capacity is doubled (to 3.1 GW), the required wind capacity can be reduced by 64% (a dry year) and 87% (a wet year). However, quadrupling gas capacity from the all-renewable case does not reduce the required renewable capacity noticeably compared with the 2x gas cases. Although the influence is slightly lower than the 2x gas case, doubling the transmission capacity (2x transmission case) also reduces the required wind capacity, requiring 77.9 GW of wind capacity for dry and 36 GW for wet years. Paradoxically, the electricity generation from gas and distributed

Table 4 Life-cycle greenhouse-gas emissions of each generation option in kg MWh−1 (Vattenfall, 2012). kg MWh−1 Condensing Hydro Wind Nuclear Distributed CHP Gas PV Industrial CHP

1269 9 15 5 15 1269 44 503

Fig. 2. Generation capacity (top, in GW electric) and annual generation (bottom, in TWh) of modeled future energy cases for Sweden, including the current status quo, the Söder case, the all-renewables case and the other variations based on dry year (left) and wet year (right) assumptions. Distributed combined–heat-and-power plants (CHP) followed the historical electricity generation information.

CHP in a wet-year is higher than the equivalent case for a dry year excluding the Söder case, despite the higher share of hydroelectric power. Our model must increase wind and solar photovoltaic capacity during a dry year to cover the brief intervals when hydroelectric power cannot provide sufficient generation due to the low-fill levels. The increased share of wind and photovoltaic power, in turn, reduces the share of hydroelectric power for a dry year. Although the hydroelectric potential of the wet-year cases is much larger than the dry-year cases, the limited transmission capacity between the generation areas and the consumption area in Sweden is a significant obstacle during the peak demand (or low supply). The locally sited backup systems, therefore, cover the short-term peak demand for the dry-year cases (bottom left in Fig. 3) while providing intermediate electricity supply for the wet-year cases (bottom right in Fig. 3). As a result, the consumption level of Sweden (black line in Fig. 3) becomes much lower than the total generation profile in the renewables-focused cases. Fig. 3 illustrates a large amount of electricity from wind power that cannot be consumed in Sweden. This excess generation

1%. Hydroelectric power generates 68.3 TWh in this case, which is slightly higher than for the dry years and the status quo; 20.7 GW of wind and 9.9 GW of photovoltaic provide 64 TWh. Since the power system must be able to meet demand in both wet and dry years, the projected gas power plants (10.3 GW) required in a dry year will sit idle more often during a wet year, which may challenge the long-term economic viability of such a system. The installed wind-power capacity needs to be substantially higher if natural-gas backup is limited to the current level. The total installed capacities of wind power in the all-renewable cases are 178.3 GW for a dry year, and 153.8 GW for a wet year. The all-renewable case for a dry year requires 154.0 GW (427.1 TWh) of wind and < 0.1 GW (< 0.01 TWh) of solar photovoltaic, while the equivalent case for a wet year requires 105.0 GW (307.4 TWh) of wind and 24.6 GW (35.9 TWh) of solar photovoltaic. Electricity generation from hydroelectric power (17.3 TWh for a dry year and 30.8 TWh for a wet year) is significantly lower than the status quo, due to the high penetration of wind power. The capacity factors of hydroelectric power are 12.2% for a dry year

60

Energy Policy 112 (2018) 56–66

S. Hong et al.

Fig. 3. Generation profiles of four representative cases including the Söder cases (left) and the all-renewables cases (right), for a dry year (top) and a wet year (bottom), from Monday to Sunday during the first week of March. The black line is the electricity consumption, and area charts are electricity generation by source from all bidding areas. The total generation can be higher than the consumption profile due to the limited transmission capacity, and exceed generation of wind power. Even when wind and photovoltaic is sufficient to supply the demand, hydroelectric power may be needed due to the limited transmission capacity. Note that the area where the demand is higher than the generation is supported by the international transmission.

could be exported to neighboring countries, or alternatively, wind farms might be shut down to stabilize the national electricity grid (leading to an increase in the levelized cost from this source). Note that the electricity generation profiles of Fig. 3 are the sum of the electricity generation from all four bidding areas in Sweden. Therefore, the total sum of generation in the figure can be larger than the demand due to the limited transmission capacity between bidding areas. 3.2. Electricity costs In general, the status quo (including nuclear) has both the lowest annual costs and levelized cost of electricity compared with the alternative cases (Fig. 4). The levelized cost of generated electricity in the alternative cases is about twice that of the status-quo case, and up to five times higher when framed in terms of consumption. Note that the price does not include the additional balancing costs such as transmission costs and flexibility costs that will increase substantially if the intermittent renewable energy share increases. The Söder case for a dry year requires about doubling of annual electricity costs (US$ 15.9 billion) compared with the status quo (US$ 7.9 billion). The other cases require similar levels of the annual costs to the Söder case, except for the all-renewables case: for a dry year, the latter requires US$ 43.5 billion. Costs in the wet-year cases are lower the dry-year by between US$ 3.0 (the 2x gas case) and 7.3 billion (the 2x transmission case), excluding the Söder case. The key component leading to increased electricity costs are the large installed capacities of wind and solar photovoltaic of the alternative cases. The average-generation cost of electricity consumption is lowest for the status quo (US$ 55.5 MWh−1) and highest for the all-renewables case (dry year) at US$ 302.8 MWh−1. This is followed by the 2x transmission case (US$ 136.9 MWh−1) and the 2x gas case (US$ 127.7 MWh−1). For the wet-year cases, the all-renewables case requires

Fig. 4. Annual cost of electricity (right y-axis) and annual average electricity cost (left yaxis) of the cases for a dry year (top) and a wet year (bottom). Levelized cost of electricity is estimated by the total electricity consumption in Sweden (triangle) or the total electricity production (square).

US$ 303.0 MWh−1, whereas the Söder case (with substantial gas) requires US$ 98.2 MWh−1. The average-generation cost of electricity production of the alternative cases is lower, as spillage is ignored. The average-generation cost of electricity production raises the possibility of reducing losses by exporting spillage to neighboring countries. 61

Energy Policy 112 (2018) 56–66

S. Hong et al.

Fig. 5. Annual greenhouse-gas emissions of electricity (right y-axis) and greenhouse-gas emissions intensity of electricity (left y-axis) of cases for a dry year (top) and a wet year (bottom). Greenhouse-gas emissions intensity of electricity is estimated by the total electricity consumption in Sweden (triangle) or the total electricity production (square). IPCC general guideline data is used to calculate greenhouse-gas emissions during the combustion process.

Fig. 6. Annual greenhouse-gas emissions during the combustion process (Mt), annual lifecycle greenhouse-gas emissions (Mt) and annual electricity costs (billion USD, right) of cases for a dry year (top) and a wet year (bottom). Note that the lifecycle emissions become lower than the domestic emissions due to the high share of bioenergy.

However, even for the average-generation cost of electricity production, the current status-quo case has the lowest average-generation cost.

combustion process analysis, a case based on a wet year leads to higher greenhouse gases than for a dry year. The total lifecycle emissions are lower than the emissions during the combustion process in Sweden due to the lower lifecycle greenhouse-gas emissions of bioenergy (between 15 and 100 kg CO2e MWh−1) Vattenfall (2012) compared with the emissions during the combustion process (> 1400 kg CO2e MWh−1) (IPCC, 2006), which is the main source of distributed CHP. The current status-quo case emits 1.2 Mt CO2e during the life-cycle of electricity generation compared with 11.9 Mt CO2e for the emissions during the combustion process. The Söder case for a dry year emits 40.0 Mt CO2e, with the equivalent case for a wet year being 16.3 Mt CO2e. The allrenewable case for a dry year emits 7.8 Mt CO2-e and the wet-year equivalent is 8.3 Mt CO2e.

3.3. Greenhouse-gas emissions A case for a wet year emits more greenhouse gases than the equivalent case for a dry year due to the high penetration of dispatchable backup sources including distributed CHP and natural gas (Fig. 5). Despite the higher hydro potential, a wet-year case requires higher penetration of dispatchable power due to the limited transmission between bidding areas and the lower wind power capacity, as explained above. The level of the annual greenhouse-gas emissions during the electricity generation of the status-quo model is about 11.9 Mt (82.8 kg MWh−1). The emissions level is different with the actual greenhouse-gas emissions (between 6.4 and 13.5 Mt CO2e during the last three decades) because of the different fuel mixes of dispatchable generations such as distributed CHP. The Söder case for a dry year has far larger greenhouse gas emissions (46.1 Mt) and higher emission intensities for electricity generation (277.2 kg MWh−1) or consumption (321.0 kg MWh−1) compared with the other cases for a dry year due to the high reliance on backup sources, including natural gas and distributed CHP. The Söder case for a wet year emits 23.1 Mt (160.6 kg MWh−1 for consumption). By contrast, the all-renewables cases for a dry year and a wet year emit the lowest greenhouse gases which are 38.8 kg MWh−1 (5.6 Mt) for a dry year and 66.6 kg MWh−1 (9.6 Mt) for a wet year. However, we note that this is based on general IPCC guideline data for emissions of different energy sources, and is not specific to the Nordic region. Detailed life-cycle analysis, in particular, for the Nordic region by Vattenfall (2012), show significantly lower relative emissions from nuclear power than from renewable sources. If the Vattenfall (2012) numbers (life-cycle greenhouse-gas emissions) are used, the all-renewables case emissions are higher than the status quo (Fig. 6). Similar to the greenhouse-gas emissions during the

3.4. Sustainability comparison Fig. 6 compares the annual costs (right blue bar) and greenhousegas emissions (left red bar) of modeled cases. It is clear from this that the status quo delivers the lowest electricity costs. The Söder case failed to achieve either lower cost or lower greenhouse-gas emissions compared to the status quo. Although the all-renewable cases achieved slightly lower greenhouse-gas emissions (ignoring life-cycle emissions), its costs are almost five times higher. Overall, from our analysis, none of the cases replace nuclear with renewables achieved lower greenhousegas emissions and lower generation costs. 4. Discussion Our modeling study demonstrates that attempting to replace Swedish nuclear power with non-dispatchable renewable sources, such as onshore wind and solar photovoltaic (and natural gas as backup), will either: (1) massively increase greenhouse gas emissions, with a substantial increase in cost or (2) keep greenhouse gas emissions during 62

Energy Policy 112 (2018) 56–66

S. Hong et al.

Grid, 2016a, 2016b). Moreover, given that neighboring European countries are aiming at increasing the share of renewables (European Commission, 2016) and some countries such as Germany and Switzerland have decided to reduce or phase-out nuclear power (Renn and Marshall, 2016), the international import will become more difficult.

the combustion process low, but massively increase costs. Overall, both in terms of the generation mixes and the environmental consequences, our result (the Söder case) is aligned with the conclusions of previously published work (Wagner and Rachlew, 2016). Our result (the 2x transmission case) implies that the expansion of international and national transmission will reduce the required wind capacity, however the result is not strictly aligned with the IEA report (IEA, 2016b) since our model focused on providing ‘alternative present’ cases (a form of counterfactual), based on the national level analysis, whereas the IEA report modeled future cases for the entire Nordic region. First, we have confirmed that a unit (GWh) of electricity generation from nuclear energy in Sweden cannot be replaced by the same amount of electricity generation from intermittent renewable sources while maintaining (or constraining) electricity cost and greenhouse-gas emissions. Indeed, the emissions during the combustion process are forecast to increase four-fold (the Söder case), or else costs will inflate five-fold (the all-renewable cases), compared with today's status quo that includes a substantial contribution from nuclear energy. Secondly, expanding the transmission grid capacity is a critical factor if Sweden is to attempt to replace nuclear power with renewables. Expanded grid capacity will minimize the additional wind and solar photovoltaic capacities required. However, compared to the status quo, even a doubling of the current capacity will fail to provide additional economic or emissions-reduction benefits but will incur additional transmission installation and management costs (not included in our calculations). Although required wind and photovoltaic capacities can be decreased by decoupling electricity generation and heat production, enabling distributed CHP to become reserve backup (Appendix A and B), this will increase the greenhouse-gas emissions significantly and annual costs compared with today. Thirdly, under an all-renewables case, a large amount of excess generation needs to be absorbed by the network to avoid serious failure of the electricity system. However, given that most European countries are also expanding their share of renewable sources, seeking to export excess weather-dependent generation to other countries will likely become increasingly challenging (Kanellakis et al., 2013). Wind power production in the northern European region is highly correlated, meaning that at times of overproduction in Sweden, there might be no export market into which this electricity can be sold. Even today, power prices occasionally turn negative in northern European countries with high renewable penetration, such as Germany, where the market price of electricity fell to -€130 per MWh for brief periods in May 2016. National Grid, the system operator in the United Kingdom, paid £91.8 million to shut down wind power due to excess electricity supply and £216.82 million to rebalance the curtailment in FY 2015/16 (National

5. Conclusion and policy implications We have reviewed and modeled some plausible alternative electricity mixes for Sweden that seek to exclude nuclear power and optimized the resultant contributions using a simulated annealing algorithm. Based on available cost and greenhouse-gas emissions data, we have estimated the levelized cost of electricity and greenhouse-gas emissions of each modeled case. We conclude from this analysis that a large installed capacity of renewable sources focused on a substantial increase in wind and solar photovoltaic, cannot replace nuclear generation if the current electricity costs and greenhouse-gas emissions are to be maintained. A sustainable electricity mix which is simultaneously environmentally and economically robust will not be achieved by replacing nuclear power with renewables. Although nuclear power is not the only viable option for reducing greenhouse-gas emissions, it is clear that replacing nuclear with non-dispatchable renewables is neither economically viable nor environmentally friendly. In particular, for countries that already have substantial experience with the technology, nuclear power will be an economically viable and environmentally friendly option (Alonso et al., 2015; Baek, 2016; Hong et al., 2014; Lovering et al., 2016; Park et al., 2016). Given that the COP 21 agreement requiring the global efforts for reducing greenhouse-gas emissions, replacing nuclear power plants (which is a zero-emission generating source) that have decades of operational lifespan left with other zero or low-emission options, will require unnecessary investments that could be put to more productive and effective uses. Moreover, the electrification of transport sectors will increase electricity consumption from low-emission sources (Chen et al., 2015), necessitating an expansion in both nuclear power and renewables. To achieve its ambitious climate goals, our results imply that Sweden would be unwise to replace nuclear power with renewables. Instead, this nation needs to focus its future build effort on an energy pathway that is based on cost-effective and environmentally-friendly nuclear and renewable mix. Although our modeling work is applied to the case study of Sweden due to its current political relevance, our approach and results will be relevant to any other countries attempting to replace nuclear power with renewables.

Appendix A. Alternative cases with decoupled CHP See Appendix Fig. A1 Appendix B. Generation profile of alternative cases with decoupled CHP See Appendix Fig. B1 Appendix C. Greenhouse-gas emissions and annual cost of electricity See Appendix Fig. C1 Appendix D. Validation of the model See Appendix Fig. D1

63

Energy Policy 112 (2018) 56–66

S. Hong et al.

Fig. A1. Generation capacity (top in GW) and annual generation (bottom in TWh) of modeled future energy cases for Sweden, including the current status quo, the Söder case, the allrenewables case and the other variations based on dry year (left) and wet year (right) assumptions. Distributed combined heat and power plants did not follow the historical electricity generation information. Note that the all renewable case is a baseline case for increasing gas and transmission capacities.

Fig. B1. Generation profiles of four representative cases including the Söder cases (left) and the all-renewables cases (right) for a dry year (top) and a wet year (bottom) from Monday to Sunday during the first week of March. The black line is the electricity consumption, and area graphs are electricity generation by sources from all bidding areas. The total generation can be higher than the consumption profile due to the limited transmission capacity, and exceeded generation of wind power. Note that the area where the demand is higher than the generation is supported by the international transmission.

64

Energy Policy 112 (2018) 56–66

S. Hong et al.

Fig. C1. Annual domestic greenhouse-gas emissions (Mt, red) and annual electricity costs (billion USD, right) including O & M, capital and fuel costs of cases for a dry year (top) and a wet year (bottom).

Fig. D1. Generation profiles from the real consumption and generation data (left) from Monday to Sunday during the first week of March in 2014 (top) and a calendar year in 2014 (bottom); and generation profiles from our algorithm when applying the current generation capacity (right) from Monday to Sunday during the first week of March in 2014 (top) and a calendar year in 2014 (bottom). The difference between modeled wind and photovoltaic output and real power output, and market mechanisms makes the gap between the real-world data and our algorithm results. However, our model (right) follows the real-world generation profile reasonably well. evidence from the United States. Ecol. Indic. 66, 352–356. http://dx.doi.org/10. 1016/j.ecolind.2016.01.059. BP Solar, 2011. BP Solar 270 and 280W photovoltaic modules. Chen, F., Taylor, N., Kringos, N., 2015. Electrification of roads: Opportunities and challenges. Appl. Energy 150, 109–119. http://dx.doi.org/10.1016/j.apenergy.2015.03. 067. Cherp, A., Jewell, J., Vinichenko, V., Bauer, N., Cian, E.D., 2016. Global energy security under different climate policies, GDP growth rates and fossil resource availabilities. Clim. Change 136, 83–94. http://dx.doi.org/10.1007/s10584-013-0950-x. ElektronikTidningen, 2016. Hur Sverige kan ersatta karnkraften. Elliston, B., Diesendorf, M., MacGill, I., 2012. Simulations of scenarios with 100% renewable electricity in the Australian National Electricity Market. Energy Policy 45, 606–613. http://dx.doi.org/10.1016/j.enpol.2012.03.011. Energikommissionen, 2016. Memorandum on the financial conditions of the existing Swedish electricity production.

References Alonso, A., Brook, B.W., Meneley, D.A., Misak, J., Blees, T., Erp, J.B. van, 2015. Why nuclear energy is essential to reduce anthropogenic greenhouse gas emission rates. EPJ Nucl. Sci. Technol. 1, 3. http://dx.doi.org/10.1051/epjn/e2015-50027-y. Amoli, N.A., Sakis Meliopoulos, A.P., 2015. Operational flexibility enhancement in power systems with high penetration of wind power using compressed air energy storage. In: Power Systems Conference (PSC), 2015 Clemson University. Presented at the Power Systems Conference (PSC), 2015 Clemson University, pp. 1–8. http://dx.doi.org/10. 1109/PSC.2015.7101694. Bäcklund, B.-L., 2017. Miljöpartiet om Vindkraft [WWW Document]. Miljöpartiet Gröna. URL 〈https://www.mp.se/boras/just-nu/miljopartiet-om-vindkraft〉 (accessed 6 June 2017). Baek, J., 2016. Do nuclear and renewable energy improve the environment? Empirical

65

Energy Policy 112 (2018) 56–66

S. Hong et al.

OECD/NEA, 2012. Nuclear Energy and Renewables - System effects in low-carbon electricity system. Paris, France. Palizban, O., Kauhaniemi, K., 2016. Energy storage systems in modern grids—matrix of technologies and applications. J. Energy Storage 6, 248–259. http://dx.doi.org/10. 1016/j.est.2016.02.001. Park, S.-H., Jung, W.-J., Kim, T.-H., Lee, S.-Y.T., 2016. Can renewable energy replace nuclear power in Korea? An economic valuation analysis. Nuclear Engineering and Technology 48, 559–571. http://dx.doi.org/10.1016/j.net.2015.12.012. Qvist, S.A., Brook, B.W., 2015. Environmental and health impacts of a policy to phase out nuclear power in Sweden. Energy Policy 84, 1–10. http://dx.doi.org/10.1016/j. enpol.2015.04.023. Renn, O., Marshall, J.P., 2016. Coal, nuclear and renewable energy policies in Germany: from the 1950s to the “Energiewende. Energy Policy 99, 224–232. http://dx.doi.org/ 10.1016/j.enpol.2016.05.004. SCB, 2016. Statistics Sweden [WWW Document]. Stat. Cent. URL 〈http://www.scb.se〉 (accessed 21 February 2017). Siden, G., Hylander, J., Winkler, C., Stenqvist, C., 2016. 100 procent förnybar elenergi år 2020 – 2025 – 2030? Halmstad University, Halmstad. http://www.greenpeace.org/ sweden/Global/sweden/klimat/dokument/2016/100%20procent%20-%20tre %20scenarier.pdf. SMHI, 2015. SMHI Öppna Data | Meteorologiska Observationer [WWW Document]. URL 〈https://www.smhi.se/〉 (accessed 11 December 2015). Söder, L., 2014. På väg mot en elförsörjning baserad på enbart förnybar el i Sverige [WWW Document]. URL 〈http://kth.diva-portal.org/smash/get/diva2:727697/ FULLTEXT01.pdf〉. Svensk energi, 2015a. Statistik - Svensk energi [WWW Document]. URL 〈http://www. svenskenergi.se/Elfakta/Statistik/〉 (accessed 11 December 2015). Svensk energi, 2015b. Elåret 2015. [WWW Document]. URL 〈https://www. energiforetagen.se/statistik/elstatistik/elaret2/〉 (accessed 27 July 17). Svensk Vindkraftforening, 2017. Årets vindkraftpris till Prof. Lennart Söder. Swedish Energy Agency, 2015. Energy in Sweden 2015. Vattenfall, 2012. Life Cycle Assessment - Vattenfall's electricity generation in the Nordic countries. Vestas, V.-, 2016. V90-3.0 MW® [WWW Document]. URL 〈https://www.vestas.com/en/ products/turbines/v903_0_mw〉 (accessed 3 May 17). Wagner, F., Rachlew, E., 2016. Study on a hypothetical replacement of nuclear electricity by wind power in Sweden. Eur. Phys. J. Plus 131, 1–8. http://dx.doi.org/10.1140/ epjp/i2016-16173-8. WNA, 2017. Nuclear Energy in Sweden - World Nuclear Association [WWW Document]. URL 〈http://www.world-nuclear.org/information-library/country-profiles/ countries-o-s/sweden.aspx〉 (accessed 6 January 2017). WNN, 2017. Sweden reverses its nuclear phase out [WWW Document]. URL 〈http:// www.world-nuclear-news.org/newsarticle.aspx?id=24606〉 (accessed 6 January 17).

European Commission, 2016. Renewable energy directive - Energy - European Commission [WWW Document]. Energy. URL /energy/en/topics/renewable-energy/ renewable-energy-directive (accessed 12 December 2016). Fortum, 2017. Condensing power | Fortum [WWW Document]. URL 〈https://www. fortum.com/countries/in/production/condensing-power/pages/default.aspx〉 (accessed 29 May 2017). Hong, S., Bradshaw, C.J.A., Brook, B.W., 2013. Evaluating options for sustainable energy mixes in South Korea using scenario analysis. Energy 52, 237–244. http://dx.doi.org/ 10.1016/j.energy.2013.02.010. Hong, S., Bradshaw, C.J.A., Brook, B.W., 2014. Nuclear power can reduce emissions and maintain a strong economy: rating Australia's optimal future electricity-generation mix by technologies and policies. Appl. Energy 136, 712–725. http://dx.doi.org/10. 1016/j.apenergy.2014.09.062. Huber, M., Dimkova, D., Hamacher, T., 2014. Integration of wind and solar power in Europe: assessment of flexibility requirements. Energy 69, 236–246. http://dx.doi. org/10.1016/j.energy.2014.02.109. IEA, NEA, OECD, 2015. Projected Costs of Generating Electricity 2015. Organisation for Economic Co-operation and Development, Paris. IEA, 2016a. IEA - Sweden [WWW Document]. URL http://www.iea.org (accessed 12 May 16). IEA, 2016b. Nordic Energy Technology Perspectives 2016. Ingenjören, 2016. Ingenjören » Sidan kunde inte hittas. IPCC, 2006. IPCC - Task Force on National Greenhouse Gas Inventories [WWW Document]. URL 〈http://www.ipcc-nggip.iges.or.jp/public/2006gl/〉 (accessed 19 December 2015). Kanellakis, M., Martinopoulos, G., Zachariadis, T., 2013. European energy policy—a review. Energy Policy 62, 1020–1030. http://dx.doi.org/10.1016/j.enpol.2013.08.008. Köberle, A.C., Gernaat, D.E.H.J., van Vuuren, D.P., 2015. Assessing current and future techno-economic potential of concentrated solar power and photovoltaic electricity generation. Energy 89, 739–756. http://dx.doi.org/10.1016/j.energy.2015.05.145. KTH Magazine, 2015. Sveriges vag mot helt fornybar el. Lovering, J.R., Yip, A., Nordhaus, T., 2016. Historical construction costs of global nuclear power reactors. Energy Policy 91, 371–382. http://dx.doi.org/10.1016/j.enpol.2016. 01.011. National Grid, 2016a. Monthly Balancing Services Summary 2015/16. National Grid, 2016b. Future Energy Scenarios | National Grid. Nordin, L., Asenius, J., 2017. Svar till Lennart Back [WWW Document]. dt.se. URL 〈http://www.dt.se/opinion/insandare/replik-tilll-lennart-back〉 (accessed 6 June 2017). Nordin, L., Persson, J.P., 2017. Så kan kärnkraften avvecklas [WWW Document]. vlt.se. URL 〈http://www.vlt.se/opinion/debatt/sa-kan-karnkraften-avvecklas〉 (accessed 6 June 17). NREL, 2016. Transparent Cost Database | Transparent Cost Database [WWW Document]. URL 〈http://en.openei.org/apps/TCDB/〉 (accessed 12 May 2016).

66