Long-term scenario analysis of nuclear energy and variable renewables in Japan's power generation mix considering flexible power resources

Energy Policy 83 (2015) 169–184

Contents lists available at ScienceDirect

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

Long-term scenario analysis of nuclear energy and variable renewables in Japan's power generation mix considering flexible power resources Ryoichi Komiyama a,n, Yasumasa Fujii b a b

Resilience Engineering Research Center, The University of Tokyo, Hongo 7-3-1, Bunkyo-ku, Tokyo 113-8656, Japan Department of Nuclear Engineering and Management, The University of Tokyo, Hongo 7-3-1, Bunkyo-ku, Tokyo 113-8656, Japan

H I G H L I G H T S








Authors analyze Japan's long-term scenarios of nuclear and variable renewables. The analysis is performed by a dynamic optimal power generation mix model. Nuclear phase-out and carbon regulation quadruple power generation cost in 2050. Higher PV shares present challenges to make LNGCC a profitable ramp generator. Power saving is an economical option to treat an imbalance caused by PV output.

art ic l e i nf o

a b s t r a c t

Article history: Received 1 October 2014 Received in revised form 25 February 2015 Accepted 4 April 2015

This paper comprehensively analyzes an optimal deployment of variable renewables (VRs) with flexible power resources, such as electricity saving and rechargeable battery, in Japan's long-term power generation mix to 2050 under possible nuclear energy scenarios. The study is performed, employing a dynamic high time-resolution optimal power generation mix model which is formulated as a large-scale linear programming model. Simulation results show that both complete nuclear phase-out and carbon reduction by 80% in 2050 from 2010 encourage VR expansion in the country's power system and cause a quadruple increase of power generation cost at 2050 compared with that under current nuclear capacity and no carbon regulation policy; long-term cost reduction of VR energy system is necessary if VR is positioned as a mainstream for future sustainable power supply. Secondly, higher levels of VR integration decrease the capacity factor of LNG combined cycle (LNGCC), which implies the challenge to assure LNGCC serving as a remunerated ramp generator for VR intermittency. Finally, as an economically optimal solution, electricity saving serves as an important option to integrate massive VR and to treat a seasonal imbalance of its power output in an efficient way. & 2015 Elsevier Ltd. All rights reserved.

Keywords: Nuclear Variable renewable Power generation mix Rechargeable battery

1. Introduction After the Fukushima nuclear accident in Japan on March 2011, a lot of attention has been concentrated on renewable energy and energy efficient technology as alternative sources replacing nuclear energy. Recently, the Japanese government has encouraged electric utilities to enhance the proportion of renewable energy in their power grid and to ensure demand flexibility, for minimizing the dependency on nuclear power in the country's energy mix. Particularly after the implementation of feed-in-tariff (FIT) in 2012 by the government, the cumulative installed PV capacity rapidly increased from 7.3 GW in FY 2012 to 14.3 GW in FY 2013 (METI, n

Corresponding author. E-mail address: [email protected] (R. Komiyama).

http://dx.doi.org/10.1016/j.enpol.2015.04.005 0301-4215/& 2015 Elsevier Ltd. All rights reserved.

2014a). Moreover, PV capacity, which is certified to be built for the future and eligible for FIT, amounts to 65.7 GW as of March 2014 against 231 GW of total utility capacity in Japan (METI, 2014a; IEEJ, 2014). The effects of feed-in tariff (FIT) in Japan for renewable energy have been powerful, because the country's total installed PV capacity, in particular, almost doubled since the start of FIT in July 2012 and Japan has thus experienced a massive PV expansion. In July 2012, Ministry of economy, trade and industry (METI) set the purchase rate in FIT for PV as 42 yen per kW (METI, 2012a). However, since then, METI continuously revised down the rate to reflect lower PV system prices. Currently in FY2014, METI announced that the rate in FIT, for example, for household PV is revised downward to 37 yen per kW (METI, 2014b). By contrast, METI shows preferential renewable energy policies about the set of the rate to offshore wind and other renewables. In FY2014, for

170

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

instance, the rate of offshore wind is newly established, and that of most of other renewables remains unchanged (METI, 2014b). This suggests that the ministry expects the extensive installation of not only PV but also other types of renewable energy. Considering the current trend, PV installation is likely to grow in the power system. Until now, the government sets the target of PV cost reduction and attempts to maximize PV share in the power system from a longterm perspective (NEDO, 2004, 2009). Solar PV, as well as wind power, is expected to play a critical role for achieving a sustainable energy system. However, the power output of those variable renewables (VRs) is largely constrained by the intermittent availability of climate conditions such as solar insolation and wind speed. If VR penetration becomes predominant in the power grid, a key technical challenge is to maintain the adequate balance of power demand and supply all of the time. Under massive VRs' penetration, power system should effectively control their intermittency, aiming at cost-effectively integrating them with power grid and building an optimal power generation mix. For treating the VR variability, flexible power resources have attracted electricity market, which could facilitate the integration of VR into power system; those could contribute to accommodate higher levels of VR penetration and to guarantee system adequacy through those controllability, such as electricity saving, flexible power generator, tie-line interconnection and electricity storage. For instance, electricity saving, in the form of demand curtailment during periods of high electricity prices, could actively contribute to ensure system flexibility. By implementing those flexible measures, VRs are expected to be actively integrated in power systems. In addition, rechargeable battery is expected to serve as a measure to smooth the volatile output from variable renewable resources which causes voltage and frequency fluctuations in power grid network (Hadjipaschalis et al., 2009; Divya and Østergaard, 2009), besides conventional storage system such as pumped hydro power storage. Recently, for example, NaS (sodium sulfur) and Li-ion batteries have penetrated as commercial energy storage technology finding applications in electric grid support and in wind and PV power integration, bringing favorable benefits to power system control. Furthermore, the introduction of plug-in hybrid vehicle (PHEV) and electric vehicle (EV) has a possibility to serve as distributed storage system (Hu et al., 2014), contributing to manage the output variation of variable renewables. For planning a sustainable energy system, a high priority should be placed on analyzing the maximization of VR contribution in the nation's power system by taking advantage of those flexible power resources and on discussing the sustainable pathway of future power generation mix. Country such as Japan, where VRs begin to be deployed, should organize better policy to tackle those integration challenges, and this requires the employment of energy system model that can analyze VR installable potential for building a cost-effective and low-carbon power system. Based on the background, this paper analyzes future possible scenarios of nuclear energy and variable renewables in Japan's power generation mix to 2050, considering flexible power resources, and the authors attempt to identify the challenges associated with integrating large shares of VRs into power systems. The analysis is performed through the development of a dynamic high time-resolution optimal power generation mix model by extending the authors' previous manuscript (Komiyama and Fujii, 2014) which focuses on a single-period analysis (Komiyama et al., 2013a; Komiyama and Fujii, 2013b). The developed model here is a cost minimization model and allows us to assess a long-term optimal deployable pathway of VRs under nuclear energy scenarios in Japan, which can support to make VR investments required to address long-term energy security and climate change imperatives. The highlight of the model is that considered time

resolution is 10 min on 365 days of respective forecast horizon to 2050 as well as that flexible power resources are included, that is, quick load-following thermal plants, rechargeable battery and electricity saving. This manuscript comprehensively analyzes the best mix of short-term resources, such as rechargeable battery and electricity saving, and long-term resources, such as nuclear and variable renewables, in a long-term energy system planning, because the deployment of short-term resources influences the load curve profile and has a large impact on the long-term decisionmaking of the investment for nuclear and variable renewable (IEA, 2014; METI, 2010c). The advantage of a 10-min resolution consists in assessing the impact of short-cycle PV and wind variability on power generation mix in a detailed way. Until now, elaborate analysis on a power generation mix considering VR intermittency has been conducted using a specific energy model (Gallestey et al., 2002; Lund, 2005; Ummels et al., 2007; Xie and Ilic, 2008; Kiviluoma and Meibom, 2010; Troy et. al., 2010a, 2010b; Cheung and Rios-Zalapa, 2011; Denholm and Lund, 2011; Hug-Glanzmann, 2011; Hart and Jacobson, 2011; Keane et al., 2011; NREL, 2012; Schaber et al., 2012; Deanea et al., 2014; Palchak and Denholm, 2014). As far as the authors survey, however, many assessments have not yet been done, comprehensively discussing the positioning of large-scale VR deployment, together with those flexible power resources, in the country's long-term power generation mix with a higher temporal approach such as a 10-min. This paper is composed of following sections: Section 2 provides the methods including mathematical formulation of a dynamic high time-resolution optimal power generation mix model; Section 3 explains simulation results under respective nuclear and CO2 regulation scenario, and discusses a positioning of nuclear and VRs in the country's power generation planning; in Section 4, conclusions and political implications are explained and future direction of research is described.

2. Methods 2.1. Dynamic high time-resolution optimal power generation mix model Expected larger VR penetration raises the significance of energy system model that can analyze the contribution of VR in a longterm energy planning, and positioning of VR should be harmonized with a long-term development of the whole power system. In addition, as VRs considerably penetrates in the system, investments in system flexibility are necessary and the discussion about the optimal timing of those investments become required. In this paper, the authors discuss long-term scenarios of Japan's power generation mix to 2050 by developing a dynamic high time-resolution optimal power generation mix model in an annual time resolution at a 10-min under various technical constraints. The model is developed, employing a linear programming technique. The minimization of multi-period objective function, which is a summation of discounted annual facility and fuel cost from 2010 to 2050, allows us to identify the best mix strategy of power generation and capacity of power plants in the forecast time period and to estimate the cost-effective planning of power system flexibility to integrate large-scale VRs from a long-term perspective. In order to minimize total system cost under higher VR penetration, strategic energy policy for transforming the power grid is required, and the approach here is suitable to derive power system planning to support such a policy formulation. In the model, the number of constraints is 18 million, and that of endogenous variables (Table 1) is 7.6 million. An annual calendar year is described by 52,560 time segments (¼ 6 time points per hour  24 h per day  365 days per year). Exogenous variables

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

Table 1 Endogenous variables. J: discounted total cost from 2010 to 2050 ($) TCy: annual total cost in year y ($/year) CSj,y: annual cost of j-th storage facility in year y ($/year) CSAVEd,t,y: energy saving cost in day d, time t and year y ($/year) AvKi,d,y: available capacity of i-th power plant in day d, year y (GW) Chaj,d,t,y: input of j-th storage facility in day d, time t and year y (GW) Disj,d,t,y: output of j-th storage facility in day d, time t and year y (GW) DMaxi,d,y: maximum output level of i-th power plant in day d and year y (GW) Ki,y: capacity of i-th power plant in year y (GW) KS1j,y: kW capacity of j-th power storage facility in year y (GW) KS2j,y: kWh capacity of j-th power storage facility in year y (GWh) MtKi,m,y: unavailable capacity of i-th power plant in m-th maintenance schedule in year y (GW) Nki,y: newly constructed capacity of i-th power plant in year y (GW) Nks1j,y: newly constructed kW capacity of j-th power storage facility in year y (GW) Nks2j,y: newly constructed kWh capacity of j-th power storage facility in year y (GWh) Saved,t,y: electricity demand saving in day d, time t and year y (GWh) SSj,d,t,y: stored energy of j-th storage in day d, time t and year y (GWh) Suppi,d,t,y: suppressed output of i-th type of power plants in day d, year y and time t (GW) TChaj,y: annual total charged electricity of j-th type of storage facility in year y (kWh/year) Xi,d,t,y: output of i-th power plants in day d, time t and year y (GW) where i ∈{1: Nuclear, 2: Coal fired, 3: Gas CC, 4: Gas fired 5: Oil fired, 6: Biomass, 7: Hydro, 8: Geothermal, 9: PV, 10: Wind} j∈{1: Pumped-hydro, 2: Battery (stationary sodium–sulfur battery)} d∈{1, 2, …, D} D: number of the day per year (D ¼ 365 or 366) t∈{1, 2, …, T} T: number of the time steps per day (T ¼24n6¼ 144) y ∈{0, 1, 2, 3, Y} Y: number of the yearly steps (Y ¼4) yeary⊂{year0 ¼ 2010, year1 ¼2020, year2 ¼2030, year3 ¼ 2040, year4 ¼2050}

such as cost and technical assumptions are shown from Tables 2 to 5, based on Komiyama and Fujii (2014). PV cost (Table 4) is assumed to be incrementally decreased according to National Policy Unit (2011), and fuel price (Table 3) is set based on (IEA, 2013). It should be noted, however, that the effect of FIT on the cost of renewable energy such as PV is not explicitly considered in this paper. Regarding power demand curve, this paper adopts electricity load curve developed in Komiyama and Fujii (2014) and the shape of the load curve is assumed to remain unchanged in the

171

forecast period. Annual power demand is projected to grow towards 2050 at the same rate as that in electricity demand forecast of reference scenario by IEEJ (2013), that is, 0.5% per annum. In the assumption of the scenario, an annual average growth rate of GDP is assumed to be 1.4%, that of population,  0.4%, and that of GDP per capita, 1.7%, under the probable penetration of energy efficient technology. According to IEEJ (2013), alternative scenario, called as advanced technology scenario where advanced efficient technology will be expanded, shows that an electricity demand growth rate is  0.1% per annum. This decreasing trend of electricity demand will influence the required capacity of alternative power sources such as PV, wind and rechargeable battery. The assessment for the impact of decreasing electricity demand is considered as an important research topic for the future. Regional scope in this paper is the whole region of Japan as a single region, and the electricity market is assumed as monopoly market, although the model developed here is easily applicable to other region or country with the change of exogenous variables. Mathematical formulation of the model is described as follows: 2.1.1. Objective function Objective function is a discounted total cost occurred from 2010 to 2050; annual total costs among the representative years (decennial years from 2010 to 2050) is estimated through linear interpolation, and the objective function, as formulated in Eq. (1), is arranged as the integral of those interpolated annual costs through the forecast period. Total costs in the representative years are composed of annual fixed cost, fuel cost and electricity saving cost (Eq. (2)), and its formulation is based on Komiyama and Fujii (2014). Annual fixed cost is estimated as the multiplication of capital recovery factor, unit fixed cost ($/kW) and capacity (kW). Discount rate in this paper is assumed as 5 percent. First and second items in the right-hand side of Eq. (2) correspond to annual facility and fuel cost respectively; third item represents energy storage cost described in Eq. (3). The formulations of Eqs. (3) and (4) are based on Komiyama and Fujii (2014) as well; fourth item stands for energy saving cost as later shown in Eqs. (5) and (6).

Table 2 Exogenous variables of power plants. Type

Nuclear

Coal

LNG GCC

LNG ST

Oil

Biomass

Unit construction cost [$/kW] Life time [year] Annual O&M cost rate Maximum increase rate of output [1/h] Maximum decrease rate of output [1/h] Efficiency Own consumption rate Fuel cost [cent/specific unit] Heat content [kcal/specific unit] Carbon content [kg-C/specific unit] Seasonal peak availability Annual average availability Share of daily start and stop Minimum output level Specific unit

3500 40 0.040 0.00 0.00 1 0.04 1.67 860 0 0.85 0.85 0 0.3 kWh

2300 40 0.048 0.31 0.58 0.42 0.06 See Table 3 6139 0.618 0.85 0.78 0 0.3 kg

1200 40 0.036 0.82 0.75 0.57 0.02 See Table 3 13043 0.746 0.90 0.83 0.5 0.2 kg

1200 40 0.036 0.82 0.75 0.40 0.04 See Table 3 13043 0.746 0.90 0.80 0.3 0.2 kg

1900 40 0.039 1.00 1.00 0.39 0.05 See Table 3 9126 0.788 0.90 0.80 0.7 0.3 l

3500 40 0.048 0.31 0.58 0.20 0.13 12.25 3585 0 0.85 0.78 0 0.3 kg

Type

Hydro

Geothermal

PV

Wind

Unit construction cost [$/kW] Life time [year] Annual O&M cost rate Maximum increase rate of output [1/h] Maximum decrease rate of output [1/h]

8500 60 0.02 0.05 0.05

8000 50 0.01 0.05 0.05

See Table 4 17 0.01

2750 17 0.02

172

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

Table 3 Fuel price assumption.

2010 2020 2030 2050

Coal (cent/kg)

LNG (cent/kg)

Oil (cent/l)

9.8 10.4 10.8 10.8

50.1 64.7 65.6 70.2

45.8 66.3 71.0 79.5

Table 4 PV cost assumption.

PV Cost [$/kW]

2010

2020

2030

2040

2050

5700

4000

3340

2740

2120

Table 5 Exogenous variables of power storage technologies. Type

Pumped

Battery (NAS)

Unit kW construction cost [$/kW] Life time [year] Annual O&M cost rate Unit kWh construction cost [$/kWh] Unit non-durable material cost [$/kWh] Life cycle [times] Cycle efficiency Self discharge loss [1/h] Maximum kWh ratio to kW Maximum capacity factor

2400 60 0.01 10 0 1 0.7 0.0001 6 0.9

1200 15 0.01 40 160 4500 0.9 0.001 1 0.9

min J= Y−1

⎧ ⎪

∑ ⎨⎪∫

y=0

⎩

yeary + 1

(yeary + 1 − τ) × TCy + (τ − yeary) × TCy + 1 yeary + 1 − yeary

yeary

1

×

(1 + γ)

τ − 2010

⎫ ⎪ dτ ⎬ ⎪ ⎭ D

10

TCy =

∑ (gi × pfi, y

× Ki, y +

T

D

2

∑ ∑ pvi, y × Xi, d, t, y) + ∑ CSj, y d=1 t=1

i=1

+

(1)

j=1

T

∑ ∑ CSAVEd, t, y d=1 t=1

(2)

where, gi: annual fixed charge rate of i-th type of power plant (capital recovery factor), pfi,y: unit fixed cost of i-th type of power plants ($/kW), and pvi,y: unit variable cost of i-th type of power plants ($/kWh).

⎛ TChaj, y ⎞ ⎟⎟ CSj, y = (gs1j ⋅pfs1j ⋅KS1j, y ) + (gs2j ⋅pfs2j ⋅KS2j, y ) + ⎜⎜pfs 3j ⋅ cyclej ⎠ ⎝ D

TChaj, y =

(3)

T

∑ ∑ Chaj, d, t, y d=1 t=1

(4)

where, gs1j: annual fixed charge rate for power component of j-th type of storage facility, pfs1j: unit fixed cost for power component of j-th type of storage facility(cost for kW capacity, $/kW), gs2j: annual fixed charge rate for energy component of j-th type of storage facility, pfs2j: unit fixed cost for energy component of j-th type of storage facility (cost for kWh capacity, $/kWh), pfs3j: unit fixed cost for consumable material, such as electrode, electrolyte and separator of j-th type of storage facility($/kWh), and cyclej: maximum recharge times of j-th type of storage facility.

Fig. 1. Assumed demand curve. Reference price P0,t is assumed to be constant through a representative year. P0,t is assumed at 16 yen/kWh in 2010 by reference to Japanese average electricity price (IEEJ, 2014), although P0,t is normally necessary to be set equal to the shadow price of electricity demand in no CO2 regulation scenario (Fujii and Yamaji, 1998). Price elasticity is assumed as  0.5 which value is estimated as long-term elasticity in household sector by Murakami (2012).

Rechargeable battery cost depends on its power (kW) and energy (kWh) capacity. The cost information here is derived from METI (2010a, 2012b). Electricity saving has been regarded as one of key measures to integrate intermittent supply sources into power grid in an efficient manner. Hence, energy saving is endogenously included in the optimal power generation mix model by a top-down approach; total cost minimization endogenously determines the optimal introduction of electricity saving, considering its cost-effectiveness against supply-side technologies. This paper assumes typical demand curves where reference price is P0 and reference demand is loadd,t,y (Fig. 1). Based on this demand curve, the promotion of energy saving from reference point causes the increase in electricity price, and eventually, the integral of the demand curve from reference demand “loadd,t,y” to curtailed demand “loadd,t,y minus Saved,t,y” corresponds to energy saving cost (Fujii and Yamaji, 1998). Concerning the modeling of the electricity conservation cost, there are two methods. One is bottom-up approach and the other is top-down approach. In the bottom-up approach, the quantity of electricity saving is assumed to be calculated as the sum of electricity saving in each end-use device or facility, while in the top-down approach the electricity conservation cost or additional equipment purchase expenses can be estimated as the loss of consumers' utility as described in Fig. 1 (Fujii and Yamaji, 1998). This paper adopts the top down approach instead of the bottom-up approach, because, in the bottom-up approach, it is necessary to prepare tremendous amount of a nationwide data on energy usages in detail, and in many cases, it is almost impossible to appropriately calculate a nationwide quantity of electricity conservation with this approach due to lack of statistical information. Endogenous variable Saved,t,y is determined through an optimization, considering its cost competitiveness towards supply side technologies. The electricity saving cost, the fourth item in the right-hand side of Eq. (2), is mathematically formulated in Eqs. (5) or (6), depending on the value of price elasticity. Eqs. (5) and (6) are developed on the basis of Fujii and Yamaji (1998).

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

173

1

Saved , t , y

CSAVEd, t, y = ∫ 0

Saved , t , y

P (s)ds = ∫ 0

P0, t ×

⎛ loadd, t , y − s ⎞− β ⎜ ⎟ ds ⎝ loadd, t , y ⎠

⎧ β− 1 ⎧ ⎫ ⎪⎛ loadd, t , y − Saved, t , y ⎞ β ⎪ ⎪ β ⎟ ⎬ (5) − 1 ⎪ 1 − β × loadd, t, y × P0, t × ⎨⎜⎝ loadd , t , y ⎠ ⎪ ⎪ ⎪ ⎩ ⎭ ⎪ =⎨ ⎪ (β ≠ 1) ⎪ loadd , t , y ⎛ ⎞ ⎪ load ⎜ ⎟(β = 1) (6) d, t, y × P0, t × log⎝ load ⎪ d , t , y − Saved , t , y ⎠ ⎩ where, β: price elasticity ( ¼0.5 (Murakami, 2012)), P0,t: reference price ( ¼16 yen/kWh in 2010), Japanese average electricity price (IEEJ, 2014), and loadd,t,y: power load in day d, time t and year y (GWh). Regarding price elasticity β, this paper assumes 0.5 (Murakami, 2012), although the actual demonstration of demand response (critical peak pricing, CPP) in Japan reports that the value is less than 0.5 (Asano and Yamaguchi, 2014), for example. In this paper, reference electricity price P0,t is assumed to be changed to 2050. As a first step, in each given scenario, annual power generation cost (yen per kWh) to 2050 is calculated by the model without considering electricity saving. That is, the power generation cost is estimated by the model without considering Eqs. (5) and (6) in objective function (Eqs. (1) and (2)) and then, reference power price P0,t to 2050 (in 2010, 16 yen per kWh) is estimated by using the annual growth rate of those derived power generation costs to 2050. Secondly, using the estimated reference prices, optimal power generation mix to 2050 is calculated again by the model considering electricity saving, that is, considering Eqs. (5) and (6).

Fig. 2. Shutdown occurrence rate for plant maintenance in Japan. 4 types of the schedules are considered (Komiyama and Fujii, 2014).

(c) Constraints on available capacity and annual maintenance schedules: Parameter urm,d is shutdown occurrence rates of plants in day d due to maintenance of m-th schedule. This paper assumes four time profiles of annual maintenance schedule as illustrated in Fig. 2 (Komiyama and Fujii, 2014). The parameter upai is annual average availability of i-th type of thermal power plant, and uppi corresponds to seasonal peak availability. Following equations are developed by Komiyama and Fujii (2014). 4

AvKi, d, y +

∑

(urm, d × MtKi, m, y) = Ki, y

(i = 1, 2, ... , 6)

m=1

(11)

4

∑

(ursm × MtKi, m, y) = (1 − upai) × Ki, y

m=1

(i = 1, 2, ... , 6), D

2.1.2. Constraints (a) Power demand and supply balances: Following equation is based on Komiyama and Fujii (2014) and newly includes electricity saving. 10

2

∑ Xi, d, t, y + ∑ (Disj, d, t, y − Chaj, d, t, y) = loadd, t, y − Saved, t, y i=1

j=1

(7)

y

∑ remi(y, j) × Nki, j

(i = 1, 2, ... , 10) (8)

j=0 y

KS1i, y = ks1inii, y +

∑ remsi(y, j) × Nks1i, j

(i = 1, 2)

j=0

∑ urm, d

(12)

d=1 4

∑

(urm, d × MtKi, m, y) ≥ (1 − uppi ) × Ki, y

(i = 1, 2, ... , 6)

m=1

where, loadd,t,y: electric load in day d, time t and year y. (b) Constraints on capacity expansion and intertemporal capacity balances: Total accumulated capacity of each plant is determined, considering newly constructed and retired plants over a forecast horizon.

Ki, y = kinii, y +

ursm = (1/D) ×

where, urm,d: occurrence rate of plant shutdown in day d due to maintenance of m-th schedule (Fig. 2), upai: annual average availability of i-th type of power plant(“Annual Average Availability” in Table 2), and uppi: seasonal peak availability of i-th type of power plant(“Seasonal Peak Availability” in Table 2). (d) Available capacity constraints of considered technologies: In PV and wind, the output of unit capacity ufi,d,t in Eq. (16), a yearly profile of PV and wind outputs at a 10-min in a year, are given as later shown in Fig. 3 and Fig. 4, using Japanese Meteorological database (Japan Meteorological Agency, 2007), which offers observed data of solar insolation and wind speed on a 10-min resolution. following equations are included in the model, based on Komiyama and Fujii (2014).

(9)

y

KS2i, y = ks2inii, y +

∑ remsi(y, j) × Nks2i, j (i = 1, 2) j=0

(10)

where, kinii,y: remaining initial capacities of i-th type of power plant in year y, ks1inii,y: remaining initial power (kW) capacities of i-th type of energy storage technology in year y, ks2inii,y: remaining initial energy (kWh) capacities of i-th type of energy storage technology in year y, remi(y,y′): lifetime matrix of i-th type of power plant in year y, y′ (if the plant remains in year y, the value is “1”; otherwise, “0” ), remsi(y,y′): lifetime matrix of i-th type of energy storage technology in year y, y′ (if the storage facility remains in year y, the value is “1”; otherwise, “0”).

(13)

Fig. 3. PV output profile of Japan in 365 days at a 10-min interval.

174

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

Xi, d, t + 1, y ≤ Xi, d, t, y + increasei × {(1 − λ i)Xi, d, t, y + λ i⋅uci, d, t⋅Ki, y} (i = 7, 8)

(25)

Xi, d, t + 1, y ≥ Xi, d, t, y − decreasei × {(1 − λ i)Xi, d, t, y + λ i⋅uci, d, t⋅Ki, y} (i = 7, 8)

Fig. 4. Wind output profile of Japan in 365 days at a 10-min interval.

Xi, d, t, y ≤ AvKi, d, y

(i = 1, 2, ... , 6)

Xi, d, t, y ≤ uci, d, t × Ki, y

(14)

(i = 7, 8)

ufi, d, t × Ki, y = Xi, d, t, y + Suppi, d, t, y

(15)

(i = 9, 10)

where, increasei: maximum output increase rate per unit time of ith type of power plant, decreasei: maximum output decrease rate per unit time of i-th type of power plant, λi: capacity weight in the present output level (¼0.5 for default setting in this study). (h) Constraint on minimum output threshold of thermal power plant: Eq. (27) shows that thermal power plant operates more than a minimum threshold, excluding thermal plant with Daily Start and Stop (DSS) mode. For sharp demand variations within a day, thermal plant with DSS mode generates electricity under rapid heat-up and cool-down cycle. Following equations are developed by Komiyama and Fujii (2014).

Xi, d, t, y ≥ (DMaxi, d, y − dssi⋅AvKi, d, y)⋅moli

(27)

DMaxi, d, y ≥ Xi, d, t, y

(28)

DMaxi, d, y ≥ Xi, d + 1, t, y

(29)

(16)

Chaj, d, t, y + Disj, d, t, y ≤ us1j, d × KS1j, y

(17)

SSj, d, t, y ≤ us2j, d × KS2j, y

(18)

where, uci,d,t: availability factor of i-th type of power plant (hydro and geothermal) in time t and day d, ufi,d,t: availability factor of i-th type of power plant (PV and wind) in time t and day d, us1j,d: kW availability factor of j-th type of storage facility, and us2j,d: kWh availability factor of j-th type of storage facility. (e) Constraints on upper and lower installable capacity: Installable plant capacity is regulated under its maximum and minimum deployable limitation (Komiyama and Fujii, 2014).

Ki, y ≥ K0i, y, Ki, y ≤ Kupper, i, y

(19)

KS1j, y ≥ KS10j, y , KS1j, y ≤ KS1upper, j, y

(20)

KS2j, y ≥ KS20j, y , KS2j, y ≤ KS2upper, j, y

(21)

where, K0i,y, KS10j,y, KS20j,y: existing capacity, and Kupper,i,y, KS1upper, KS2upper,j,y: capacity upper limit. (f) Capacity reserve constraints for power supply reliability: To maintain electricity supply reliability, reserve capacity is assured on the basis of following equation (Komiyama and Fujii, 2014).

where, DMaxi,d,y: maximum output level of i-th type power plant in day d and dþ 1 at year y, dssi: share of daily start and stop operation (DSS) of i-th type of power plant, and moli: minimum output level ratio of operation of i-th type of power plant. (i) Charge and discharge balances of energy storage technology: Eq. (30) describes the power charge and discharge balance for stored electricity in energy storage technology like pumped-Hydro and rechargeable battery (Komiyama and Fujii, 2014). effstorage in Eq. (30) stands for the round-trip efficiency of rechargeable battery. charge cycle efficiency and discharge cycle efficiency of the battery are equal to the square root of the round-trip efficiency effstorage respectively.

SSj, d, t + 1, y = (1 − sdj) ⋅SSj, d, ty + ( effstorage, j Chaj, d, t, y −

6

8

2

i=1

i=7

× KS1j, y

j=1

≥ (1 + δ) × (loadd, t, y − Saved, t, y)

(22)

where, d: reserve margin (¼ 5–8%). (g) Constraints on load following capability of power plants: Following equations reflect load following capability of each power plant. more detailed explanations are available in Komiyama and Fujii (2014).

Xi, d, t + 1, y ≤ Xi, d, t, y + increasei × {(1 − λi)Xi, d, t, y + λi⋅AvKi, d, y} (i = 1, 2, ... , 6)

(23)

Xi, d, t + 1, y ≥ Xi, d, t, y − decreasei × {(1 − λi)Xi, d, t, y + λi⋅AvKi, d, y} (i = 1, 2, ... , 6)

(24)

1 effstorage, j

Disj, d, t, y) (30)

× Tw

j,y,

∑ AvKi, d, y + ∑ uci, d × Ki, y + ∑ us1j, d

(26)

(j) Available capacity constraint of battery technology: Eq. (31) is based on Komiyama and Fujii (2014) as well.

SSj, d, t, y ≤ mstorage, j × us1j, d × KS1j, y

(31)

where, sdj: self-discharge rate of j-th type of electricity storage, effstorage,j: round-trip efficiency of j-th type of electricity storage, mstorage,j: Energy storage capacity per generation capacity of j-th type of electricity storage, Tw: step width of the unit time (10 min). (k) CO2 emissions regulation: Eq. (32) is set to regulate CO2 emissions, based on Komiyama and Fujii (2014). 5

⎛

i=2

⎝

D

T

⎞

∑ ⎜⎜carboni × ∑ ∑ Xi, d, t, y⎟⎟⋅Tw ≤ CO2uppery d=1 t=1

⎠

(32)

where, carboni: Carbon intensity of fuel of i-th type power plant, and CO2 upper: Upper limit of CO2 emissions. Concerning installed capacity of coal-fired power plant, the maximum upper limit is assigned on the basis of METI (2010b), reflecting on future tighter environmental regulation. However, as nuclear power plants in Japan remain shut down after the

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

Fukushima nuclear accident, Japanese utilities actually began planning the new operation of coal-fired power plants in order to reinforce the power supply capability. For example, Tokyo, Kansai, Chubu, Kyushu and Tohoku electric power companies plan to add 1.0 GW (TEPCO, 2014a), 1.2 GW (KEPCO, 2014), 1.0 GW (CHUDEN, 2014), 1.2 GW (KYUDEN, 2014) and 0.6 GW (Tohoku Electric Power Company, 2014) of coal-fired power plant in their power generation mix respectively. But it should be noted that the influence of those new coal-fired plants on CO2 emissions is not explicitly considered in this paper. Maximum installable wind capacity is set at 63 GW, considering its onshore resource potential in Japan. The yearly profile of average outputs of PV and wind in a 10-min, corresponding to ufi,d, t in Eqs. (16), are estimated, as shown in Figs. 3 and 4 (Komiyama and Fujii, 2014), using the Japanese meteorological database (Japan Meteorological Agency, 2007). This database provides observed data of solar insolation and wind speed etc. on every 10-min. Estimated PV outputs of unit capacity in Japan during the year 2007 are shown in Fig. 3, which indicates that PV output intensity tends to be higher in summer and lower in winter, while the intensity of wind power output (Fig. 4) becomes higher in winter and lower in summer. However, it should be noted that those VRs' outputs are assumed to be perfectly foreseeable in the planning period and the uncertainty of the VRs' outputs is not considered in this paper. In addition, wind and PV outputs actually show a variety of variability profiles depending on weather conditions, although this paper deals with a single pattern of those outputs. The required capacity of rechargeable battery and ramp generator such as LNGCC will change, depending a various type of wind and PV outputs. Hence, the assessment for the impact of those output patterns on optimal power generation mix is considered as an important future work. 2.2. Scenario assumption about nuclear and CO2 regulation policy This paper investigates Japan's long-term power generation mix under multiple nuclear and CO2 regulation scenarios. In each scenario, an optimal power generation mix is identified, employing the model developed here. 2.2.1. Nuclear After the Fukushima nuclear accident, the Japanese government has not yet published the specific deployable target of nuclear power plant in the country's power system, although the possible fractions of nuclear in the power generation mix was roughly discussed (METI, 2012c). Since the governmental nuclear outlook is currently not available, Fig. 5 describes four possible scenarios about Japan's nuclear power plant capacity towards 2050, which enables us to analyze the impact of nuclear policy on the country's best mix of power generators and flexible power resources. Before the Fukushima, the government planned to build additional 14 nuclear power plants by 2030, according to Japan's national energy policy (METI, 2010b) officially formulated before the Fukushima. Although current outlook on nuclear remains quite uncertain, “Status quo” scenario assumes the decommissioning of Fukushima nuclear power station and the continuous operation of all existing plants to 2050. “Decommission” case supposes the decommissioning of nuclear power plant in excess of 60 years, 50 years and 40 years in operation as well as the demolishment of the Fukushima nuclear power station. Nuclear capacity in 2050 is almost zero, one-fifth and half from the capacity in 2010 respectively on decommission after 40, 50 and 60 years in operation. Until 2030, nuclear capacity in “Status quo” is almost equivalent to that in “Decommission after 60 years”. Under the four scenarios regarding the country's nuclear policy, this paper consistently

175

Fig. 5. Japan's long-term nuclear scenarios to 2050, consisting of Status quo, Decommission after 40 years, Decommission after 50 years and Decommission after 60 years in operation.

investigates the impact of nuclear power development on possible power generation mix in Japan. It should be kept in mind, however, that this paper only deals with the nuclear front-end scenario, not the back-end scenario including the outlook of spent fuel reprocessing facility, fast breeder reactor and nuclear waste disposal, though Japan Atomic Energy Commission evaluated the political options associated with the back-end issues towards 2030 (JAEC, 2012). 2.2.2. CO2 regulation policy Since 1990, energy-related CO2 emissions in Japan have increased at almost the same pace as energy supply. The fuel switching of energy supply from fossil fuel has so far made little progress. CO2 emissions in 2011 amounts to 1242 Mt, up 8.8% from 1990 level (IEEJ, 2014). In order to reduce the domestic total CO2 emissions, more technical and political effort should be placed on the emissions particularly in power generation sector which accounts for 36% of the total emissions in 2011 (Fig. 6). For discussing a sustainable pathway of power generation mix by using the model developed here, four scenarios including no regulation scenario are assumed in this paper, taking the CO2 emissions in 2010 as a benchmark (Fig. 7). The considered scenarios are composed of no regulation, 30% reduction, 50% reduction and 80% reduction by 2050 from the level of the emissions in 2010. In respective CO2 regulation scenario, a long-term power generation mix in Japan is evaluated, simultaneously considering the aforementioned nuclear policy scenario. According to IEEJ

Fig. 6. Breakdown of Japan's CO2 emissions in FY2011. The total emissions in 2011 amounts to 1168 t.

176

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

Fig. 7. CO2 regulation scenarios towards 2050. CO2 reduction by 30%, 50% and 80% in 2050 from the emissions in 2010 are assumed, together with no regulation scenario.

(2013), the Japanese total CO2 emissions in 2040 is projected to be almost halved from the levels of 2010 if efficient supply-side and demand-side technologies are incrementally penetrated in Japan's energy mix in the forecast horizon.

3. Results and discussion This paper analyzes total 16 simulation cases under both nuclear and CO2 regulation scenarios (¼4 scenarios  4 scenarios) explained in the previous section. 3.1. Power generation mix in 2050 Figs. 8 and 9 show the configuration of capacities and power generation mix respectively at 2050 in the individual scenario which is shown in the horizontal axes. CO2 regulation encourages fuel switching from coal to LNG combined cycle (LNGCC), wind, PV, rechargeable battery and electricity saving. In nuclear “Status quo” case, CO2 80% reduction scenario encourages the expansion of PV system, together with wind power deployed to its maximum limitation. At this case, the fraction of PV generation accounts for one-fifth of the total power generation. In nuclear “Decommission after 40 years” under CO2 80% reduction scenario, a radical decommissioning of nuclear significantly introduces PV system furthermore, and the ratio of PV in the total power generation mix amounts to 40%, which suggests that ensuring power supply security, derived from PV variability, becomes more critical. And in this case, massive energy saving is implemented; 300 TWh or 30% of reference power demand is

Fig. 8. Capacity mix in Japan at 2050 under nuclear and CO2 regulation scenarios.

Fig. 9. Power generation mix in Japan at 2050 under nuclear and CO2 regulation scenarios.

conserved in 2050, mainly because large-scale VR integration causes the rising of power generation cost and electricity saving holds cost competitiveness against supply side technology. It should be cautiously recognized, however, that the phase-out of nuclear in this model is mainly compensated by the increased deployment of PV and electricity saving, because additional thermal power plant is constrained due to CO2 regulation, maximum limitation of wind power is assigned and the capacities of other renewables such as hydro and geothermal are fixed by those maximum deployable potential. And the sufficient introduction of rechargeable battery is only observed in nuclear decommissioning after 40, 50 and 60 years in operation under CO2 80% reduction scenario where the fraction of PV in the mix becomes more than 30%, because an economic profile of rechargeable battery is less favorable due to its higher cost and other more affordable flexible power sources such as LNGCC mainly compensates a VR variability. The cost reduction of storage technology is an important agenda to be investigated, and Komiyama and Fujii (2014) performed a sensitivity analysis on battery cost to prove the economical trade-off between battery and the curtailment of VR surplus output. 3.2. Power generation cost Modeling of power system on a 10-min basis in a cost minimization approach enables us to investigate the effect of higher VR fractions on total system cost. In addition, the model here is capable of assessing a VR integration cost at an overall power system level, instead of separately calculating the costs associated with the impact of VR integration on power plants, battery and energy saving and adding them together. Practically, a portfolio of technologies accommodates VR in the grid under the minimization of total system cost. Therefore, the overall approach in this paper is preferable to evaluate the impact of VR integration on total system cost. Fig. 10 shows an annual total cost per electricity supply at 2050 in different scenario, and it reveals the escalation of power generation cost, as severe CO2 regulation policy is implemented and nuclear is decommissioned. From the figure, PV system cost and energy saving cost are major components to cause the cost increase of the system. Under nuclear “Decommission after 40 years” and “CO2 80% reduction” scenario (right edge of the figure), power generation cost shows a quadruple increase compared with that under current nuclear capacity (“Status quo”) and no carbon regulation policy (left edge of the figure), chiefly because of a higher penetration of PV in the system with rechargeable battery. VR integration into power system has positive impacts like fuel cost reduction and negative impacts like increasing investment for

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

Fig. 10. Power generation cost in Japan at 2050 under nuclear and CO2 regulation scenarios.

capital-intensive VR and battery. In this paper, even considering future PV cost reduction, it turns out that a larger fraction of VR in the grid significantly increases total system cost.

177

alternative generators even under severe CO2 regulation with a certain amount of energy saving introduced in 2050. However, in nuclear decommission after 40 years and CO2 80% reduction scenario (right figures on Figs. 11(b) and 12(b)), LNGCC cannot play a central role in 2050 as an alternative power source; on a power generation basis, LNGCC maintains the largest share until 2030, but thereafter, its fraction represents a shrinking trend due to stringent CO2 constraint, while PV together with rechargeable battery enormously expands and electricity saving is observed to be actively introduced from 2030. In this case (right figures on Fig. 11(b)), a required capacity of rechargeable battery in 2050 is rather smaller than that of PV, because other flexible resources such as electricity saving, VR suppression control and flexible LNGCC operation together treat a VR intermittency. Based on CO2 80% reduction scenario (right figures on Fig. 11), moreover, total capacity of the system shows a significant increase to 2050 compared with other CO2 regulations, because the capacity factor of PV is lower and its massive installation eventually causes a considerable growth of the total system capacity. 3.4. Optimal dispatch of power sources in 2050

3.3. Transition of power generation mix to 2050 Figs. 11 and 12 depict the trajectory of capacity and power generation mix to 2050 under nuclear (Status quo, Decommission after 40 years) and CO2 regulation scenarios (No regulation, 50% reduction, 80% reduction). In Fig. 12, the power generated by power plants includes the power charged into power storage technologies. In nuclear decommission after 40 years and CO2 50% reduction scenario (central figures on Figs. 11(b) and 12(b)), LNGCC is one of

Monthly optimal operations of power generators in May and December on 2050 are illustrated from Figs. 13 to 16 under the mix of two nuclear scenarios (Status quo, Decommission after 40 years in operation) and two CO2 regulation policies (No regulation, 80% reduction). According to Figs. 13(b)–16(b), CO2 80% reduction policy actively encourages the VR installation, and it is observed that VR variability is technically managed with electricity saving, rechargeable battery, pumped-hydro, flexible operation by thermal power plant and VR output suppression; multiple measures

Fig. 11. Capacity mix from 2010 to 2050 in Japan under nuclear (Status quo and Decommission after 40 years) and CO2 regulation scenarios (No regulation, 50% reduction and 80% reduction).

178

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

Fig. 12. Power generation mix from 2010 to 2050 in Japan under nuclear (Status quo and Decommission after 40 years) and CO2 regulation scenarios (No regulation, 50% reduction and 80% reduction). In this figure, the power generated by power plants includes the power charged into power storage technologies.

dynamically serve as a whole to control short-cycle variations of VR. It is important to recognize that, compared with conventional power dispatch, massive integration of VR leads to a significant transformation of power system operation to maintain the costeffectiveness of electricity supply. Thus, comprehensive employment of those demand-side and supply-side measures is critical in enhancing the ratio of intermittent sources that can be economically integrated in the power grid. Increase in system flexibility by those measures makes it possible to integrate larger VR penetration. Suppression control of VR output below its maximum output capacity serves as an

economical option to curtail VR excessive output and to control extreme VR intermittency which is costly to accommodate in a grid. In May (Figs. 13 and 15), solar insolation shows a higher intensity, while electricity demand is modest due to the moderate climate condition in Japan. When massive PV is installed, PV output turns out to frequently outstrip the electricity demand in this season even when sunshine periods coincide with high electricity demand, and surplus PV output is observed to be suppressed. In December (Figs. 14 and 16), solar insolation shows a lower intensity and the scale of PV output suppression decreases

Fig. 13. Monthly power generation profile in May at 2050 on nuclear “Status quo” scenario. “Demand (original)” indicates electricity load before electricity saving implementation.

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

179

Fig. 14. Monthly power generation profile in December at 2050 on nuclear “Status quo” scenario. “Demand (original)” indicates electricity load before electricity saving implementation.

as a result. Thus, these calculated results suggest that technically resolving the seasonal imbalance between solar insolation and electricity demand is indispensable to integrate a large-scale PV in Japan. This seasonal imbalance of PV output has an impact on the introduction of electricity saving in each season as well. Under high levels of PV penetration, as compensating measure for PV output decline, electricity saving is more implemented in December when the intensity of PV output decreases (Figs. 14(b) and 16(b)); electricity saving complements seasonal PV output decline and is economically justified to minimize the use of thermal power generation for complying with CO2 regulation. By contrast in May when a PV output intensity increases, electricity saving is not implemented particularly in the time of PV output peak (Figs. 13(b) and 15(b)), because, at the peak time, PV integration in the system becomes maximized and greatly contributes to comply with CO2 regulation policy.

3.5. Capacity factor of LNGCC As shown from Figs. 13 to 16, LNG combined cycle (LNGCC) serves as a ramp generator for regulating a VR variability, and it is noteworthy to confirm the operating behavior of LNGCC under large-scale VR integration. The study analyzes the impact of VR integration on a capacity factor of LNGCC, and examines the dynamic effects of VR on the country's electricity system. Fig. 17 depicts an annual profile of LNGCC capacity factor in different nuclear and CO2 regulation scenario at 2050 (Nuclear: Status quo, Decommission after 40 years, CO2 policies: No regulation, 80% reduction). Vertical and horizontal axes show the time of the day in 10-min and the day of the year respectively. In no CO2 regulation scenario (Fig. 17(a) and (c)) where PV constitutes only minor fraction in the system, LNGCC shows a higher operational level, almost a full-time operation, through a year in both daytime and evening of the day. When PV is largely deployed in the system (Fig. 17(b) and (d)), however, LNGCC capacity factor

Fig. 15. Monthly power generation profile in May at 2050 on nuclear “Decommission after 40 years”. “Demand (original)” indicates electricity load before electricity saving implementation.

180

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

Fig. 16. Monthly power generation profile in December at 2050 on nuclear “Decommission after 40 years”. “Demand (original)” indicates electricity load before electricity saving implementation.

shows a significant decline, a part-time operation, particularly in May and maintains a certain level in other months, while a level of LNGCC operation in daytime decreases throughout a year. In the months around May when solar insolation shows a higher intensity and electricity load exhibits modest level, large-scale PV install satisfies a majority of electricity demand, and LNGCC supply is not required so much in this season (Fig. 17(b) and (d)). Thus, due to this seasonal imbalance, a large-scale PV deployment might cause seasonal ununiformity of LNGCC operation. Up to now, LNGCC are an expected ramp generator for extreme short-cycle variations of VR output. Those results imply that an economic performance of LNGCC becomes unremunerated due to the considerable reduction of its load factor in specific seasons under massive VR penetration. Technical challenge is that higher VR penetrations and more flexible operation in the system make utilities with LNGCC under economic pressure. The value of LNGCC in complementing VR output fluctuation should be recognized, and it is necessary to incentivize utility to assure LNGCC operation even when it copes with a greater VR intermittency in the system. Higher VR expansion in the grid without specific policies leads to raise concerns regarding the investment for flexible power generators and the sufficient adequacy of the system. 3.6. Electricity saving under massive VR penetration Fig. 18 presents an annual profile of energy saving rate by electricity saving at 2050 under nuclear and CO2 scenarios (Nuclear: Status quo, Decommission after 40 years, CO2 policies: No regulation, 80% reduction), and the saving rate is defined as energy saving amount divided by reference electricity load (before electricity saving). In the figure, vertical and horizontal axes show the time of the day in a 10- min and the day of the year respectively. In no CO2 regulation policy (Fig. 18(a) and (c)), electricity saving implementation is limited due to its higher implementable cost compared with power supply technologies. By contrast, particularly in nuclear decommission after 40 years and CO2 80% reduction (Fig. 18(d)) where 40% of power generation supply is accounted by PV output, extensive electricity saving is observed in most of a year and annual total electricity demand is saved by 30%. Electricity saving is not so much introduced, however, in the time around daytime peak of PV output on May when PV output intensity remains higher, because surplus PV output is sufficiently available to cover the majority of power demand and a fraction of

PV power supply becomes maximized for complying with CO2 regulation. This paper assumes monopolized electricity market where long-term power generation mix is planned through the cost minimization approach. However, the Japanese electricity market is planned to be deregulated, based on the roadmap formulated by Ministry of Economy, Trade and Industry, Japan (METI, 2013). In the deregulated electricity market, various end-use measures such as demand response (DR) are expected to play a critical role for balancing the market, and those are not fully considered in this paper. DR provides an opportunity for end-user to serve a vital role in power system by mitigating or shifting the electricity demand in response to time of use (TOU) rate, real-time pricing (RTP) and critical peak pricing (CPP). They have attracted attention against the flat rates where, during a period of time, electricity usage is charged the same rate. TOU is usually assigned on usage over on-peak and off-peak periods when the price is predetermined and constant for each period, and RTP is a rate scheme under which rates change momentarily based on the wholesale electricity price. CPP employs a high price that becomes effective during critical peak periods of higher electricity demand with advance notice to the consumer. Ministry of Economy, Trade and Industry (METI) in Japan launched a DR demonstration project to assess its introductory potential in balancing demand and supply (METI, 2014c). One of highlight in the project is to investigate the DR potential through dynamic pricing (CPP). Smart community projects in Yokohama, Toyota, Kyoto and Kitakyushu of Japan are supported by METI. The impact of CPP is studied in those power systems (Asano and Yamaguchi, 2014; METI, 2014c).

4. Conclusions and policy implications This paper evaluates long-term scenarios of nuclear and VRs in a consistent way by developing a dynamic high time-resolution optimal power generation mix model; the highlight of the model is dealing with various type of flexible power resources such as flexible thermal power plant, rechargeable battery, electricity saving and VR suppression control in a 10-min resolution. The energy model here is useful because it enables us to evaluate the flexibility options through an optimization of the system as a whole, rather than focusing solely on VR technology itself. This

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

Fig. 17. Annual profile of LNGCC capacity factor in a 10-min resolution on 365 days at 2050 under nuclear and CO2 regulation scenarios.

181

182

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

Fig. 18. Annual profile of energy saving rate through electricity saving in a 10-min resolution on 365 days at 2050 under nuclear and CO2 regulation scenarios.

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

model will contribute to provide the informative data of VR integration in energy policy formulation and may offer suggestive insight regarding VR deployment. However, the analysis assumes VR output as completely predictable variable and does not consider the uncertainty of VR intermittency, and technical measures for voltage and frequency controls under massive VR penetration are not considered as well. Simulation results show that both nuclear phase-out and carbon regulation policies make VRs economically justified and encourage its expansion in Japan's long-term power generation mix (Figs. 8 and 9). When nuclear is totally decommissioned and CO2 is mitigated by 80% from the level of 2010 by 2050, power generation cost shows a quadruple increase compared with that in current nuclear capacity after the Fukushima (“Status quo”) and no carbon regulation scenario (Fig. 10). Massive VR integration has both positive impacts like fuel saving and negative impacts like capitalintensive investment for VRs. Even taking into consideration a cost decline of PV system for the future (Table 4), larger PV integration causes a considerable increase in total system cost as a whole. Those results reveal that simultaneous implementation of nuclear phase-out and severe low-carbon policy has a negative impact on cost-effective power supply in the country. Monthly electricity payment of a typical household in Tokyo showed a 32 percent increase from 6405 yen on Oct. 2010, before the Fukushima, to 8423 yen on October 2014 (TEPCO, 2014b), mainly because of a complete nuclear shutdown and the associated increase in expensive fossil fuel import. Under the current situation, extreme VR expansion will push up electricity price furthermore. From the calculated results, technical innovation and long-term cost reduction of VR energy system is necessary if VR is politically positioned as a mainstream for future sustainable power supply. In that case, large-scale VR integration decreases the capacity factor of LNG combined cycle (LNGCC) (e.g. Fig. 17(d)) and make it non-profitable for the investment and operation, which implies the altering role of LNGCC under higher penetrations of VR. In Japan, the Act for Partial Revision of the Electricity Business Act was enacted on 13 November 2013. Based on this, the Japanese government expects to fully liberalize the electricity market by 2020. In those deregulated electricity markets, the profitability of a ramp generator such as LNGCC needs to be guaranteed for integrating massive VR installation. Hence, it is imperative to establish capacity market for utility companies to assure LNGCC serving as a profitable ramp generator for the VR variability. Finally, as an economically rational solution, electricity saving serves as an important option to integrate large-scale VR and to solve the seasonal imbalance of VR output; electricity saving is largely introduced when VR output shows a lower intensity causing supply scarcity in the grid, while electricity saving is not so much implemented when VR provides surplus power enough to cover the most electricity demand with low carbon intensity. From the results, electricity saving is considered to be economically suitable to resolve the seasonal imbalance of electricity market particularly in Japan where seasonal changes, such as four seasons, of solar insolation and wind speed is strongly observed. Political focus needs to be put on designing appropriate frameworks of electricity pricing and advanced system that motivate end-use sectors to show a dynamic behavior of electricity demand. Future work consists in assessing the impact of VR on power grid by considering transmission line, refining the electricity saving mechanism with sectoral difference of price elasticity, including advanced technology such as carbon capture and sequestration (CCS) technology and combining demand extension technologies like electric vehicle and heat-pump water heater.

183

Acknowledgment This work was supported by JSPS KAKENHI Grant no. 25870176.

References Asano, H., Yamaguchi, N., 2014. The demonstration and application of demand response in Japan and overseas. Electr. Rev. 99 (10), 46–50 (in Japanese). Cheung, K.W., Rios-Zalapa, R., 2011. Smart dispatch for large grid operations with integrated renewable resources. in: Proceedings of the IEEE PES Innovative Smart Grid Technologies Conference, Anaheim, USA. Chubu Electric Power Company (CHUDEN), 2014. Press Release. 〈http://www.chu den.co.jp/corporate/publicity/pub_release/press/3254876_19386.html〉 (in Japanese). Deanea, J.P., Draytonb, G., Gallachóira, B.P., Ó., 2014. The impact of sub-hourly modelling in power systems with significant levels of renewable generation. Appl. Energy 113, 152–158. Denholm, P., Lund, H., 2011. Grid flexibility and storage required to achieve very high penetration of variable renewable electricity. Energy Policy 39, 1817–1830. Divya, K.C., Østergaard, J., 2009. Battery energy storage technology for power systems—an overview. Electr. Power Syst. Res. 79 (4), 511–520. Fujii, Y., Yamaji, K., 1998. Assessment of technological options in the global energy system for limiting the atmospheric CO2 concentration. Environ. Econ. Policy Stud. 1 (113–139), 1998. Gallestey, E., Stothert, A., Antoine, M., Morton, S., 2002. Model predictive control and the optimization of power plant load while considering lifetime consumption. IEEE Trans. Power Syst. 17 (1), 186–191. Hadjipaschalis, I., Poullikkas, A., Efthimiou, V., 2009. Overview of current and future energy storage technologies for electric power applications. Renew. Sustain. Energy Rev. 13 (6), 1513–1522. Hart, Elaine K., Jacobson, Mark Z.,, 2011. A Monte Carlo approach to generator portfolio planning and carbon emissions assessments of systems with large penetrations of variable renewables. Renew. Energy 36 (8), 2278–2286. Hu, X., Murgovski, N., Johannesson, L., Egardt, B., 2014. Comparison of three electrochemical energy buffers applied to a hybrid bus powertrain with simultaneous optimal sizing and energy management. IEEE Trans. Intell. Transp. Syst. 15 (3), 1193–1205. Hug-Glanzmann, G., 2011. A Hybrid Approach to Balance the Variability and Intermittency of Renewable Generation. IEEE PowerTech, Trondheim, Norway. International Energy Agency (IEA), 2013. World Energy Outlook 2013. OECD, Paris, France. International Energy Agency (IEA), 2014. The Power of Transformation—Wind, Sun and the Economics of Flexible Power Systems, OECD, Paris, France. 〈http:// www.iea.org/bookshop/465-The_Power_of_Transformation〉. Institute of Energy Economics Japan (IEEJ), 2013. Asia/World Energy Outlook 2013. Tokyo, Japan. Institute of Energy Economics Japan (IEEJ), 2014. EDMC Handbook of Energy & Economic Statistics in Japan, Energy Conservation Center. Tokyo, Japan. Japan Atomic Energy Commission (JAEC), 2012. Result of Study About the Options of Nuclear Fuel Cycle Policy. 〈http://www.aec.go.jp/jicst/NC/iinkai/teirei/sir yo2012/siryo22/siryo1-1.pdf〉 (in Japanese). Japan Meteorological Agency, 2007. Automated Meteorological Data Acquisition System (AMeDAS) (in Japanese). Kansai Electric Power Company (KEPCO), 2014. Press Release. 〈http://www.kepco. co.jp/corporate/pr/2014/1128_1j.html〉 (in Japanese). Keane, A., Tuohy, A., Meibom, P., Denny, E., Flynn, D., Mullane, A., et al., 2011. Demand side resource operation on the Irish power system with high wind power penetration. Energy Policy 39 (5), 2925–2934. Komiyama, R., Shibata, S., Nakamura, Y., Fujii, Y., 2013a. Analysis of possible introduction of PV systems considering output power fluctuations and battery technology, employing an optimal power generation mix model. Electr. Eng. Jpn., 182; a, pp. 9–19. Komiyama, R., Fujii, Y., 2013b. Assessment of Japan's optimal power generation mix considering massive deployment of variable renewable power generation. Electr. Eng. Jpn., 185; b, pp. 1–11. Komiyama, R., Fujii, Y., 2014. Assessment of massive integration of photovoltaic system considering rechargeable battery in Japan with high time-resolution optimal power generation mix model. Energy Policy 66, 73–89. Kiviluoma, J., Meibom, P., 2010. Influence of wind power, plug-in electric vehicles, and heat storages on power system investments. Energy 35 (3), 1244–1255. Kyushu Electric Power Company (KYUDEN), 2014. Press Release. 〈http://www.kyu den.co.jp/press_h141119-1.html〉 (in Japanese). Lund, H., 2005. Large-scale integration of wind power into different energy systems. Energy 30 (10), 2402–2412. Ministry of Economy, Trading and Industry, Japan (METI), 2010a. Future of Battery Industry in Japan, 2010 May. 〈http://www.meti.go.jp/report/downloadfiles/ g100519a02j.pdf〉 (in Japanese). Ministry of Economy, Trading and Industry, Japan (METI), 2010b. The Strategic Energy Plan of Japan—Meeting Global Challenges and Securing Energy Futures, 2010 June. 〈http://www.meti.go.jp/english/press/data/pdf/20100618_08a.pdf〉 (revised in June 2010). Ministry of Economy, Trading and Industry, Japan (METI), 2010c. Development of

184

R. Komiyama, Y. Fujii / Energy Policy 83 (2015) 169–184

Next-Generation Transmission and Distribution Grid for Low Carbon Society. Research Committee of Next-Generation Transmission and Distribution Grid, 2010 April. 〈http://www.enecho.meti.go.jp/committee/council/electric_power_ industry_subcommittee/001_03〉 (in Japanese). Ministry of Economy, Trading and Industry, Japan (METI), 2012a. Feed-in-tariff of renewable energy, July 2012 〈http://www.enecho.meti.go.jp/category/saving_ and_new/saiene/kaitori/dl/120522setsumei.pdf〉 (in Japanese). Ministry of Economy, Trading and Industry, Japan (METI), 2012b. About Battery and Hydrogen, July 2012 〈http://www.enecho.meti.go.jp/info/committee/ki honmondai/28th/28-5.pdf〉 (in Japanese). Ministry of Economy, Trading and Industry, Japan (METI), 2012c. Tentative summary of energy mix option, July 2012 〈http://www.enecho.meti.go.jp/commit tee/council/basic_problem_committee/017/pdf/17-3-1.pdf〉 (in Japanese). Ministry of Economy, Trading and Industry, Japan (METI), 2013. Report of expert committee for electric power system reform, February 2013. 〈http://www.en echo.meti.go.jp/category/electricity_and_gas/electric/system_reform001/pdf/ 20130515-1-1.pdf〉 (in Japanese). Ministry of Economy, Trading and Industry, Japan (METI), 2014a. About introduction situation of renewable energy, June 2014. 〈http://www.meti.go.jp/press/ 2014/06/20140617003/20140617003-2.pdf〉 (in Japanese). Ministry of Economy, Trading and Industry, Japan (METI), 2014b. News Release (March 25, 2014), March 2014. 〈http://www.meti.go.jp/press/2013/03/ 20140325002/20140325002.pdf〉 (in Japanese). Ministry of Economy, Trading and Industry, Japan (METI), 2014c. Demonstration Project of Next Generation Energy and Social System—Progress and Achievement, April 2014 〈http://www.meti.go.jp/committee/summary/0004633/pdf/ 016_01_00.pdf〉 (in Japanese). Murakami, K., 2012. Research on the Impact of Electricity Price Rise on Economy and Environment: Example Study on Greenhouse Gas Tax and Feed-In-Tariff. APIR Discussion Paper Series No. 23. National Policy Unit, 2011. Report of Cost Verification Committee, Tokyo, 2011 December (in Japanese). National Renewable Energy Laboratory (NREL), 2012. In: Hand, M.M., Baldwin, S.,

DeMeo, E., Reilly, J.M., Mai, T., Arent, D., Porro, G., Meshek, M., Sandor, D. (Eds.), Renewable Electricity Futures Study, vols. 4. National Renewable Energy Laboratory, Golden, CO (NREL/TP-6A20-52409). New Energy and Industrial Technology Development Organization (NEDO), 2004. Report about Japanese PV Roadmap for 2030, 2004 June, Tokyo (in Japanese). New Energy and Industrial Technology Development Organization (NEDO), 2009. Revised Report about Japanese PV Roadmap for 2030, 2009 June, Tokyo (in Japanese). Palchak, D., Denholm, P., 2014. Impact of Generator Flexibility on Electric System Costs and Integration of Renewable Energy. National Renewable Energy Laboratory, Golden, CO (NREL/TP-6A20-62275). Schaber, K., Steinke, F., Hamacher, T., 2012. Transmission Grid Extensions for the Integration of Variable Renewable Energies in Europe: Who benefits Where? Energy Policy 43, 123–135. Tohoku Electric Power Company, 2014. Press release. 〈http://www.tohoku-epco.co. jp/news/normal/1188477_1049.html〉 (in Japanese). Tokyo Electric Power Company (TEPCO), 2014a. Press release. 〈http://www.tepco.co. jp/cc/press/2014/1236420_5851.html〉 (in Japanese). Tokyo Electric Power Company (TEPCO), 2014b. Announcement About Fuel Cost Adjustment for Nov. 2014. 〈http://www.tepco.co.jp/e-rates/individual/fuel/ad just/index-j.html〉 (in Japanese). Troy, N., Denny, E., O’Malley, M., 2010a. Base-load cycling on a system with significant wind penetration. IEEE Trans. Power Syst. 25 (2), 1088–1097. Troy, N., Flynn, D., Milligan, M., O’Malley, M., 2010b. Unit commitment with dynamic cycling costs. IEEE Trans. Power Syst. 27 (4), 2196-2. Ummels, B.C., Gibescu, M., Pelgrum, E., Kling, W.L., Brand, A.J., 2007. Impacts of wind power on thermal generation unit commitment and dispatch. IEEE Trans. Energy Convers. 22 (1), 44–51. Xie, L., Ilic, M.D., 2008. Model predictive dispatch in electric energy systems with intermittent resources. In: Proceedings of the 2008 IEEE International Conference on Systems, Man and Cybernetics.