Application value of energy storage in power grid: A special case of China electricity market

Application value of energy storage in power grid: A special case of China electricity market

Energy 165 (2018) 1191e1199 Contents lists available at ScienceDirect Energy journal homepage: www.elsevier.com/locate/energy Application value of ...

1MB Sizes 0 Downloads 8 Views

Energy 165 (2018) 1191e1199

Contents lists available at ScienceDirect

Energy journal homepage: www.elsevier.com/locate/energy

Application value of energy storage in power grid: A special case of China electricity market Wei Wu a, Boqiang Lin b, * a

The School of Economics, China Center for Energy Economics Research, Xiamen University, Xiamen, Fujian, 361005, PR China School of Management, China Institute for Studies in Energy Policy, Collaborative Innovation Center for Energy Economics and Energy Policy, Xiamen University, Fujian, 361005, China

b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 5 April 2018 Received in revised form 28 August 2018 Accepted 30 September 2018 Available online 2 October 2018

With the increase of renewable energy permeability and the development of distributed grid, energy storage plays an increasingly important role in the power system. A lot of studies have shown that energy storage can already be economically feasible. However, most previous studies concentrated on the value of energy storage in the free electricity market. In China, the power grid monopolizes the process of electricity transmission, distribution and retail, and the feed-in tariff and retail prices of electricity are regulated by government. It is difficult to analyze the application value of energy storage for China's electricity due to the lacking of data. The major contribution of this paper is to evaluate the application value according to the data of a provincial power grid. The results support the argument that energy storage can generate positive returns. The optimal storage capacity and operational strategy are also discussed in this paper. © 2018 Elsevier Ltd. All rights reserved.

Keywords: Energy storage Power grid China electricity market Optimal storage scale

1. Introduction In 2016, China's electricity consumption was 5219 TW-hours, accounting for one-fourth of the world's electricity consumption. With the rapid economic growth, electricity consumption also maintained an upward trend. In response to the incensement of electricity demand, power grid investment also maintained rapid growth. The power system design is load-following, that is, the power supply terminal needs to be adapted to the demand. Due to the volatility of electricity demand, grid investment often retains greater redundancy. In recent years, the amount of investment in the power grid even exceeded the amount of power investment. Considering that the share of electricity consumption in the tertiary industry and household consumption in China is still increasing, and that the proportion of unstable renewable energy sources is also rising, the grid operation will face more challenges in the future. Energy storage system (EES) is considered as an important technology to enhance the flexibility of power systems, transferring loads and reducing the cost of power grids [1,2]. Currently, more than 99% of the energy storage capacity is large-scale energy

* Corresponding author. E-mail addresses: [email protected], [email protected] (B. Lin). https://doi.org/10.1016/j.energy.2018.09.202 0360-5442/© 2018 Elsevier Ltd. All rights reserved.

storage devices such as pumped hydroelectric storage (PHS) and compressed air energy storage (CAES), which accounts for 3% of the global generating capacity [3]. In recent years, with the decline of battery cost, the battery energy storage (BES) has become to receive widespread attention [4]. The BES has the advantage of high energy conversion efficiency, especially for some battery such as lithiumion battery can reach 94% conversion efficiency [5]. In addition, the BES is always applied in the distributed energy storage (DES), which can be distributed allocated on the demand side and reduce the grid reinforcement investment. Most of the current research of ESS applications is focused on the free electricity market. The applied value of energy storage is mainly derived from price arbitrage [6]. However, China's electricity market is a typical regulated market whose electricity price is regulated by government [7]. So it is difficult to obtain the true marginal cost of electricity and assess the value of energy storage. The major contribution of this paper is to evaluate the application value of energy storage in China according to the load data of a provincial power grid. We estimate the revenue of energy storage from the reduction of grid reinforcement and peak load shift. Considering that the operating mode of the ESS will affect the investment return, we apply the dynamic programming method to address the optimal storage capacity and operational strategy. The results of the relevant analysis can provide useful information for

1192

W. Wu, B. Lin / Energy 165 (2018) 1191e1199

regulators and grid companies to arrange energy storage configurations more reasonably. It should be note that we only consider the application of battery energy storage which is allocated on the terminal side in this paper. Some other storage technologies such as pumped storage and compressed-air energy storage are not included. Due to the limitation of the data, we can only provide a preliminary calculation for the allocation value of energy storage in China's power grid. However, we still believe the result might be useful for the policy maker and stakeholders. This paper is organized as follows. Section 2 briefly reviews the application of energy storage in the power grid and the method to solve the optimal operation mode of ESS. Section 3 gives an overview of the status of China's electricity market, and presents the optimal storage model for the regulated market. Section 4 describes the result and examines the influence of exogenous variables. Conclusions, as well as policy implications, are drawn in Section 5. 2. Literature review The electrical system is a complex system which needs to be analyzed by using techniques from complexity science [8]. The permeability improvement of intermittent renewable energy and the proportion increase of household and commercial in the electricity consumption structure may threaten the reliability of grid, and increase the operational cost of conventional power systems. The energy storage is a key section to deal with these challenges [9]. The adoption of ESS can bring a number of high-value applications for the power grid [5,10], and increase the total social surplus [11,12]. figured out that the role of ESS can be recognized either on the generation side or the demand side, and providing the ancillary services for the power grid. These advantages include [1]: On the supply side, ESS can improve the power grid's ability to absorptive renewable energy [13e15], reducing investment and operating costs of power supplies [16,17] and bringing down the peak feed-in tariffs [18]. On the demand side, ESS enables the consumer implement the demand side management [19,20], conduct price arbitrage [21,22] and deploy distributed photovoltaic systems [23]. Energy storage is also of great value in grid systems [24]. The ESS can enhance the reliability of grid, and making more efficient use of the network [25], eliminate peak-valley gap and smooth the load [26] and improve the operation efficiency of the power grid [27,28]. In recent years, a lot of researches are focusing on economic viability and optimal scale of energy storage [21] estimated the economic viability of energy storage in US electricity market and calculate the profit-maximizing size. The results showed that the energy storage can achieve an attractive internal rate of return for some regions [29] investigated the optimal procurement and scheduling of battery storage in distribution system with high photovoltaic (PV) penetration [30] assessed the economic viability of storage projects in the power grid under increasing wind penetration levels [31] quantified the economics of energy storage participating in arbitrage and regulation services within different markets [9] evaluated the economic viability of battery energy storage, and estimated the optimal storage scales for energy storage in China. The energy storage on the terminal side is growth fast in medium and smaller scale systems [32]. In distribution, EES can be installed in various places from the substation down to the end customers [33], which is often treated as DES. The DES can bring a lot benefits such as improving the reliability, enhancement of distributed generator and reducing equipment cost [34]. Batteries are probably the best option for DES [35]. The battery systems can offer a number of high-value opportunities in energy storage for the

grid [5,36] conducted the techno-economic analysis for residential battery storage systems and determined the optimal sizing [37] proposed a strategy for optimal integration of battery energy storage and illustrated that the DES can improve the hosting ability of the utility grid. 3. Methodology 3.1. Market description The electric utility of China is monopolized by two grid companies, the State Grid Corporation of China (SGCC) and the China Southern Grid Corporation (CSGC) [38]. These two giants control the process of transmission and distribution (T&D), distribution and retail. Although the power plants have been separated from the grid company since the last round reform, the feed-in tariff is still regulated by power grid and local regulators [39,40]. Due to the monopoly in T&D process, the amount of feed-in electricity is also dispatched by the grid companies. During operation, the power grid enterprises play the role of guaranteeing the power supply. In each province, there are local grid companies which are affiliates of SGCC and CSGC. The two power grid companies implement vertical management of provincial power grid companies, and provincial power grid companies undertake the actual operation of network planning, construction, maintenance and power reliability guarantee. The electrical grids represent both a natural monopoly and an essential facility. The power grids have to be regulated to avoid the abuse of market dominant position [41]. In China, the provincial government remains directly involved in investment and planning decisions, the feed-in tariff and retail price are different in each province [42]. The current price mechanism is that “retail price ¼ feed-in tariff þ T&D price þ T&D loss þ government fund” [43]. Feed-in tariff is the price purchased by power grids when access to the main grid. T&D price is the transmission and distribution cost included in the sales price. It mainly includes the depreciation of T&D facility and operating costs. The price variance between purchase and retail of each province often represents the cost in T&D process. In the actual implementation, the pricing of electricity often adopts a fixed price mechanism. For example, the retail price for the general industrial and commercial department is usually identical [43]. Although there the peak-valley prices policy is implemented, the electricity price is also conducted a fix price mode, which without taking into account the real-time cost of electricity. For such monopolistic state-owned enterprises as the power grid, economic interests are not their primary purpose. Grid companies often sacrifice economic benefits to ensure the reliability of power supply. At the peak of electricity demand, electricity prices are lower than the marginal cost of electricity supply. In the actual operation process, due to the lack of flexible price mechanism, and the mandatory task of the power grid for power supply, the power grid infrastructure often maintains a large degree of redundancy, which results in an inefficient of grid investment. 3.2. The model design Taking into accounts the special condition of China, the method to analyze the application value of energy storage in the previous studies may be hard to be adopted. On the one hand, the retail price of electricity is regulated and couldn't reflect the actual marginal cost of electricity supply. On the other hand, under the fix price mechanism, it's difficult to evaluate the consumer's reservation price of the electricity. To deal with these problems, we propose an analytical

W. Wu, B. Lin / Energy 165 (2018) 1191e1199

framework to evaluate the benefit of energy storage in the power grid which is installed on the terminal side, which can be regarded as a kind of demand side management tool. Fig. 1 provided the illustration. The EES includes battery system and inverter. By allocating the energy storage, the terminal load of the demand side can be smoothed and the infrastructure investment of the power grid can be reduced [34]. The benefits of the electricity energy storage (EES) for power grid are mainly in two aspects. First, EES can shift the load and reduce the peak load, so that the power grid can reduce the investment for the facility. The savings in investment costs can be seen as a benefit. The second is the reduction of feed-in cost. The marginal feed-in tariffs of the grid at different time points are different. In the peak period, the marginal feed-in electricity is supplied by high cost generators such as gas generation and pumped storage. The load shift from peak to valley will decrease the share of high feed-in tariff supply. The optimization objective function is presented in Eq. (1), which aims to maximize the profit of the power grid for a certain customer load mode when applying the EES. The symbol Rgc y represents the reduction of grid reinforcement costs, Rls is the revenue of load shift and CESS y is the investment cost of the DES facility. Of course, there are some other benefit by allocating the EES such as the value of ancillary and

max Profit ¼ Rgc

y

þ Rls  CESS

(1)

y

To evaluate the reduction of grid reinforcement costs, it's required to know the unit construction cost the power grid. However, since the China's power grids never published the specific investment data, we need to push backwards through the actual operating costs of the grid. As illustrated in Eq. (2), we decompose the operation cost of the power grid into two parts: the T&D loss cost and the grid capacity cost. Wherein, CGrid represents the annual T&D and feed-in cost; E is annual power purchase, hloss represents the line loss rate, PB represents the average feed-in tariff, Lmax represents the maximum load of the power grid, and Clg represents the annual unit load capacity cost. It should be noted that the T&D costs of the power grid are simplified as a linear function of the power grid capacity. This implicitly assumes that the grid is constant returns to scale. While this may introduce some bias, making such simplifying assumptions is useful in analyzing key issues where data sources are limited.

CGrid ¼ E*hloss *PB þ Lmax *Clg

(2)

In the condition of price regulation, the profitability of China's power grid enterprises is often very low. The rate of return on the grid may be hard to cover the economic rent (ie, opportunity cost) of the grid assets. Therefore, according to the research of [9], we assume that the grid has zero economic profit, which means the sales revenue equal to the grid cost (Eq. (3)). In the equation, RGrid is

AC BUS Power Plant

Load

Grid Inverter

Battery Storage Fig. 1. The illustration of energy storage on terminal side.

1193

the annual electricity retail revenue of the grid, PS is the average retail price. In fact, RGrid equivalents to the difference between the feed-in and retail process.

RGrid ¼ PS Eð1  hloss Þ  PB E ¼ CGrid

(3)

Combining Eq. (2) and Eq. (3), we can acquire Eq. (4), which can be utilized to calculate the unit cost of grid capacity.

Clg ¼ ðPS Eð1  hloss Þ  PB Eð1 þ hloss ÞÞ=Lmax

(4)

By adopting EES, the peak load can be reduced. According to the unit cost of grid capacity, we can calculate the revenue from the reduction of grid reinforcement costs Rgc y , which can be expressed as Eq. (5). In the equation, the return of the grid is a linear function of the maximum load variation of the grid.

Rgc

y

¼ DCGrid ¼ DLmax *Clg

(5)

The variation of peak load DLmax will be affected by the installed capacity of the DES and the specific operating mode. When the energy storage facility is put into operation, it will affect the grid load. The variation of grid load can be expressed by Eq. (6). lg is the spot load of the power grid, lc is the spot load of the demand side, and is the charge-discharge volume from the energy storage system to the grid.

lg ðtÞ ¼ lc ðtÞ  xðtÞ

t ¼ 1; 2; /8760

(6)

In the absence of energy storage devices, the grid load lg will equal to the demand side load lc . After connecting the DES, the grid load will change due to the load transfer. According to the maximum load before and after the adoption of DES, the variation of the peak load can be calculated which is shown in Eq. (7). The symbol lg represents the annual maximum load of grid (Eq. (8)), and lc represent the annual maximum load of demand side (Eq. (9)).

DLmax ¼ lg  lc

(7)

  lg ¼ sup lg ðtÞ

t ¼ 1; 2; /8760

(8)

lc ¼ supflc ðtÞg

t ¼ 1; 2; /8760

(9)

The stored quantity of electricity is given by Eq. (10). The symbol xðtÞ represents the net output electricity of DES in period t. When xðtÞ  0, the DES is in the process of discharge, the energy conversion efficiency of the battery hES and the efficiency of the inverter hinver need to be taken into consideration. If xðtÞ < 0, the DES is in the process of charge, it may need to consider the efficiency of the inverter. The stored quantity in each hour SðtÞ should be greater than 0 and less than the max storage capacity SC (Eq. (11)). The charge and discharge volume should satisfy the maximum power constraint of the energy storage system (Eq. (12)). The upper limit of power PowH is proportional to the max capacity of the energy storage system (Eq. (13)), where the scaling factor x is used to denote the ratio between system capacity and maximum power.

 Sðt þ 1Þ ¼

SðtÞ  xðtÞ=ðhES hinver Þ SðtÞ  xðtÞ*hinver

xðtÞ  0 xðtÞ < 0

(10)

0  SðtÞ  SC

(11)

1*PowH  xðtÞ  PowH

(12)

1194

W. Wu, B. Lin / Energy 165 (2018) 1191e1199

PowH ¼ xSC

(13)

In China, the feed-in tariff is also regulated by the government. A fix price mode is adopted for the generator. However, the feed-in tariff is different for different kind of power generator. The coal power and hydropower is often implemented a low feed-in tariff, and supply for most of the needs. But in the period of high demand load, the marginal generator is gas generator, which adopts a high feed-in tariff. The revenue of load shift is the price deviation between charge and discharge process, which can be calculated by Eq. (14). The symbol Rls represents the revenue of load shift, and PE is the marginal feed-in tariff. PE will change with the capacity of storage. As shown in Eq. (15), when the capacity of energy storage is small, all the marginal feed-in power which is substituted by energy storage is source from high-cost generator such as natural gas power generation. With the extending of energy storage capacity, the ratio of high feed-in tariff electricity in the marginal electricity support will decrease. The symbol PH represents the high feed-in tariff, and the symbol PL is the low feed-in tariff. The gH is the volume of high feed-in tariff power generator.

Rls ¼

8760 X

xðtÞ*PE ðtÞ

(14)

t¼1

 PE ðtÞ ¼

PH fPH gH þ PL ½xðtÞ  gH ðtÞg=xðtÞ

xðtÞ  gH ðtÞ xðtÞ > gH ðtÞ

(15)

In general, EES on the terminal side includes two main sections: power conversion system (PCS) and energy storage section [44]. In the model of [9], the investment cost of EES can be calculated by the cost of storage battery and inverter. According to Eq. (16), the investment of storage system CESS is a function of storage capacity, unit cost of storage battery CS unit and unit cost of inverter Cinv unit . Since we calculate the annualized investment income in Eq. (1), investment cost should also be apportioned to the annual investment cost according to the lifetime of the EES and the discount rate. Eq. (17) and Eq. (18) give the specific calculation method.

CESS ¼ SC CS unit þ PowH Cinv unit ¼ SC CS unit þ xSC Cinv unit

CESS ¼

lifetime X i¼1

0CESS

y

CESS

(16)

y

(17)

ð1 þ rÞi

¼ CESS rð1 þ rÞlifetime

.h

ð1 þ rÞlifetime  1

i

(18)

The above equations could be organized as the form of Eq. (19). The decision variables in the optimal problem include the installed capacity of energy storage SC and the discharge volume of EES at 8760 time points xðtÞ. 8760   X max Profit ¼  lg  lc Clg þ xðtÞ*PE ðtÞ 8760 SC ;xðtÞt¼0 t¼1 .h i    SC CS unit þ xSC Cinv unit rð1 þ rÞlifetime ð1 þ rÞlifetime  1

electricity. The capacity of the energy storage is determined by the peak load deviation and the price deviation. Some previous studies have used dynamic programming (DP) algorithm to solve this kind of problem [45] [46]. To solve our model, we combine the enumeration method and the DP algorithm. First, we enumerate all the storage capacity, and then adopt the DP algorithm to determine the optimal operating mode under each storage capacity. According to the revenue and cost under different storage capacity, we can get the optimal storage scale. When conducting the DP algorithm, there was the need to decompose a large optimization problem into some sub problems. The solution of a sub problem depends on the state of the solution of previous states. Since electricity demand tends fluctuate in days, we decompose the whole operational problem into sub problems which treat one day as a cycle. The time interval between sub questions is T ¼ ½T1 ; T2 ; /Tn . Under a certain storage capacity, the annual cost of EES is also determined, and can be regarded as a constant when solving the optimization problem. In the initial state, that is, for the first sub problem, the optimal solution is given by Eq. (20). The storage volume at the end of the first sub problem should be used as the initial value of the next sub problem.

dð1Þ ¼ fxðtÞ; SðT1 Þg

t ¼ 1; 2; /T1

(20)

Eq. (21) provides the optimal function of the sub problem. The net output volume xðtÞ will influence the peak load and should be treated as the state variable. T1   X max lg ð1Þ  lc Clg þ xðtÞ*PE ðtÞ T

1 xðtÞt¼1

(21)

t¼1

The peak load lg is a function of the state which will change with the progress of the sub problem. As can be observed in the previous equations, the maximum load on the grid depends on the initial load and the discharge volume of the energy storage system. When solving for the k-th sub-problem, we can get the maximum load change of the power grid based on the charge-discharge volume from the initial time to the state k (Eq. (22)).

  lg ðkÞ ¼ fl xðtÞ; lc

t ¼ 1; 2; /Tk

(22)

For the non-initial state solution, the new state contains the state of the previous sub problem and the solution of the current sub problem. Specifically, as presented in Eq. (23), the decision variables in the interval ½Tn1 ; Tn  need to satisfy the condition of Eq. (24).

dðnÞ ¼ fdðn  1Þ; xðtÞ; SðTn Þg

max

n xðtÞTt¼T



t ¼ Tn1 ; Tn1 þ 1; /Tn

Tn  X lg ðnÞ  lc Clg þ xðtÞ*PE ðtÞ

n1

(23)

(24)

t¼Tn1

3.3. Data summary

(19) The system operating conditions can be described in Fig. 2. The reduction of peak load can reduce the cost of power grid. The price deviation between charge and discharge can bring the revenue from arbitrage. When the load is low, the marginal feed-in tariff PE is low, and when the load is increase, the marginal feed-in tariff will increase until all the marginal supply are from the high cost

To verify the value of the model, we use the data of a provincial grid, which includes the spot load in the whole year, and the marginal power plant in each time point. . In 2015, Anhui's electricity consumption was 179.5 TWh, which exceeded the electricity consumption of the Netherlands, Poland, roughly 60% of UK's and one-third of Germany's electricity consumption. Fig. 3 illustrates the probability density distribution and cumulative density distribution of Anhui power gird's annual spot load in 2015. It can be seen from the figure that the load distribution in Anhui Province is

W. Wu, B. Lin / Energy 165 (2018) 1191e1199

1195

lc lg

lc lg

PE

Fig. 2. The illustration of running condition.

maximum load is small. However, the marginal cost to supply the peak load will not decrease. In fact, all the investment cost of the power grid will be spread into the retail price and finally transfer to the consumers. If the energy storage system is used to cut the valley and fill the peak to help improve the utilization of power grid equipment, the investment scale of the power grid could be curtailed effectively. The main parameters of the model are summarized in Table 1. The investment cost and system parameters of DES source from Ref. [43], the electricity consumption is from Anhui statistical yearbook 2016; the annual average feed-in tariff and annual average retail price source from Ref. [47]; the peak load, spot load, high feed-in tariff and low feed-in tariff are provide by the Anhui provincial electric power company. 4. Result and discussion 4.1. The optimal energy storage size

Fig. 3. China electricity market operator 1-h demand data for Anhui province in 2015 plotted as probability density and cumulative density.

close to the normal distribution, and the right side also shows the characteristics of long tail distribution. The capacity of the power grid needs to satisfy the peak load. For most of the period, the grid is operating in an unsaturated state. The case we choose had an annual maximum load of 28.68 GW, and annual minimum load of 8.59 GW. The maximum load is 3.34 times than the minimum load. The average load was 16.37 GW which was only 57.1% of the maximum load. In more than 90% of the time, the grid load is below 19.3 GW. This implies that the maximum load of the power grid is actually only satisfy the peak load of a few hours. The load characteristics of the power grid bring a lot of potential for the application of energy storage. The occurrence of extreme load is very short means the marginal supply benefit to the

Fig. 4 illustrates the revenue and cost variance with the energy storage capacity. Since the unit cost of the distributed energy storage is constant, the total energy storage cost CESS increases linearly with the increase in the capacity of DES. The revenue of power will grow with the capacity of DES. However, the marginal revenue of total revenue will decrease due to the marginal revenue of grid reinforcement reduction Rgc and load shift Rls will decrease. When the installed capacity of energy storage is small, adding one unit of energy storage can effectively reduce the peak load of the power grid and reduce the construction cost of the power grid. However, with the increase of the capacity of energy storage, the effect on peak load reduction will gradually decrease. The net revenue equal to the total revenue minus the energy storage cost, which exist an optimal revenue point. For our research case, the optimal storage capacity is 5 GWh where the net revenue can achieve a max level. This result can help us quantitatively analyze the optimal value of energy storage applications. Fig. 5 can help us better understand the reason why the marginal revenue of grid reinforcement is decrease. It can be seen from Fig. 5 that the load distribution of electric power is often unbalanced and the peak load occupies a shorter time. When the maximum load of the grid is decreasing from L1 to L2, the required

1196

W. Wu, B. Lin / Energy 165 (2018) 1191e1199

Table 1 Parameter summary. Symbol

Item

Unit

Value

E PB PS RGrid

Electricity consumption Annual average feed-in tariff Annual average retail price Total feed-in tariff and retail price differentials Peak load

TWh/year CNY/MWh CNY/MWh Billion CNY/year GW

163.98 489.67 690.14 32.87 28.68

Line loss rate Line loss cost Grid capacity unit cost Unit cost of battery Unit cost of inverter Discount rate Maximum power/storage capacity Battery efficiency Inverter efficiency Lifetime of BES system High feed-in tariff Base feed-in tariff

% Billion CNY/year CNY/(kW*year) CNY/kWh RMB/kW

7.67 5.72 946.75 350*6.25/0.8 800 5% 0.25 92% 97% 10 850 408.5

lc

hloss

E*hloss *PB Clg CS unit Cinv unit r

x hES hinver lifetime PH PL

Fig. 4. The revenue and cost variation with the energy storage.

L1 L2

L L

B

energy storage capacity is equivalent to the area of ABC in the diagram. When the load is reducing from L2 to L3, the required energy storage capacity is equal to the area of BCDE in the diagram, which is higher than the area of ABD intuitively. Therefore, with the drop of high load, the requisite capacity to further reduce the maximum load of the same amplitude is increased. In addition to the revenue of power grid reinforcement, the revenue from load shift will also decrease with the increase of the DES capacity, which is mainly due to the limited quantity of highpriced electricity. When the capacity of energy storage is small, most of the energy storage can be used to substitute the high-price electricity. However, as the capacity of energy storage increases, most of the high-priced electricity will has been substituted, and it will be more difficult to obtain peak shift benefits. Fig. 6 shows the load curves of one week in different seasons to show how the energy storage system will influence the power grid. The customer load represent the load demand of the customer side, and the grid load represent the load of grid support which equal to the customer load plus the storage load. It can be seen from the figure that the adoption of the energy storage system can effectively reduce the peak load and achieve a load shifting. The load of grid will be smoother compare with the actual customer load. Energy storage on the terminal side can bring great benefits to the power grid, but because it often needs to be deployed on the

C

D

E

Load

L3

A

e e e year CNY/MWh CNY/MWh

Time Fig. 5. The illusion for why the marginal revenue of EES is diminishing.

Fig. 6. The load curve variation in different seasons.

W. Wu, B. Lin / Energy 165 (2018) 1191e1199

demand side, the grid needs to transfer part of the revenue to the customers. Therefore, how to design an appropriate benefit allocation mechanism is essential to realize the value of EES. In the free electricity market, the price of electricity fluctuates in a relatively large range, so the profit of EES can be realized through arbitrage. Simultaneously, the power grid can also benefit from the arbitrage process [48]. However, for a regulated electricity market like China, demand-side electricity prices are strictly regulated because power grid enterprises monopolize the transmission, distribution, and sales of electricity. At present, many regions implement peak-valley pricing, that is, electricity prices are divided into three periods, which are peak, valley and flat. However, the research of [43] pointed out that peak-valley electricity prices and peak-valley time periods are often fixed. They argued that the arbitrage of energy storage in China's electricity market may make the deployed of EES unreasonable under some conditions. In other words, when the price is fixed, if the net present value of the energy storage investment is negative, there will be no installed energy storage capacity. And if the net present value is positive, it may lead to excessive capacity of energy storage and make the load distribution more unbalanced. 4.2. Sensitivity analysis The parameters such as discount rate, battery cost, battery system life, peak electricity price, grid cost, which are considered as exogenous variables in the model, may affect energy storage investment. The values of these exogenous variables may change constantly. To analyze the influence of exogenous variables, sensitivity analysis is conducted in this section. Fig. 7 describes the effect of the variance of different variables. The first figure in Fig. 7 is about the influence of discount rate on the optimal energy storage capacity and the optimal return. As the discount rate decreases, the optimal investment scale will increase and the optimal return will increase. This is mainly due to the fact that the present value of future income will increase after the discount rate decreases, which brings higher applied value to energy storage. Since the investment expenditure is paid at the launch period, the marginal cost of energy storage does not vary, so the optimal energy storage capacity and income increase at the equilibrium point. The cost of the battery also will influence the optimal scale of energy storage. The lower the cost is, the higher the optimal energy storage capacity will be. In addition, the cost of batteries is probably one of the most variable factors. According to the estimation of [49], the annual cost of lithium batteries fell by an annual average of 8% during 2007e2014 and the trend is still likely to pursue. The cost of battery is the major part of EES investment. The application value of energy storage will increase with the decline of battery cost in the future. In practice, a more flexible approach can be taken to reduce the cost of the energy storage system. As indicated by Ref. [50], the current secondary energy storage devices which are reused based on decommissioned vehicle batteries are also one way to improve

1197

utilization efficiency. China is currently the largest market for electric vehicles in the world, and sales of electric vehicles are still growing rapidly [51]. Therefore, there will be considerable retired vehicle batteries in the future, which may be suitable for the EES system. Of course, the energy conversion efficiency of the retired batteries will be low, so the exact value of the application remains to be further studied. System lifetime will also affect the application value of energy storage. A longer service life of energy storage facility would enhance the corresponding optimal energy storage scale and income. However, because of the limitations technology, it is often more difficult to extend battery life than to reduce costs. To lithium batteries, for example, the main impact on the lifetime is the stability of the electrode, which requires a breakthrough in material technology. This requires the government to increase investment in basic research of battery materials. The increase of high feed-in tariff will enhance the optimal capacity of EES and the profit. The main reason for this phenomenon is that the increase of high feed-in tariff will enhance the revenue from load shift. China's current high feed-in tariff generators are mainly sourcing from natural gas generation and pumped storage, which is used to respond to peak load of. China's natural resource endowment is characterized by a lack of natural gas, which led to a high natural gas prices. Although the price of natural gas is also regulated by the government, the consequence of this is the lack of supply of natural gas. At present, the annual operation hours of China's natural gas power generation is 2340 h, which are equivalent to running with only 26.7% of the load. Therefore, if the demand for natural gas goes up in the future, the high feed-in tariff still has the possibility to rise. Moreover, LNG accounts for a large proportion of the natural gas supply. The price of LNG tends to be more volatile. For example, for SeptembereDecember 2017, LNG prices in the Chinese market have risen by 156% due to a shortage of LNG supplies. Considering the influence of natural gas price increase, the application value of energy storage in the future will be further enhanced. The grid reinforcement cost will also affect the optimal result. Grid reinforcement cost analysis can be adopted to evaluate the new electricity infrastructure. On the one hand, as the price of land, labor and material rise, grid reinforcement costs will rise. Also, the expansion of the power grid may be based on the existing grid and the corresponding cost may be lower than the original construction cost. Here, we use the total asset turnover rate of SGCC as the indicator to evaluate the variation trend of grid reinforcement cost. The SGCC accounts for more than 80% of China's electricity market and can be representative of China's power grid. The total asset turnover rate equals operating income divided by total assets. The operating income of grid comes from sales revenue, which is linearly related to the electric sales volume. The total asset of the grid is the value of grid infrastructure. In 2008, the aggregate asset turnover rate of SGCC was 0.6814, while its total asset turnover rate in 2016 was 0.6434. The slowing down of asset turnover means that when the same unit of electricity is supplied, the cost of the grid is on the rise in recent years. If the trend will continue in the future, the value of energy storage will increase. 5. Conclusion and policy implication

Fig. 7. The optimal storage capacity and profit under different condition.

Energy storage has the application value in the field of load shift and peak load reduction. At present, the frequent business mechanism of energy storage is price arbitrage. In a free market, the price arbitrage of electricity can bring benefits to both the grid and the arbitrageurs. The arbitrageurs can acquire revenue from arbitrage, and the grid can smooth the load demand. However, for a regulated market such as China, the electricity price is always regulated by

1198

W. Wu, B. Lin / Energy 165 (2018) 1191e1199

government and does not reflect the true cost of electricity supply. However, China is the world's largest electricity market, accounting for 1/4 of the world's electricity consumption. Therefore, it is of profound significance for government and power grid enterprises to study the application value of energy storage in regulated electricity markets. Based on the previous literature, and from the perspective of power grid enterprises, this paper sets up a framework for analyzing the application value of energy storage in a regulated market. In this model, the influence of the application of energy storage on the grid reinforcement cost and load shift revenue is examined. A dynamic programming method is adopted to determine the operation mode, which divides the optimal problem into some sub problems. The results demonstrate that the energy storage can bring net revenue for the power grid. However, with the expansion of EES scale, the marginal revenue of grid reinforcement and load shift will gradually decrease, which implies that the application of EES is diminishing marginal returns. For a specified load mode and electricity demand, there is an optimal EES scale. Our model also calculates the optimal operation mode for the EES system. It is found that the EES can effectively reduce the load fluctuation of the power grid. To further analyze the impact of exogenous variables, we conducted a sensitivity analysis to examine how the discount rate, battery costs, battery life, peak electricity prices and grid reinforcement costs will influence the optimal energy storage installed capacity. The results of this paper can provide valuable information and suggestions for power grid enterprises and policy maker: First, since EES is often deployed on the demand side, a reasonable profit distribution mechanism needs to be designed between the power grid and the customers. The subsidy mechanism can be designed referring to the distributed photovoltaic policy of China, which provide cross-subsidize according to the capacity of power generation. The exact amount of subsidy should be determined according to the revenue of the power grid. Under this subsidy model, the customer and grid companies can share the benefit of energy storage, which lead to Pareto improvements. Meanwhile, due to the existence of the optimal storage capacity, a quota system can be adopted to ensure the optimal size of EES. Second, the optimal energy storage scale is influenced by the battery cost and service lifetime. Battery cost is likely to remain declining, so the potential for future energy storage may be further expanded. For the government, it is necessary to increase R&D subsidies to speed up the progress of battery technology. In addition, an appropriate policy should be designed to promote the secondary utilization of the batteries of retired vehicles battery to reduce the cost of energy storage system. Third, increasing the high feed-in tariff will increase the value of energy storage. At present, the peak demand of China's power is major relying on natural gas generation. However, as the price of natural gas is also regulated by the government and kept purposely low, the price of China's natural gas feed-in tariffs has been depressed. The efficiency of the electricity market can be improved by reducing the demand of peak natural gas power generation. Therefore, the government should reduce the regulation of natural gas price and stimulates the application of energy storage facility.

Funding This work was supported by the China's state grid corporation technology project “Key Technologies and Application of Energy and Electricity Price Forecast Analysis under the Background of New Electricity Reform” (grant numbers SGFJJY00JJJS1700025), Report Series from Ministry of Education of China (No. 10JBG013) and China National Social Science Fund (No. 17AZD013).

References [1] IEC. Electrical energy storage white paper. Geneva: IEC; 2011. [2] Metz D. Economic evaluation of energy storage systems and their impact on electricity markets in a smart-grid context. Portugal: Universidade do Porto; 2016. [3] Luo X, Wang J, Dooner M, Clarke J. Overview of current development in electrical energy storage technologies and the application potential in power system operation. Appl Energy 2015;137:511e36. https://doi.org/10.1016/j. apenergy.2014.09.081. [4] Larcher D, Tarascon JM. Towards greener and more sustainable batteries for electrical energy storage. Nat Chem 2015;7(1):19e29. https://doi.org/10. 1038/nchem.2085. [5] Dunn B, Kamath H, Tarascon JM. Electrical energy storage for the grid: a battery of choices. Science 2011;334(6058):928e35. https://10.1126/science. 1212741. [6] Krishnamurthy D, Uckun C, Zhou Z, Thimmapuram P, Botterud A. Energy storage arbitrage under day-ahead and real-time price uncertainty. IEEE Trans Power Syst 2018;33(1):84e93. https://10.1109/TPWRS.2017.2685347. [7] Mou D. Understanding China's electricity market reform from the perspective of the coal-fired power disparity. Energy Pol 2014;74:224e34. https://doi.org/ 10.1016/j.enpol.2014.09.002. [8] Bompard E, Maser M, Nuttall WJ. High tension electricity: new multi-scale complexities in the electricity system. Technol Forecast Soc Change 2015;96:327e33. https://doi.org/10.1016/j.techfore.2014.07.006. [9] Lin B, Wu W. Economic viability of battery energy storage and grid strategy: a special case of China electricity market. Energy 2017b;124:423e34. https:// doi.org/10.1016/j.energy.2017.02.086. [10] Roskilly AP, Taylor PC, Yan J. Energy storage systems for a low carbon future e in need of an integrated approach. Appl Energy 2015;137:463e6. https://10. 1016/j.apenergy.2014.11.025. [11] Kanakasabapathy P. Economic impact of pumped storage power plant on social welfare of electricity market. Int J Electr Power Energy Syst 2013;45(1): 187e93. https://doi.org/10.1016/j.ijepes.2012.08.056.  [12] Zidar M, Georgilakis PS, Hatziargyriou ND, Capuder T, Skrlec D. Review of energy storage allocation in power distribution networks: applications, methods and future research. IET Gener Transm Distrib 2016;10(3):645e52. https://doi.org/10.1049/iet-gtd.2015.0447. [13] Olabi AG. Renewable energy and energy storage systems. Energy 2017;136: 1e6. https://doi.org/10.1016/j.energy.2017.07.054. [14] Kabir MN, Mishra Y, Ledwich G, Dong ZY, Wong KP. Coordinated control of grid-connected photovoltaic reactive power and battery energy storage systems to improve the voltage profile of a residential distribution feeder. IEEE Trans Ind Inform 2014;10(2):967e77. https://doi.org/10.1109/TII.2014. 2299336. [15] Ma T, Yang H, Lu L. Development of hybrid batteryesupercapacitor energy storage for remote area renewable energy system. Appl Energy 2015;153: 56e62. https://doi.org/10.1016/j.apenergy.2014.12.008. [16] Ippolito MG, Di Silvestre ML, Sanseverino ER, Zizzo G, Graditi G. Multiobjective optimized management of electrical energy storage systems in an islanded network with renewable energy sources under different design scenarios. Energy 2014;64:648e62. https://doi.org/10.1016/j.energy.2013.11. 065. [17] Pudjianto D, Aunedi M, Djapic P, Strbac G. Whole-systems assessment of the value of energy storage in low-carbon electricity systems. IEEE Trans Smart Grid 2014;5(2):1098e109. https://doi.org/10.1109/TSG.2013.2282039. [18] Awad AS, Fuller JD, El-Fouly TH, Salama MM. Impact of energy storage systems on electricity market equilibrium. IEEE Trans Sustain Energy 2014;5(3): 875e85. https://doi.org/10.1109/TSTE.2014.2309661. [19] Poudineh R, Jamasb T. Distributed generation, storage, demand response and energy efficiency as alternatives to grid capacity enhancement. Energy Pol 2014;67:222e31. https://doi.org/10.1016/j.enpol.2013.11.073. [20] Nguyen HK, Song JB, Han Z. Distributed demand side management with energy storage in smart grid. IEEE Trans Parallel Distr Syst 2015;26(12): 3346e57. https://doi.org/10.1109/TPDS.2014.2372781. ~ o-Echeverri D. Economic viability of energy [21] Bradbury K, Pratson L, Patin storage systems based on price arbitrage potential in real-time US electricity markets. Appl Energy 2014;114:512e9. https://doi.org/10.1016/j.apenergy. 2013.10.010. [22] Zafirakis D, Chalvatzis KJ, Baiocchi G, Daskalakis G. The value of arbitrage for energy storage: evidence from European electricity markets. Appl Energy 2016;184:971e86. https://doi.org/10.1016/j.apenergy.2016.05.047. [23] Parra D, Gillott M, Norman SA, Walker GS. Optimum community energy storage system for PV energy time-shift. Appl Energy 2015;137:576e87. https://doi.org/10.1016/j.apenergy.2014.08.060. [24] Eyer JM, Iannucci JJ, Corey GP. Energy storage benefits and market analysis handbook, a study for the doe energy storage systems program. Sandia National Laboratories; 2004. [25] Saboori H, Hemmati R, Jirdehi MA. Reliability improvement in radial electrical distribution network by optimal planning of energy storage systems. Energy 2015;93:2299e312. https://doi.org/10.1016/j.energy.2015.10.125. [26] Han X, Ji T, Zhao Z, Zhang H. Economic evaluation of batteries planning in energy storage power stations for load shifting. Renew Energy 2015;78: 643e7. https://doi.org/10.1016/j.renene.2015.01.056.

W. Wu, B. Lin / Energy 165 (2018) 1191e1199 [27] Oh H. Optimal planning to include storage devices in power systems. IEEE Trans Power Syst 2011;26(3):1118e28. https://doi.org/10.1109/TPWRS.2010. 2091515. [28] Lawder MT, Suthar B, Northrop PW, De S, Hoff CM, Leitermann O, Subramanian VR. Battery energy storage system (BESS) and battery management system (BMS) for grid-scale applications. Proc IEEE 2014;102(6): 1014e30. https://10.1109/JPROC.2014.2317451. [29] Ansari B, Shi D, Sharma R, et al. Economic analysis, optimal sizing and management of energy storage for PV grid integration. In: Transmission and distribution conference and exposition (T&D), 2016 IEEE/PES. IEEE; 2016. p. 1e5. https://doi.org/10.1109/TDC.2016.7520090. [30] Das T, Krishnan V, McCalley JD. Assessing the benefits and economics of bulk energy storage technologies in the power grid. Appl Energy 2015;139: 104e18. https://doi.org/10.1016/j.apenergy.2014.11.017. [31] Berrada A, Loudiyi K, Zorkani I. Valuation of energy storage in energy and regulation markets. Energy 2016;115:1109e18. https://doi.org/10.1016/j. energy.2016.09.093. [32] Carpinelli G, Celli G, Mocci S, Mottola F, Pilo F, Proto D. Optimal integration of distributed energy storage devices in smart grids. IEEE Trans Smart Grid 2013;4(2):985e95. https://doi.org/10.1109/TSG.2012.2231100. [33] Nguyen CP, Flueck AJ. Agent based restoration with distributed energy storage support in smart grids. IEEE Trans Smart Grid 2012;3(2):1029e38. https://doi. org/10.1109/TSG.2012.2186833. [34] Nourai A. Installation of the first distributed energy storage system (DESS) at American electric power (AEP). Sandia National Laboratories; 2007. No. SAND2007-3580. [35] Zogg R, Tyson L, Ofer D, Brodrick J. Distributed energy storage. ASHRAE J 2007;49(5):90. [36] Hesse HC, Martins R, Musilek P, Naumann M, Truong CN, Jossen A. Economic optimization of component sizing for residential battery storage systems. Energies 2017;10(7):835. https://doi.org/10.3390/en10070835. [37] Jayasekara N, Masoum MA, Wolfs PJ. Optimal operation of distributed energy storage systems to improve distribution network load and generation hosting capability. IEEE Trans Sustain Energy 2016;7(1):250e61. https://doi.org/10. 1109/TSTE.2015.2487360. [38] Fan J, Zhao D, Wu Y, Wei J. Carbon pricing and electricity market reforms in China. Clean Technol Environ Policy 2014;16(5):921e33. https://doi.org/10. 1007/s10098-013-0691-6.

1199

[39] Wang Q, Chen X. China's electricity market-oriented reform: from an absolute to a relative monopoly. Energy Pol 2012;51:143e8. https://doi.org/10.1016/j. enpol.2012.08.039. [40] Ngan HW. Electricity regulation and electricity market reforms in China. Energy Pol 2010;38(5):2142e8. https://doi.org/10.1016/j.enpol.2009.06.044. [41] Zweifel P, Praktiknjo A, Erdmann G. In: Energy economics. Berlin, Heidelberg: Springer Texts in Business and Economics. Springer; 2017. p. 297e313. https://doi.org/10.1007/978-3-662-53022-1_13. [42] Zhang S, Andrews-Speed P, Li S. To what extent will China's ongoing electricity market reforms assist the integration of renewable energy? Energy Pol 2018;114:165e72. https://doi.org/10.1016/j.enpol.2017.12.002. [43] Lin B, Wu W. Cost of long distance electricity transmission in China. Energy Pol 2017a;109:132e40. https://doi.org/10.1016/j.enpol.2017.06.055. [44] Zakeri B, Syri S. Electrical energy storage systems: a comparative life cycle cost analysis. Renew Sustain Energy Rev 2015;42:569e96. https://doi.org/10. 1016/j.rser.2014.10.011. [45] Jiang DR, Powell WB. Optimal hour-ahead bidding in the real-time electricity market with battery storage using approximate dynamic programming. Inf J Comput 2015;27(3):525e43. https://doi.org/10.1287/ijoc.2015.0640. [46] Sioshansi R, Madaeni SH, Denholm P. A dynamic programming approach to estimate the capacity value of energy storage. IEEE Trans Power Syst 2014;29(1):395e403. https://doi.org/10.1109/TPWRS.2013.2279839. [47] National Energy Administration. 2013e2014 price regulation of electric power enterprises in China. Beijing: National Energy Administration; 2015 (in Chinese), http://zfxxgk.nea.gov.cn/auto92/201509/t20150902_1959.html [accessed 16 December 2016]. [48] Divya KC, Østergaard J. Battery energy storage technology for power systemsdan overview. Elec Power Syst Res 2009;79(4):511e20. https://doi.org/ 10.1016/j.epsr.2008.09.017. [49] Nykvist B, Nilsson M. Rapidly falling costs of battery packs for electric vehicles. Nat Clim Change 2015;5(4):329e32. https://doi.org/10.1038/nclimate2564. [50] Heymans C, Walker SB, Young SB, Fowler M. Economic analysis of second use electric vehicle batteries for residential energy storage and load-levelling. Energy Pol 2014;71:22e30. https://doi.org/10.1016/j.enpol.2014.04.016. [51] Lin B, Wu W. Why people want to buy electric vehicle: an empirical study in first-tier cities of China. Energy Pol 2018;112:233e41. https://doi.org/10. 1016/j.enpol.2017.10.026.