battery

Energy 118 (2017) 1110e1122

Contents lists available at ScienceDirect

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

Design and test of a new droop control algorithm for a SMES/battery hybrid energy storage system Jianwei Li*, Qingqing Yang, Francis. Robinson, Fei Liang, Min Zhang, Weijia Yuan* Dept. of Electronic and Electrical Engineering, University of Bath, Bath BA2 7AY, United Kingdom

a r t i c l e i n f o

a b s t r a c t

Article history: Received 28 December 2015 Received in revised form 9 September 2016 Accepted 29 October 2016 Available online 5 November 2016

High capacity energy storage units are desirable to maintain power system stability in the presence of power disturbances produced by renewable energy sources and fluctuating load profiles. Battery energy storage systems may be used to smooth power flow, however, the frequent, deep charge and discharge cycling required dramatically reduces battery service life. A hybrid energy storage system (HESS) using battery energy storage with superconducting magnetic energy storage (SMES) is proposed to mitigate battery cycling while smoothing power flow. A HESS power sharing control method based on the novel use of droop control is proposed. This is able to control charge/discharge prioritization and hence protect the battery from high power demand and rapid transient cycling. A sizing strategy is proposed for the battery and SMES which overcomes the oversizing problem. A hardware implementation is used to assess the control and SMES sizing methods for short time scale HESS operation. A dynamic off-grid seawave energy conversion system is simulated to assess the performance of the HESS over a longer time scale. A battery lifetime model which takes into account both battery life cycles and discharge current rate is used to estimate battery lifetime extension. A lifetime increase of 26% is obtained for the HESS design example investigated. © 2016 Elsevier Ltd. All rights reserved.

Keywords: Battery lifetime Droop control Hybrid energy storage system Superconducting magnetic energy storage

1. Introduction Energy storage systems, ESSs, have the potential to play a significant role in increasing the penetration of renewable power generation [1e3]. Previous work showed the different functions of ESSs, including power balancing [1,4], frequency control [5], voltage control [6], etc. Various kinds of ESSs are designed and widely demonstrated in renewable power applications [7e11]. Energy storage technology may be used to smooth the output power of renewable generators and provide a stable and dispatchable contribution to the power grid or local load. Battery energy storage systems, which are characterized by high efficiency and large energy density, are increasingly being used with renewable power generation. However, a battery has a limited number of charge/ discharge cycles, which limits its service life [3,12]. The operating stress is particularly onerous when batteries are used with renewable sources because they are subjected to many short-term charge/discharge cycles and high transient current levels [13,14].

* Corresponding authors. E-mail addresses: [email protected] (J. Li), [email protected] (W. Yuan). http://dx.doi.org/10.1016/j.energy.2016.10.130 0360-5442/© 2016 Elsevier Ltd. All rights reserved.

Moreover, when performing in a long-term energy storage capacity [15], the battery's relatively low power density means it cannot respond very quickly to high transient currents [16]. Therefore, many renewable power ESS investigations involve integrating high power density supercapacitors with batteries [17e20]. Compared with supercapacitors and other energy storage technologies, however, SMES devices have higher power density, lower selfdischarge rate, unlimited cycle life, and higher peak current handling capabilities [21e23], and potentially offer a significantly better component for hybridisation. In this work a novel control method and sizing procedure for a battery and SMES hybrid energy storage system, suitable for smoothing renewable source power fluctuations, is proposed and investigated. The control method for a hybrid energy storage system is more complicated than for a single energy storage technology because the required integrated control system needs to take into account the different characteristics of different energy storage technologies. Many control methods for HESSs have been studied in the literature. Ise. et al. in Ref. [21] Li. et al. in Ref. [13] introduced the filter based power distribution control for the battery and SMES hybrid system. In this method, the SMES is controlled to deal with high-frequency power while the battery is used to respond to long-

J. Li et al. / Energy 118 (2017) 1110e1122

Nomenclature C d1 d2 Emax Ibat IB_bat IB_SMES Im Imaxc(x) Imax_bat Imax_SMES Iref ISMES Ix kbat kSMES Pbat(t) Pi PL(t) Pm Pmean Pnet(t) Pnet_max

DC bus capacitor duty ratio of M1 while M2 is constantly on duty ratio of M2 while M1 is constantly off largest energy deficiency battery current output current of the battery DC/DC converter output current from SMES DC/DC converter maximum discharge current limit of the battery maximum charge current of the ESS x maximal battery current maximal SMES current reference current in sliding mode control SMES current charge or discharge current of the ESS x battery droop coefficient SMES droop coefficient battery power provisional SMES power defined in SMES sizing study instantaneous load demand (t) power in the largest net power fluctuation cycle average value of SMES power instantaneous net power peak value of the net power

term power fluctuations. However, the power filter parameters are very difficult to decide and are normally selected empirically. The battery current in a HESS can be limited to a desired range using model predictive control introduced in Ref. [24], but the utilization factor of the short-term energy storage system (the supercapacitor, the SMES etc.) is not very high. A power grading strategy [25] is proposed to share the fluctuated power between the SMES and the battery. Although the full power capacity of the SMES is fully used by this control method, the battery still undergoes high power fluctuations after the SMES is fully discharged. A decentralized control strategy using a droop controller has been previously demonstrated in a multi-terminal DC system [26,27], which is able to achieve the optimal sharing of various kinds of power sources when satisfying different load demand profiles. This paper introduces the novel use of a droop control method for controlling the power sharing between the SMES and battery. Less complex than previous approaches, the proposed control method is able to maintain power system stability by compensating both the high-frequency and low-frequency power fluctuations. Furthermore, compared with previous hybrid control methods, the new control has several advantages. Firstly, it is capable of prioritizing the operation of the SMES over the battery for short-term power fluctuations. Secondly, the charge/discharge level and rate-of-change for the SMES and the battery can be controlled by the droop factor. Thirdly, the proposed control is able to apply different energy storage system charge and discharge rates according to their distinct characteristics by setting different droop factors. A new sizing study of the HESS is proposed in this paper. The new sizing algorithm is performed in two steps: first, it sizes the battery capacity based on the system energy requirement; second, it determines the SMES capacity based on power requirements. In the first step, the “Loss of Power Supply Probability” (LPSP) optimization method [28e30], is used to estimate battery size for a required system reliability. In the second step, a new iterative sizing

1111

PSMES(t) Power dealt by the SMES PSMES_max maximal power dealt by the SMES Pw(t) instantaneous wave power output Vbat battery voltage Vbus DC bus voltage Vlow_end lower DC bus voltage boundary Vmax(x) upper voltage limit of the ESS x Vmax_bat upper voltage limit of the battery Vmax_SMES upper voltage limit of the SMES Vmin(x) lower voltage limit of the ESS x Vr DC bus reference voltage VSMES SMES voltage Vup_end upper DC bus voltage boundary DP power increment for SMES sizing study List of abbreviations BOS battery only system DDLWEC direct drive linear wave energy converter DOD depth of discharge EES energy storage system HESS hybrid energy storage system LPSP loss of power supply probability SC Super-capacitor SMES superconducting magnetic energy storage

method is proposed. In the second step, a new iterative sizing method is proposed. This method advances the previous sizing study of SMES/battery HESS by overcoming the oversizing problem and improving the SMES utilization factor. The performance of the proposed control algorithm for a SMES/ battery HESS is tested using both hardware experiments and simulation studies. In the simulation studies a direct drive linear wave energy converter (DDLWET) is the assumed renewable source. Battery lifetime extension is one of the most important advantages of the HESS with the proposed control. A battery lifetime model [13] is used in this paper, which is able to calculate the effect of both the charge/discharge cycles and the discharging rate on battery degradation. The battery depth of discharge (DOD) data extracted from a long duration simulation, is used to predict the battery lifetime extension. In this study, the battery lifetime is estimated to increase by 26% from 6.72 years to 7.92 years. DDLWEC systems have advantages as renewable sources, such as high reliability, lower initial investment and maintenance cost and less system complexity [11,31e34]. However, the unfiltered power output of DDLWECs is deeply fluctuant, and generally unsuitable for direct connection to the power grid or local loads, hence it is a good choice to assess the proposed HESS design. Several important challenges must be addressed in DDLWEC designs. First, the slow, random and cyclic sea wave induces a similarly shaped electromagnetic force waveform in the DDLWEC varying both in frequency and amplitude [35]. Secondly, the extracted power will drop to zero with a frequency twice that of the electromagnetic force waveform frequency which leads to large fluctuation (from seconds to tens of seconds) in output power [31]. Also, the energy in ocean waves is dependent on weather and tidal conditions [36]. This paper is organized as follows. The methodology including the concept of the droop control, the DC/DC controller for each ESS, the droop coefficient study and the system sizing study are described in detail in Section 2. In Section 3, a case study based on a wave energy conversion system is developed to verify the proposed

1112

J. Li et al. / Energy 118 (2017) 1110e1122

methodology in both the hardware experiments and simulation studies. Moreover, the battery lifetime quantitative analysis are introduced in this section. Section 4 gives the conclusions in the end. 2. Methodology 2.1. Droop control The typical arrangement of power converters and energy storage components of a SMES/battery hybrid energy storage system is shown in Fig. 1. The initial objective of the hybrid energy storage system is to maintain the DC bus voltage within a target range. The battery and the SMES are both connected to the DC bus through the DC/DC converters. The purpose of the power sharing control is to generate different converter references to charge and discharge the different energy storage devices at different rates. Decentralized control of the power contributions may be implemented using a droop controller as successfully demonstrated in a multi-terminal DC system [6,26,27]. This paper proposes the novel use of active current droop control in power sharing between the SMES and the battery by adjusting the droop factors for different ESSs. The working principles of the voltage versus current active droop control for HESS is illustrated by Fig. 2. Based on the DC bus voltage measurement, the energy storage systems ESSx (where x refers to the one of a number of ESS) charge or discharge at a level controlled by the following four conditions: 1. When the DC bus voltage Vbus
Vmax(x) (where Vmax(x) is the upper voltage limit of the ESSx), the ESS is charged at maximum charge current Imaxc(x). 2. When VbusVmin(x) (where Vmin(x) is the lower voltage limit of the ESSx), the energy storage system discharges at maximum discharge current Imaxdis(x). 3. In between Vmax(x) and Vmin(x), the current is controlled based on the current vs. DC bus voltage relationship:

Ix ¼ ðVbus  Vr Þ,kx

(1)

In Eq. (1), for a particular ESSx , Ix is the charge or discharge current, kx is the droop coefficient and Vr is DC bus reference voltage.

Fig. 2. Hybrid energy storage system voltage vs. current characteristic for active droop control.

Vbus is derived from measurement. The different energy storage devices have different charge/discharge rates for different droop coefficients (kx). To control the power ratio between different ESS, the droop coefficients for the battery and the SMES are set based on their different energy storage characteristics and operating constraints. Additionally, the battery and the SMES are protected from over charge/discharge by setting the voltage upper and lower boundaries, Vup_end and Vlow_end. 2.2. DC/DC control design for the energy storage systems 2.2.1. DC/DC controller for the battery The battery DC/DC converter with hysteresis current control is shown in Fig. 3. The hysteresis control has been shown to have good tracking capability and good robust performance [37,38]. The inductor current IL in Fig. 3, is regulated to track the reference

4. When Vbus ¼ Vr, the energy storage device is in stand-by and there is no charge or discharge current. The upper and lower voltage limit can be pre-defined based on manufacturer's data for energy storage devices. The DC bus voltage

Fig. 1. Typical control topology for the SMES/battery hybrid scheme.

Fig. 3. Battery DC/DC converter with hysteresis current control.

J. Li et al. / Energy 118 (2017) 1110e1122

current Iref using the sliding mode control as described in Refs. [39,40]. Based on the droop control method in Eq. (1), the droop control for the battery can be written:

IB

bat

¼ ðVbus  Vr Þ,kbat

(2)

The output current of the battery DC/DC converter IB_bat can also be written as Eq. (3)

IB

bat

V ¼ Iref $ bat Vbus

ISMES is the current flowing through the superconducting coil, IB_SMES is the output current from SMES DC/DC converter and d1 is conduction duty ratio of M1 while M2 is constantly on. SMES charge current can be regulated by controlling d1. 2. Discharge mode (Vbus < Vr) Applying Kirchhoff's current law at the positive capacitor node gives Eq. (6).

(3) C

Based on Eq. (2) and Eq. (3), the reference current Iref should be set as:

V Iref ¼ kbat $ðVbus  Vr Þ$ bat Vbus

(4)

2.2.2. DC/DC controller for the SMES The asymmetric full bridge DC/DC converter of the SMES in Fig. 4 (a) comprises diodes (D1, D2), MOSFETs (M1, M2) and an output capacitor (C). Three operating modes are used: charge mode by switching between conduction paths in Fig. 4 (b) and (d); discharge mode by switching between conduction paths in Fig. 4 (c) and (d); and stand-by mode with ISMES circulated as in Fig. 4 (d). The relative values of the bus and SMES voltages determine the operating mode at any time as illustrated below.

1113

dVbus ¼ IB dt

SMES

þ ISMES ,ð1  d2 Þ

(6)

d2 is the conduction duty ratio of M2 while M1 is constantly off. The discharge current can be regulated by controlling d2. 3. Stand-by mode (Vbus ¼ Vr) During this condition, the bus voltage is equal to the SMES nominal voltage and hence no output current is required from the SMES DC/DC converter. If the DC/DC converter capacitor is chosen to be sufficiently large, its voltage may be assumed approximately constant during a switching interval. Eqs (5) and (6) may then be approximated by Eqs. (7) and (8) for the charging and discharging conditions, respectively.

IB

SMES

¼ ISMES ,d1

(7)

1. Charge mode (Vbus > Vr):

IB

SMES

¼ ISMES ,ð1  d2 Þ

(8)

Applying Kirchhoff's current law at the capacitor positive node and state space averaging the switching interval gives Eq. (5).

By applying the general droop control equation Eq. (1) to the SMES gives Eq. (9). Using Eqs. (7) and (9) in this gives Eqs. (10) and (11).

C

dVbus ¼ IB dt

SMES

 ISMES ,d1

(5)

IB

SMES

¼ ðVbus  Vr Þ,kSMES

Fig. 4. SMES DC/DC converter showing current paths in different modes: (a) circuit topology; (b) charge mode; (c) discharge mode; (d) standby mode.

(9)

1114

d1 ¼

J. Li et al. / Energy 118 (2017) 1110e1122

ðVbus  Vr Þ,kSMES ISMES

(10)

Based on Fig. 2 and Eq (1), the battery maximal current can be obtained as:

Imax ðV  Vr Þ,kSMES d2 ¼ 1 þ bus ISMES

(11)

ISMES and Vbus values are both determined by measurement. Vr is the pre-defined value. Therefore, the conduction duty ratio d1 of M1 during the charge condition and duty ratio d2 of M2 during discharge condition may be readily calculated. As a result, the charge/discharge current of the SMES may be easily regulated using a DC/DC converter and droop control. In addition, during the standby condition where the Vbus is equal to predefined reference value Vr, d1 ¼ 0 and d2 ¼ 1, which simply requires that M1 is held constantly off and M2 constantly on. 2.3. Droop coefficient study 2.3.1. The droop coefficient for the battery Reduction of peak battery current is one of the advantages of HESS. The battery manufacturers often give an optimal range of battery discharge current beyond which the voltage of the battery will drop more substantially. Therefore, the battery maximal current Imax_bat should not be higher than the maximum current limit Im:

Imax

bat

 Im

(12)

The largest energy deficiency can be estimated by integrating the power in the largest net power fluctuation cycle with the fluctuation duration t:

Zt Emax ¼

Pm ðtÞdt

(13)

0

The battery provides long-term energy storage in the HESS, which offers energy support for the system. The battery charge/ discharge currents are determined by the droop controller. The battery should be able to buffer the largest energy deficiency Emax. Hence, the battery maximal current in the duration t should meet the condition as Eq. (14)

Imax

bat ,Vbat ,t


Emax

(14)

Based on Fig. 2 and Eq (1), the battery maximal current is given by:

Imax

bat

 ¼ Vmax

bat

  Vr ,kbat

(15)

Hence, using Eq. (12) to Eq. (15), a range for kbat can be obtained:

Emax Im    kbat  Vmax bat  Vr Vbat Vmax bat  Vr $t

(16)

2.3.2. The droop coefficient for the SMES The SMES has low energy density but high power capacity and is controlled to charge or discharge quickly to deal with most of the immediate power surplus or deficiency. Hence the maximal SMES current should be high enough to meet the maximum net power requirement, as given in Eq. (17):

Imax

SMES




Pnet max VSMES

(17)

SMES

 ¼ Vmax

SMES

  Vr ,kSMES

(18)

Hence:

kSMES
 Vmax

Pnet max   Vr ,VSMES

(19)

SMES

2.4. System sizing study In HESS design, the battery and the SMES should be sized to meet both the power and energy requirements to mitigate both long-term and short-term power fluctuations. The optimal sizing of a battery-only ESS for use in renewable power generations has been investigated previously [41e44], but it is rare to find published sizing examples that effectively combine the battery and the SMES together and then make use of their different energy storage characteristics. A HESS should meet both the energy and power requirements to fulfill a required smoothing duty. Therefore, the sizing study is divided into two steps: the energy dispatching requirement sets battery size to make sure that the load energy demand is always met; the short-term power fluctuation mitigation function determines the size of SMES. The method of “Loss of Power Supply Probability” (LPSP) has been successfully used in previous work [28e30]. Using the LPSP method for the off-grid domestic load and selected DDLWEC in a case study, the battery is sized to be 2.27 kWh to meet the energy requirement. As a first step, the battery is sized conservatively without consideration of the energy stored by the SMES. Considering the SMES has a small energy density and the relatively cheap price of batteries, the slight oversize of the battery is acceptable. Previous studies [13,31,45] have presented an approach to sizing the SMES based on the largest power requirement and the mean power demand, as illustrated in Fig. 5. The SMES should be able to handle the power difference between the maximum power (PSMES_max) and average power (Pmean) during the cycle period. However, this may result in oversizing the SMES, as implied in Fig. 5. The intermittent and unpredictable renewable source may cause extremely high or low power output, which produces an extremely high value of PSMES_max. As a result, the big difference between maximum power and average power implies a large SMES is required. Currently, SMES technology is expensive and in order to improve the utilization of the SMES, a new sizing algorithm is shown in Fig. 6. Fig. 6 shows the iterative process of the new SMES sizing method. The proposed dynamic off-grid wave energy conversion system model gives the power difference data between the generated power and load demand. The power PSMES processed by the SMES can be estimated using the proposed HESS controller. Then, the maximum value and the average value of PSMES during the simulation time may be determined. A power increment DP may be obtained by introducing a factor n to divide the difference between PSMES_max and Pmean. The provisional SMES power is defined as Pi ¼ Pmean þ i$DP (0  i  n). The size of SMES is next determined based on the provisional SMES power. By applying the battery size and provisional SMES size into the simulation model, the power specifications of the SMES PSMES and the battery Pbat are obtained. During the full simulation time, if condition PSMES(t) þ Pbat(t)
Pnet(t) is satisfied, the SMES capacity meets the requirements of the system. Otherwise, an increment is added to the provisional SMES power and the SMES is resized. Apart from

J. Li et al. / Energy 118 (2017) 1110e1122

1115

Fig. 5. Diagram used to describe the previous SMES sizing method.

Fig. 6. New sizing algorithm for SMES.

the off-grid wave system variables, the value of n has a significant effect on the final size of the SMES, as shown in Fig. 7. Higher n value gives a reducing size of SMES. When n ¼ 1, the SMES is sized by the maximum power requirement, as in the previous SMES sizing method and returns an oversized value of 1.82 kJ. A horizontal asymptote can be drawn in Fig. 7 which gives the optimum size of SMES for the proposed system as 0.72 kJ. Once the capacity of the SMES is determined, the optimum

design configuration should be considered in order to reduce the cost of expensive superconducting materials. The double pancake SMES coils configuration was chosen as suggested in Ref. [46]. 2.5. Methodology discussion The power fluctuations on the AC side cause DC bus voltage oscillations. Based on the droop control method described in

1116

J. Li et al. / Energy 118 (2017) 1110e1122

Fig. 7. Relationship between n value and the size of the SMES.

Section 2.1, the energy storage units are controlled to contribute to regulating the DC bus voltage, and hence attenuate the power fluctuations. Therefore, the presented control method is applicable more generally to address power fluctuation mitigation. Moreover, as has been seen in Sections 2.3 and 2.4, the system power/energy requirements are the dominant parameters for both the droop coefficient study and the system sizing study. This further indicates that the proposed hybrid energy storage design can be applied more generally to compensate power fluctuations. The SMES and the battery work together as a voltage source to maintain the DC bus voltage within the desired range, as implied by the hybrid energy storage system configuration shown in Fig. 1. The energy storage units (SMES and battery) can be replaced by other energy storage devices e.g. supercapacitors, full cells. However, the key benefit of the proposed droop control is that it is able to control the contributions of different ESS units, to best utilize their unique characteristics. Consequently, to fully exploit droop control, the energy storages unit characteristics are expected to be complementary to each other, such as in a supercapacitor/battery scheme or SMES/battery scheme. 3. Case study based on the DDLWEC system The generated power produced by a DDLWWEC is intermittent and varies in both frequency and amplitude. It cannot normally be fed directly to the AC grid or local loads hence is used in this paper as a very good case to verify the performance of the HESS with the new control method.

Table 1 Parameters for the battery droop coefficient study. Im Emax t Vr Vbat Vmax_bat

37 A 7.8 kJ 13.6 s 120 V 24 V 175 V

Table 2 Parameters for the SMES droop coefficient study. Pnet_max VSMES Vr Vmax_SMES

392 W 32 V 120 V 135 V

Based on the parameters in Table 2, the required droop coefficient value of the SMES, kSMES, is seen to be 0.817 or more. Unlike the battery, the SMES is able to charge/discharge quickly at high current and high frequency without degradation. Hence, the maximum current is not limited by the SMES itself but by other factors such as the rated current of the interfacing DC/DC converter and the storage capacity or size of the SMES. To reduce the size of the SMES, the droop coefficient of SMES should be limited to a lower value. Considering the observational error is in the range 0.3e0.5%, a kSMES value of 0.82 is selected to enhance system error tolerance.

3.1. Droop factors for the SMES and the battery 3.2. Experimental verification Tables 1 and 2 list the parameters for the battery and SMES droop coefficients study, respectively. The parameters are obtained from battery manufacturers' data and unfiltered DDLWWEC system simulations. Based on Table 1 and the method described in Section 2.3.1, the range of kbat is calculated as 0.434 to 0.67. A small kbat value is selected in order to decrease the battery peak current as much as possible. Considering the measurement error (1%) of battery voltage, kbat ¼ 0.44 is selected in the case study.

3.2.1. Laboratory test system configuration In order to verify the proposed new control algorithm and the sizing procedure for the SMES, a hardware prototype test system was constructed as shown in Fig. 8. Since the DDLWEC system was not available for functional testing, a DDLWEC and load emulator model was developed, based on the method described in Ref. [36], which is able to generate the wave power profile based on the real wave data. The DDLWEC emulator interfaces with the external

J. Li et al. / Energy 118 (2017) 1110e1122

1117

Fig. 8. DDLWET emulator with HESS laboratory test system configuration.

experimental hardware by using a programmable DC power source LAB-SMS, as shown in Fig. 8. This is able to generate real-time DC power to represent the DDLWEC system. Fig. 9 shows the experimental test setup. The battery subsystem comprises the DC/DC converter and four 12V/50 Ah lead-acid batteries with 24 V nominal voltage. A 0.34 H SMES-inductor rated at 65A DC nominal current and a full bridge DC/DC converter make up the SMES subsystem. The topology of the DC/DC converters and the control method are as presented in Section 3. The control algorithm for the HESS is implemented using the F28069 TI micro-controller. The LabVIEW is used for measurement data logging and supervisory control. Further details of the battery and SMES system hardware are listed in Tables 3 and 4 respectively.

3.2.2. Discussion of experimental setup and results The discussion of real-time operation of the experimental system mainly focuses on the short time period performance of the proposed HESS controlled by the proposed new control algorithm.

Table 3 Battery subsystem data. Battery type DC Bus capacitor Droop factor (kbat) Battery voltage IGBT

Fig. 9. Experimental system test setup.

50Ah lead acid 5700 mF 0.44 24 V MG1502YS50

1118

J. Li et al. / Energy 118 (2017) 1110e1122 Table 4 SMES subsystem data. Conductor Length Inductance Normal current Diode MOSFET (M1, M2) DC Bus capacitor Droop factor (kSMES)

z220 m z0.34 H 65 A DSEI2X121-02A IXFN140N30P 1.2 mF 0.82

To test battery performance the operating time scale would be very large and impractical with the experimental set-up used. Therefore, system simulation results are also presented in Section 3.2 to test the long-term performance of the system. To test the short time period performance, a short term power deficiency is created using the DDLWEC emulator, which causes a short term voltage sag in the AC side of the DDLWEC, as shown in Fig. 10 (a). In addition, this experiment is able to test the sizing study of the SMES by setting the power deficiency at the equivalent maximal value. One of the essential requirements for the SMES is to compensate the largest short-term power deficiency or the deepest short term voltage sag. Based on the real wave data, the equivalent maximal short term voltage sag is set at 35% of the normal voltage for 0.08 s (four

cycles), as shown in Fig. 10 (a). Fig. 10 (b) shows the compensated AC voltage on the load side and the response of the HESS is shown in Fig. 10 (c). The AC voltage on the load side is maintained at the normal value during the voltage sag period, which means that the power delivered to the load is a constant as expected (the load demand in the very short period is fixed). The SMES current is kept at its initial value in the normal condition. During the voltage sag period, the current decreases immediately as the SMES discharges to make up the power deficit. The battery current is kept at zero and only a slight undulation can be observed. This highlights the working principle of the HESS that the SMES responds to the shortterm fluctuation to protect the battery. In addition, as shown in Fig. 10 (c), the SMES current reduces from 65 A to about 3 A at the end of the voltage sag period, which means the SMES almost fully discharges during the most severe power deficiency. Fig. 10 (b) and the SMES current in Fig. 10 (c) confirm that the capacity of the SMES is able to meet the requirement of the DDLWEC system and that the SMES is fully utilized owing to the new sizing method. 3.3. Simulation verification A model of a dynamic off-grid wave energy conversion system shown in Fig. 11 was developed in Matlab/Simulink, which included

Fig. 10. Experimental results: (a) source voltage sag; (b) load voltage; (c) variation in SMES and battery currents.

J. Li et al. / Energy 118 (2017) 1110e1122

1119

Fig. 11. Configuration of DDLWEC and HESS systems used for long time period simulations.

the DDLWEC, power converters, batteries, SMES and a domestic offgrid load. A realistic domestic household load profile and the real wave data was used in the simulation. This system was also used to demonstrate the cooperative operation of SMES and battery and to give input data for the sizing study and battery lifetime extension analysis. The tubular DDLWEC is modeled using the method described in Ref. [36]. To satisfy an off-grid domestic load requirement, a 1 kW DDLWEC was modeled. The output from the DDLWEC was connected to the rectifier and inverter and fed to the off-grid load. The DC-link bus is a suitable location to connect ESSs as buffers, to balance any surplus or deficiency in load power. A realistic domestic household load profile is used in the simulation, which contributes to the volatility of system power. The battery and SMES are modeled based on the methods described in Ref. [13]. Fig. 12 shows the variation in the (a) generated, (b) load demand,

and (c) net powers. Fig. 12 (a) shows that the directly converted power from ocean waves is highly intermittent and contains significant high-frequency components. The load power in Fig. 12 (b) also contains some rapid variations. The load demand at peak load period (18:00e23:00) is much higher than that at other times. The net power in Fig. 12 (c), illustrates the power surplus/deficit over the 24 h period. It is apparent that the generated power cannot always meet the load demand. Therefore, the ESSs is required to mitigate the power fluctuations and balance the power. Fig. 13 shows the power output from the battery and SMES. As intended, the SMES responds to the high-frequency net power requirement while the battery control responds to smooth the long time scale power fluctuation. The results in Fig. 13 confirm the effectiveness of the proposed control strategy. The SMES power flow in Fig. 13 (b) is fluctuating at a much higher frequency than

Fig. 12. Power variations in: (a) generated power (b) load demand (c) net power.

1120

J. Li et al. / Energy 118 (2017) 1110e1122

Fig. 13. Power components of (a) battery and (b) SMES.

Fig. 14. (a) Battery current and DOD in BOS, (b) battery current, DOD and SMES current in HESS.

Fig. 15. Battery lifetime estimation process used to quantify lifetime extension.

that of the battery (Fig. 13 (a)), which means the SMES has mitigated most of the high-frequency components and the battery works as an energy buffer to SMES. It can be seen in Fig. 13 (b),

during the peak load period, that many ‘fully discharged’ points can be observed because the SMES has low energy density and it cannot meet the most severe transient power requirements. Therefore, at this stage, the battery discharges more deeply to provide greater energy support for power balancing. The battery performance of the proposed HESS is compared with that of a battery only system (BOS) in Fig. 14. The results further confirm the advantages of the new control strategy. As the SMES filters the high-frequency power fluctuations, the battery DOD and current in the SMES/battery HESS undergo significantly fewer polarity reversals than that in the battery only

J. Li et al. / Energy 118 (2017) 1110e1122

system. Also, the absolute value of the peak currents in the BOS are much higher than that in the HESS. Similarly, with the current, the battery in the BOS has deeper DODs. This figure illustrates why the control strategy is able to extend battery lifetime by reducing the peak current and smoothing the sharp charge/discharge transient content. 3.4. Quantitative analysis of battery lifetime extension “Battery lifetime is one of the key factors to evaluate a hybrid energy storage design [47]. With consideration of battery durability and longevity performance, Xiong. et al. in Ref [48] proposed a near-optimal power management method for the supercapacitor and battery hybrid system using in the electric vehicle achieving a good performance against uncertain driving cycles.” In this study, the battery durability and longevity are mainly affected by the frequent cycling process with the high discharging rate. The battery lifetime model introduced in Ref. [13] was used to calculate the effect of both the charge/discharge cycles and the discharging rate on battery degradation, and is used in this paper to predict the battery service lifetime. To estimate the improvement of the battery lifetime with the HESS, battery lifetimes for both BOS and HESS operating conditions are predicted and used as indicated in Fig. 15. The estimated battery lifetime for BOS is 6.27 years, whereas with the HESS it is significantly higher at 7.92 years, resulting in a 1.65 years (over 26%) improvement. 4. Conclusion This paper proposed a control and sizing methods for a SMES and battery hybrid energy storage system, which employs the novel use of droop control to smooth the power fluctuations arising in renewable power output and transient load demand. The novel control method is able to prioritize the power sharing between the SMES and the battery and manage the charge/discharge rates of the different ESSs by adjusting the droop factor to exploit the different characteristics of the ESSs. An off-grid DDLWEC emulation system is developed in both hardware experiment and the software simulation to verify the effectiveness of the novel control method. The relative sizing of the battery and the SMES capacities within a HESS is performed using the new method, which is shown to address the SMES oversizing problem previously identified. A battery lifetime prediction model is used to quantify the battery lifetime extension in the HESS. In this study, the battery service lifetime was improved by 26% from 6.27 years in a battery only system to 7.92 years when using the HESS. The novel droop control method has potential application with other HESS storage technology combinations; for example, with battery and supercapacitor HESS applications. Acknowledgements All the Authors would like thanks the support by EPSRC EP/ K01496X/1, Royal Academy of Engineering, UK. References [1] Lund H, Salgi G. The role of compressed air energy storage (CAES) in future sustainable energy systems. Energy Convers Manag 2009;50(5):1172e9. [2] Connolly D, Lund H, Mathiesen BV, Pican E, Leahy M. The technical and economic implications of integrating fluctuating renewable energy using energy storage. Renew Energy 2012;43:47e60. [3] Ferreira HL, Garde R, Fulli G, Kling W, Lopes JP. Characterisation of electrical energy storage technologies. Energy 2013;53(0):288e98. [4] Yang Q, Gu C, Le Blond S, Li J. Control scheme for energy storage in domestic households. Conference Control scheme for energy storage in domestic households. IEEE, p. 1e6.

1121

[5] Farhadi Kangarlu M, Alizadeh Pahlavani MR. Cascaded multilevel converter based superconducting magnetic energy storage system for frequency control. Energy 2014;70:504e13. [6] Kabir MN, Mishra Y, Ledwich G, Xu Z, Bansal RC. Improving voltage profile of residential distribution systems using rooftop PVs and Battery Energy Storage systems. Appl Energy 2014;134:290e300. [7] Panayiotou G, Kalogirou S, Tassou S. Design and simulation of a PV and a PVeWind standalone energy system to power a household application. Renew Energy 2012;37(1):355e63. [8] Ansari M Mohamed Thameem, Velusami S. DMLHFLC (Dual mode linguistic hedge fuzzy logic controller) for an isolated windediesel hybrid power system with BES (battery energy storage) unit. Energy 2010;35(9):3827e37. [9] Zhao H, Wu Q, Hu S, Xu H, Rasmussen CN. Review of energy storage system for wind power integration support. Appl Energy 2015;137:545e53. [10] Zhao P, Dai Y, Wang J. Design and thermodynamic analysis of a hybrid energy storage system based on A-CAES (adiabatic compressed air energy storage) and FESS (flywheel energy storage system) for wind power application. Energy 2014;70(0):674e84. [11] Gao Y, Shao S, Zou H, Tang M, Xu H, Tian C. A fully floating system for a wave energy converter with direct-driven linear generator. Energy 2016;95: 99e109. [12] Chen H, Cong TN, Yang W, Tan C, Li Y, Ding Y. Progress in electrical energy storage system: a critical review. Prog Nat Sci 2009;19(3):291e312. [13] Li J, Gee AM, Zhang M, Yuan W. Analysis of battery lifetime extension in a SMES-battery hybrid energy storage system using a novel battery lifetime model. Energy 2015;86:175e85.
lez F, Sumper A, Gomis-Bellmunt O, Villaf
[14] Díaz-Gonza afila-Robles R. A review of energy storage technologies for wind power applications. Renew Sustain Energy Rev 2012;16(4):2154e71. [15] Dunn B, Kamath H, Tarascon J-M. Electrical energy storage for the grid: a battery of choices. Science 2011;334(6058):928e35. [16] Ibrahim H, Ilinca A, Perron J. Energy storage systemsdcharacteristics and comparisons. Renew Sustain Energy Rev 2008;12(5):1221e50. [17] Chia YY, Lee LH, Shafiabady N, Isa D. A load predictive energy management system for supercapacitor-battery hybrid energy storage system in solar application using the support vector machine. Appl Energy 2015;137: 588e602. [18] Glavin M, Chan PK, Armstrong S, Hurley W. A stand-alone photovoltaic supercapacitor battery hybrid energy storage system. Conference A standalone photovoltaic supercapacitor battery hybrid energy storage system. IEEE, p. 1688e1695. [19] Kuperman A, Aharon I. Batteryeultracapacitor hybrids for pulsed current loads: a review. Renew Sustain Energy Rev 2011;15(2):981e92.
s G, Be
langer J. Real-time simulation of a wind turbine generator [20] Li W, Joo coupled with a battery supercapacitor energy storage system. IEEE Trans Ind Electron 2010;57(4):1137e45. [21] Ise T, Kita M, Taguchi A. A hybrid energy storage with a SMES and secondary battery. IEEE Trans Appl Supercond 2005;15(2):1915e8. [22] Mahlia TMI, Saktisahdan TJ, Jannifar A, Hasan MH, Matseelar HSC. A review of available methods and development on energy storage; technology update. Renew Sustain Energy Rev 2014;33:532e45. [23] Mohammad Reza AP, Hossine Ali M, Abbas S. Voltage stabilization of VSI SMES capacitors and voltage sag compensation by SMES using novel switching strategies. Energy 2010;35(8):3131e42. [24] Hredzak B, Agelidis VG, Jang M. A model predictive control system for a hybrid battery-ultracapacitor power source. Power Electron IEEE Trans 2014;29(3): 1469e79. [25] Li J, Zhang M, Yang Q, Zhang Z, Yuan W. SMES/battery hybrid energy storage system for electric buses. IEEE Trans Appl Supercond 2016;26(4):1e5. [26] Gavriluta C, Candela JI, Citro C, Rocabert J, Luna A, Rodriguez P. Decentralized primary control of MTDC networks with energy storage and distributed generation. Ind Appl IEEE Trans 2014;50(6):4122e31. [27] Asad R, Kazemi A. A novel distributed optimal power sharing method for radial dc microgrids with different distributed energy sources. Energy 2014;72:291e9. [28] Bakelli Y, Hadj Arab A, Azoui B. Optimal sizing of photovoltaic pumping system with water tank storage using LPSP concept. Sol Energy 2011;85(2): 288e94. [29] Wu W, Christiana VI, Chen S-A, Hwang J-J. Design and techno-economic optimization of a stand-alone PV (photovoltaic)/FC (fuel cell)/battery hybrid power system connected to a wastewater-to-hydrogen processor. Energy 2015;84:462e72. [30] Yang H, Zhou W, Lu L, Fang Z. Optimal sizing method for stand-alone hybrid solarewind system with LPSP technology by using genetic algorithm. Sol Energy 2008;82(4):354e67. [31] Nie ZX, Xiao X, Kang Q, Aggarwal R, Zhang HM, Yuan WJ. SMES-battery energy storage system for conditioning outputs from direct drive linear wave energy converters. IEEE Trans Appl Supercond 2013;23(3):5000705. [32] Falcao Antonio F de O. Wave energy utilization: a review of the technologies. Renew Sustain Energy Rev 2010;14(3):899e918. €m C, Leijon M. Operation analysis of a wave energy converter under [33] Bostro different load conditions. IET Renew Power Gener 2011;5(3):245e50. [34] Shek J, Macpherson D, Mueller M. Experimental verification of linear generator control for direct drive wave energy conversion. IET Renew Power Gener 2010;4(5):395e403.

1122

J. Li et al. / Energy 118 (2017) 1110e1122

[35] Bostrom C, Waters R, Lejerskog E, Svensson O, Stalberg M, Stromstedt E, et al. Study of a wave energy converter connected to a nonlinear load. Ocean Eng IEEE J 2009;34(2):123e7. [36] Nie Z, Xiao X, McMahon R, Clifton P, Wu Y, Shao S. Emulation and control methods for direct drive linear wave energy converters. Ind Inf IEEE Trans 2013;9(2):790e8. [37] Naderi S, Pouresmaeil E, Gao WD. The frequency-independent control method for distributed generation systems. Appl Energy 2012;96:272e80. [38] Castilla M, Guerrero J, Matas J, Miret J, Sosa J. Comparative study of hysteretic controllers for single-phase voltage regulators. IET Power Electron 2008;1(1): 132e43. [39] Kim I-S. Sliding mode controller for the single-phase grid-connected photovoltaic system. Appl Energy 2006;83(10):1101e15. [40] Tan S-C, Lai Y, Tse CK. A unified approach to the design of PWM-based slidingmode voltage controllers for basic DC-DC converters in continuous conduction mode. Circuits Syst I Regul Pap IEEE Trans 2006;53(8):1816e27. [41] Aghamohammadi MR, Abdolahinia H. A new approach for optimal sizing of battery energy storage system for primary frequency control of islanded microgrid. Int J Electr Power Energy Syst 2014;54:325e33.
be
C, Ndongo M. Optimal design of a [42] Ould Bilal B, Sambou V, Ndiaye P, Ke hybrid solarewind-battery system using the minimization of the annualized

[43]

[44]

[45]

[46]

[47]

[48]

cost system and the minimization of the loss of power supply probability (LPSP). Renew Energy 2010;35(10):2388e90. Ekren O, Ekren BY. Size optimization of a PV/wind hybrid energy conversion system with battery storage using simulated annealing. Appl Energy 2010;87(2):592e8. Tan CW, Green TC, Hernandez-Aramburo CA. A stochastic method for battery sizing with uninterruptible-power and demand shift capabilities in PV (photovoltaic) systems. Energy 2010;35(12):5082e92. Li J, Zhang M, Zhu J, Yang Q, Zhang Z, Yuan W. Analysis of superconducting magnetic energy storage used in a submarine HVAC Cable based offshore wind system. Energy Procedia 2015;75:691e6. Yuan W, Xian W, Ainslie M, Hong Z, Yan Y, Pei R, et al. Design and test of a superconducting magnetic energy storage (SMES) coil. IEEE Trans Appl Supercond 2010;20(3):1379e82. Li J, et al. Design/test of a hybrid energy storage system for primary frequency control using a dynamic droop method in an isolated microgrid power system. Appl Energy 2016. http://dx.doi.org/10.1016/j.apenergy.2016.10.066. Zhang S, Xiong R, Cao J. Battery durability and longevity based power management for the plug-in hybrid electric vehicle with hybrid energy storage system. Appl Energy 2016;179:316e28.