A Life Cycle-Cost Analysis of Li-ion and Lead-Acid BESSs and Their Actively Hybridized ESSs With Supercapacitors for Islanded Microgrid Applications

The combination of supercapacitors (SCs) with Li-ion Batteries (LIBs) and Lead-Acid Batteries (LABs) as hybrid ESSs (HESSs) have widely been proposed for Microgrid (MG) applications. The SCs of HESSs eliminate the stress of surge currents on LIBs and LABs, which increases their life cycles, and decreases their life cycle costs and hence decreases the HESSs operational costs. However, the active topology of HESS, which is the most commonly used configuration, requires an extra SC and an extra DC/DC converter in comparison to the Battery Energy Storage (BESS) topology, which increases the HESS capital cost. This paper tries to investigate that the hybridization of LABs and LIBs with SCs is economically effective or not for applications in islanded MG. In this regard, an energy management and frequency control (EMFC) scheme is proposed for the operation of MG in islanded mode. Using the simulations of the proposed EMFC scheme for islanded MG, the size of main components of LIB ESS (LIBESS), LAB ESS (LABESS), LIB-SC HESS (LISHESS) and LAB-SC HESS (LASHESS) are calculated. The numerical results show that for a 10-year period operation in islanded MG, the LISHESS and LASHESS impose less cost than LIBESS and LABESS. Also, the LISHESS is cheaper (almost 11%) than LASHESS.


I. INTRODUCTION
Integration of micro-sources and energy storage systems can construct a power grid so called microgrid (MG) to meet the demands, which can be operated in both grid-connected and standalone (islanded) modes. In the grid-connected mode, the utility grid should guarantee the demand-supply balance in the MG, and therefore the frequency in this mode is maintained on its nominal value [1], [2]. The main objective of MG operator in the grid-connected mode is to minimize power import from the grid by maximum utilization of renewable distributed generators (DGs) and coordinated control of ESSs and loads, while minimizing power losses and use of fuel-based DGs [3]. In case of disturbances or failure in the main grid, the MG shifts its operation to islanded mode to ensure system stability. In the islanded mode, the MG may The associate editor coordinating the review of this manuscript and approving it for publication was Jun Shen . suffer from the imbalance between the demand and supply [4], and as the capacity of DGs is limited, they may not be able to supply the loads during some periods. Therefore, the primary concern is to meet demand by available resources [5]. The intermittent nature of renewable DGs makes the task of maintaining the supply-demand balance even harder in islanded mode, which necessitates the installation of an ESS for this task [6]- [8]. Battery ESSs (BESSs) are accepted as one of the most important and efficient ways to maintain reliable energy supply. Among various battery technologies, the Lead-Acid batteries (LABs) are the most commonly used for grid-based applications [9]. Also, the research in [10] has referred to the Lithium-Ion Battery (LIB) as the most promising battery technology. Therefore, in this paper, both LIBs and LABs are considered for microgrid applications in islanded mode. In [11], Lithium-ion battery (LIB) is used for peak shaving and load leveling. In a standalone MG, the Lead-Acid batteries (LABs) are used for energy VOLUME 8, 2020 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ management and compensating supply-demand mismatch [12], [13]. The authors of [4], [14] investigated the application of LIB/LAB ESSs (LIBESS/LABESS) for frequency regulation of islanded MGs, respectively. In various studies, it has been stated that fast power regulations and high-power applications like frequency regulation shorten the life of LIBs and LABs [15], [16]. In this regard, the combination of supercapacitors (SCs) with batteries as hybrid ESSs (HESSs) have been proposed to reduce the stress of surge currents [17]. For this purpose, SCs are used to handle sudden variations of load demand and fast power fluctuations while batteries are used to just respond to slow power fluctuations [18], [19]. In [20], the SC is used to enhance the service life of the Lead-acid battery in a standalone photovoltaic-battery power systems by mitigating life-limiting factors such as current fluctuations and surge currents. In [21], an energy management scheme is proposed to maintain power balance using a HESS in an MG with renewable generation that due to the absence of dispatchable DGs, the battery might be overcharged/overdischarged. To avoid these conditions, the authors have proposed to disconnect loads and decrease photovoltaic (PV) power generation below maximum power point in shortage and surplus scenarios of power generation, respectively. In this study, the proposed energy management scheme uses the coordination of dispatchable DGs and the battery to prevent load shedding and also to ensure that PVs always operate at maximum power point. In [22], the LIB-SC HESS (LISHESS) results in 19% increase in LIBs lifespan. Similarly, the authors of [23] discovered that the LAB-SC HESS (LASHESS) results in 30% increase in LABs lifespan. The increase in lifespans of LIBs and LABs result in less operating cost. Fig. 1 shows the topology of a BESS and a HESS. Among different HESS configurations, the active topology is the most commonly used configuration [24]. It can be observed that the HESS requires an extra SC and an extra converter, which increases HESS cost. In other words, as the HESS operation cost decreases, the HESS capital cost increases. Furthermore, in [25] the authors stated at least 50% life extension is necessary to make hybridization with the SCs cost effective for application in the electric vehicles.
The installation of an ESS with random/inappropriate size can be problematic such as imposing extra cost to the MG. In order to minimize total cost of BESS, the authors of [26], [27] have determined the minimal size of battery including its rated power and rated capacity for islanded operation of MG for a limited period using short-term frequency regulation studies. However, the authors have not clearly explained how the dimensions of the battery especially its rated capacity is determined. Considering that maintaining balance in energy demand and supply in islanded MG is the priority [5], long-term energy management studies are required to determine the required energy storage capacity for this task, which are not performed in [26], [27]. In this study, both short-term frequency regulation and long-term energy management studies are conducted to determine the required power rating and capacity of the ESS elements.
In this paper, a life cycle-cost analysis is performed to determine whether the active hybridization of LIBs and LABs with SCs is cost effective or not for MG applications. An energy management and frequency control (EMFC) scheme is proposed for MG operation in islanded mode, which ensures the demand-supply balance and maintenance of frequency deviation within allowable limits using ESS elements with minimal size. Using the proposed EMFC scheme, the size of components of LIBESS, LABESS, LISHESS and LASHESS is determined. Then, their costs are evaluated and compared for a 10-year period. The main contributions of this paper are: • To the best of the authors' knowledge, a similar research has never been conducted to economically analyze two BESS technologies and their actively hybridized ESSs with SCs for islanded MG applications.
• Unlike [26], [27] that proposed a vague approach based on frequency regulation studies, here, a comprehensible approach is proposed based on both short-term frequency regulation and long-term energy management studies to determine dimensions of the battery and the SC for the islanded operation of the MG.
• The proposed EMFC uses the coordination of DGs and battery in minimization of the required battery capacity for maintaining the demand-supply balance. The MG frequency is also kept within permitted operational limits using the optimal tuning of frequency controllers of battery/SC.
• Introducing the procedure and formulae for cost calculation of four commonly used and studied ESS technologies. The general overview of the sections, simulations, figures, tables and their relations are shown in Fig. 2.

II. PROPOSED EMFC SCHEME
The utility grid guarantees the demand-supply balance in the MG in the grid-connected mode, and thus, the frequency can be maintained in an acceptable range; though, in the islanded mode, the MG may experience fast frequency deviations due to the intermittent nature of renewable sources, load variations and etc. The frequency of the islanded MG can be governed by (1) [7]: where, f and f 0 is the MG frequency and the MG nominal frequency respectively. Also, H MG is the MG total inertia, P G , and P L are the total power generation and consumption of the MG, respectively. In islanded mode, the MG microsources are responsible for maintaining the demand-supply balance. Considering inherent slow response characteristics, the slow responding DGs are incapable of handling fast power regulations. However, the total capacity of DGs can supply the MG base load, but they cannot supply the peak load. Therefore, the installation of an ESS is necessary for fast transient power regulations and energy management during the peak-load period. Based on (1), the EMFC diagram of the islanded MG with coordinated control of an ESS and DGs is shown in Fig. 3. All the DGs have outputs representing the power generation of the DG while only dispatchable DGs have inputs to receive the power setpoint. Renewable sources like PVs are assumed to be controlled to produce their maximum power. As far as the balance between consumption and generation is maintained, the frequency is stable and remains in acceptable range.
The proposed frequency control consists of two stages: 1) primary frequency control (PFC) by the ESS using a proportional controller, which is denoted by K 1 , 2) Secondary frequency control (frequency restoration) by dispatchable DGs using an integral controller, which is denoted by K 2 . Regarding the PFC using LASHESS and LISHESS, the interception of the frequency deviation is handled by the SC. It responds to high-frequency components of MG frequency deviations using a high-pass filter. By doing this, fast power regulations are diverted to the SC. This is while the battery responds to low-frequency components of MG frequency deviations and deals with slow power regulations using a low-pass filter and a proportional controller, which is denoted by K 3 .
In case the DGs reach their maximum power, the battery also participates in the frequency restoration using an integral controller, which is denoted by K 4 . Another integral controller, which is denoted by K 5 , is also used to restore the battery power to zero, if the DGs are not generating their maximum power. In addition, an integral controller, which is denoted by K SOC , is used to control DGs power to discharge/charge the battery to maintain the (SOC BAT ) at reference value (SOC ref BAT ). The acceptable limits in frequency deviations of an islanded MG is ±1% of the rated value [28]. This paper proposes a novel approach for maintaining the MG frequency within permitted operational limits. By increasing and decreasing K 1 , the rate of power injection for frequency interception increases and decreases, respectively; the higher rate of the power injection/absorption, the less frequency deviation. Fig. 4 shows the tuning process of the K 1 by Genetic Algorithm (GA), which ensures enough   power is injected/absorbed to maintain the frequency within the permitted limits when the most severe disturbance occurs [29]- [31].
In fact, the genetic algorithm is used just once to determine the value of frequency controllers. In real-time operation, the frequency controllers, which are optimally determined by the GA, are used to handle frequency control. The most severe frequency disturbance is considered so that that the MG frequency remains within allowable limits in any less severe disturbances as well. The power rating of the ESS elements participating in PFC is determined based on the maximum power injection/absorption.
The ESS control strategy for energy management in islanding mode is charging during off-peak period, and discharging during peak hours. The mid-summer day, which has the heaviest daily load profile, is chosen as the worst case for ESS capacity sizing. If the BESS can manage power balance in this case, it certainly can take care of the MG energy management in less severe cases like any other days in a year as well.
A. CONFIGURATION OF BESS AND HESS FOR EMFC SCHEME FOR PFC Fig. 5 shows the components, topology and control system of a BESS, which is connected to the MG. It mainly consists of a battery stack, a bidirectional DC/DC converter, a bidirectional three-phase inverter and filters. In addition, current and voltage transformers are used to take current and voltage samples from the inverter output. The phase-locked loops (PLLs) are also used to determine the phase angle and frequency.
The discharge/charge control of the battery through the control of the DC/DC converter has been completely described in [32]. The power reference (P batref ) determines the required power injection/absorption by the battery. In the grid-connected mode, the MG management center determined the setpoint for the battery power. However, in the islanded mode, P batref is defined by a MG proportionalintegral (PI) controller. The error signal of integral and proportional controllers and the actual battery power determine P batref . If the DGs reached the maximum power generation, the integral controller would be blocked and only the proportional controller would be functional.  The inverter controller is the conventional dq current controller, completely discussed in [7]. I dref and I qref determine the output active and reactive powers of the inverter. The voltage of the DC link (V DC ) between the converter and the inverter is maintained at a reference level (V DCref ). To realize this task, a PI controller is used to regulate I dref . For discharging the battery energy to the MG, the converter discharges the energy of the battery into the DC link capacitor (C DC ), which results in an increase in V DC . The inverter controller tries to keep V DC at V DCref . Therefore, the inverter discharges the charged capacitor energy to the MG. For charging the battery from the MG, the converter takes energy from C DC , which causes the V DC to decrease. The inverter controller tries to keep the V DC at V DCref . Therefore, the inverter charges the discharged capacitor energy from the MG. In this paper, I qref is set to zero to maintain unity power factor [7]. Fig. 6 shows the topology, components and control system of a HESS, which is connected to the MG. The HESS consists of two ESS elements that in this paper one of them is the SC and the other is a battery, which is either an LAB or an LIB. Each of them is connected to a separate converter. Some major parts of the HESS is the same as BESS like the converters and the battery discharge/charge controller. The frequency controller of the battery in HESS is the same BESS except that the battery frequency controller in HESS just responds to low-frequency components of the MG frequency using a low-pass filter. The SC just responds to the high-frequency components using a high-pass filter and does not participate in frequency restoration. The time constants of the high-pass and low-pass filters (T ) are chosen 1.5 sec as proposed in [13]. In addition, two PI controllers are added in the power reference controller of the battery and SC. Using these PI controllers, if the SOC SC is lower than SOC ref SC , the battery discharge its energy to the SC through the DC link to charge the SC. If the SOC SC is higher than SOC ref SC , the SC discharge its energy to the battery. This process is continuously executed to maintain the SOC SC at SOC ref SC . It can be seen that the topology and control of HESS is more complex than that of BESS.

B. COST FORMULATION OF BESS AND HESS FOR PRIMARY FREQUENCY CONTROL
The BESS mainly consists of a battery stack, a converter, and an inverter, its cost can be calculated using (2): where, BC Con is the base cost ($/kW) of the bidirectional DC/DC converter. RP Con is the rated power (kW) of the VOLUME 8, 2020 bidirectional DC/DC converter. RP Inv and BC Inv are the rated power (kW) and base cost ($/kW) of the bidirectional Inverter. RC BAT and BC BAT are the rated capacity (kWh) and base cost ($/kWh) of the LIB/LAB. The rated capacity of the LIB/LAB stack can be calculated using (3), based on its maximum discharged energy (E BAT maxDis ) in the EMFC studies.
The cost of the LIB with 6000 life cycles is 1000 $/kwh [33]. The cost of an LAB cell with 2000 life cycles is 600 $/kWh [34]. Considering one discharge/charge cycle per day [35], the BC LIB for LAB and LIB can be calculated for a 10-year period:

600$ 2000 [life_cycles]
× 3650cycles Considering the inverter efficiency, the value of RP Con can be calculated as follows: where, P max Inv is the maximum absorbed/injected power of the LAB/LIB. In addition, for a 10-year period, the values for BC SC , BC Con and BC Inv are 15000 $/kW, 50 $/kWh [36] and 600 $/kWh [37], respectively. The cost of HESS, which consists of an LAB/LIB stack, an SC, two converters and an inverter, can be calculated using (7): where, RC SC and BC SC are the rated capacity (kWh) and base cost ($/kWh) of the SC. RP SCCon is the rated power (kW) of the bidirectional DC/DC converter, which is connected to the SC. As the utilization of LABs and LIB with SC in parallel results in 30% [23] and 19% [22] The SC should provide the maximum required discharge (E SC maxDis )/charge (E SC maxcha ) energy in power shortage/surplus cases to intercept frequency deviation. Therefore, it should have enough stored energy for injection in the most severe case of power shortage or enough uncharged capacity to absorb energy in the most severe case of surplus power. Considering the efficiencies of the converter and inverter, the required capacity of the SC stack can be calculated as follows: The reference value of SC (SOC ref SC ) should be carefully chosen so that the SC can provide both maximum discharged/charged energies. According to [38], the voltage of SCs can be decreased to 50% of their rated voltage, and consequently, they can only use 75% of their capacities to discharge/charge, which means 25% of its rated capacity should always be kept charged. On this basis, the SOC ref SC can be calculated using (11): The values of RP Con can be calculated based on the maximum power that is passed through the converter.
Considering that the power of converter transmits through the inverter, and the inverter efficiency of the inverter is not considered in the simulations, the value of RP Con can be calculated as follows:

III. SIMULATION STUDIES
As the HESS and BESS are responsible for EMFC, the simulation studies are divided into two sections. First section deals with short-term frequency regulations, and the second section deals with long-term energy management. Using the results of these two sections, the sizing of components of LIBESS, LABESS, LISHESS and LASHESS will be carried out. Fig. 7 shows the MG network under study, which consists of a 0.4 kV distribution feeder connected to a 20 kV distribution network through a transformer. The network is the CIGRE distribution test system [4].

B. SIMULATION STUDIES FOR FREQUENCY REGULATION OF ISLANDED MG
Based on statistical data of [4], the most severe contingency that the MG may experiences is 20 kW power shortage after an unplanned islanding. The SOFC and MT model for frequency regulation studies can be found in [39] while the DEG model is adopted from model which is presented in [40]. For modelling the PV systems, the available PV models in Simulink/Matlab are used. The type of PV panels is Sun-power SPR-305E-WHT-D. The installed ESS whether HESS or BESS, is modeled based on the model and controllers presented in section 2.A. The power injection of the BESS whether using LAB or LIB is assumed to be similar. This assumption is also considered for the HESS. Therefore, the simulations are performed once with BESS and once with HESS.

1) FREQUENCY REGULATION USING BESS
In grid-connected mode, before islanding, the total demand and generation of the MG at specific time (5th second) are 64.1 kW and 44.1 kW, respectively. Therefore, 20 kW additional power is needed to compensate for the shortcomings in the islanded mode. Fig. 8 and 9. present the DGs power generation and MG frequency in islanded mode. It is obvious the frequency is dropped present to power shortage. To maintain the frequency deviation within a safe range, the BESS injects 28.61 kW immediately. In this way, the optimization algorithm (GA) tuned K1 at 57.334. considering the fast characteristics of the LIBESS in PFC, the frequency is intercepted at 49.501 Hz. However, this value is above the lower frequency limit (49.500 Hz). The slower-responding DGs, i.e., the SOFC, MT, and DEG gradually increase their power generation. The frequency returns to 50.000 Hz, and the LIBESS power returns to zero.

2) FREQUENCY REGULATION USING HESS
The Similar conditions for power shortage and it's disturbance as occurred in section 3.B.1 is applied in this section. Fig. 10 shows that after power shortage occurrence, the MG frequency drops. It can be seen in Fig. 11 that the SC and HESS provide a fast power injection with the peak power of 27.54 kW and 28.62 kW, respectively, and the frequency drop is intercepted at 49.501 Hz. For this purpose, K 1 is tuned at 55.19. The difference between the peak power of the SC and HESS is due to the slight contribution of the battery to PFC. The performance of the SOFC, DEG and MT restores frequency and the battery power to 50 Hz and 0 kW, respectively. The SC injects energy of 5.6331 Wh from the moment it responds to the frequency deviation till the moment its power becomes zero. The battery injects maximum power of 16.26 kW and total energy of 44.9901 Wh.

C. SIMULATION STUDIES FOR ENERGY MANAGEMENT OF ISLANDED MG
In this section, the size of the LAB/LIB is determined for the daily energy management. The installed capacity of DGs may not be enough to provide the MG demand during some periods such as the peak period. Therefore, the installation of an ESS for energy management and maintaining the balance  between demand and supply is necessary for the islanded MG. As the islanding occurrence time and its duration is not known, the battery is sized for the energy management of the islanded MG for the duration of one day islanding. The mid-summer day is taken as the worst case for energy management, given that if the BESS can manage power balance in this case, it can certainly be used for energy management in less severe cases. For longer durations, the same procedure, which is presented for one day energy management, can be used. The data of MG load profile is not available in [4], [41]. Therefore, the proposed profile of MG load in the midsummer, which is shown in Fig. 12, is considered as the heaviest load profile that the MG experiences annually. The PVs power generation is also shown in this figure.
The model of the islanded MG and its components used for frequency regulation studies, cannot be used for energy management studies, because the simulation runtime is very long and the PC memory will overflow due to high volume of simulation data. Therefore, the models given in Figs. 6 and 7 are simulated in Simulink for energy management of the islanded MG with BESS and HESS, respectively. The MT and SOFC models presented in [42], and the model of DEG presented in [43] are used in the simulation studies for energy management.    Fig. 13 shows the total power generation of DGs and BESS power, and Fig. 14 shows the discharged energy of the BESS. It can be seen that during daily islanding in the mid-summer, the DGs mainly supply the loads. The BESS performs PFC and in peak-load periods that DGs reach their maximum capacity, it injects its stored energy to maintain balance between demand and supply. For this purpose, the BESS may go through several short and few long discharge/charge cycles. It should be noted that in peak-load periods, the DGs are controlled to generate their maximum power, so that the capacity of the BESS is used minimally, and Till t = 65655 sec, the DGs are successful in realizing these tasks. Till this moment, the total discharged/charged energy of the BESS is kept at zero, which means that the SOCLIB adjusts at SOC ref LIB . However, after this moment till t = 79501 sec, the MG load increases such that in order to maintain demand-supply balance, the BESS should also supply the demand while the DGs are generating their maximum power. It can be seen in Fig. 14 that the maximum discharged energy of the BESS in this period is much more than that of the previous discharge cycles. Therefore, the maximum discharged energy of the BESS in this period, determines the required capacity of the BESS for energy management, which is 61.2 kWh. In addition, the BESS maximum injected power is 32.66 kW. Fig. 15 shows the total power generation of the DGs and the power of the SC and battery. The control strategy of the DGs in energy management of the islanded MG using the HESS is the same as energy management using BESS, except that the battery just responds to low-frequency components of the frequency deviations. The SC deals with high frequency components of the MG frequency.

2) ENERGY MANAGEMENT WITH HESS
As can be seen in Fig. 16, the discharged energy curve of the battery. The duration of the major discharge cycle of the battery is 13113 seconds, and the battery discharged energy during this cycle, is 61.19 kWh. The maximum injected power of the battery during its major discharge cycle is 32.66 kW. The maximum discharge energy of the SC in all discharge cycles is 0.42 Wh, and the maximum charge energy in all charge cycles is 0.46 Wh.

IV. COST EVALUATION OF LIBESS, LABESS, LISHESS AND LASHESS
The maximum discharged energy of the BESS in the EMFC studies is 61.2 kWh, and therefore, the required battery capacity can be calculated using (3): = 75.56kWh (13) Considering that the maximum absorbed/injected power of the LIB is 32.66 kW, the value of RP BATCon is 36.29 kW, which is calculated using (6) and the value of RP Inv is In EMFC simulation studies with HESS, the maximum discharged energy of the battery is 61.19 kWh, and therefore, its required capacity can be calculated using (3): The SC discharged energy of 5.6331 Wh in shortage power scenario, and charged energy of 4.1201 Wh in surplus scenario. Using (10), the required capacity of the SC stack can be calculated as follows: 5.6331 0.9 × 0.9 + (4.1201 × 0.9 × 0.9) = 10.293Wh (17) Considering that the maximum absorbed/injected power of SC and battery are 27.54 kW and 32.66 kW, rated power of the inverter is equal to 32.66 kW. The values of RP SCCon and RP BATCon are 30.6 kW and 36.29 kW, which are calculated using (12). The cost of the LISHESS and LASHESS for a 10-year period can be achieved using (7) It can be seen that for a 10-year period operation in islanded MG, the HESSs impose less cost than BESSs. The LISHESS is cheaper (almost 11%) than LASHESS. Also, the cost of LIBESS is much less than that of LABESS. It seems that hybridizing the battery technologies with the SC results in considerable cost reduction and LIB is more cost-efficient than LAB. VOLUME 8, 2020

V. CONCLUSION
In recent years, the hybridization of the LABs and LIBs with the SCs has been proposed to prolong their life cycles and hence reduce operational cost. Therefore, in this paper, the cost effectiveness of their hybridization was investigated for islanded MG applications. For this purpose, an EMFC scheme was proposed that maintained the demand-supply balance and also kept the frequency within safe operational limits. In addition, the coordination of DGs and ESSs minimized the required capacities of the LAB and LIB for the energy management of the islanded MG. Using the simulation results and the proposed procedure for sizing and cost evaluation, their costs were determined. It was shown that the HESSs impose less cost than BESSs. The LISHESS is cheaper (almost 11%) than LASHESS, which is mostly because the cost per life cycles of the LIB is much less than that of the LAB. For the same reason, the cost of LIBESS is much less than that of LABESS. It can be summarized that hybridizing the battery technologies with the SC results in considerable cost reduction and LIB is more cost-efficient than LAB.
Although, the hybridization of LIB and LAB with SCs are more common than other HESS, the hybridization of other ESS elements with each other has also been proposed like superconducting magnetic energy storage-battery HESS. For future work, it would be interesting to also perform a techno-economic analysis between their single utilization and their hybrid utilization for ESS applications. For this purpose, the same approach which was used in this paper for LIBESS, LABESS, LISHESS and LASHESS, can be implemented.