A Comprehensive Review on Flywheel Energy Storage Systems: Survey on Electrical Machines, Power Electronics Converters, and Control Systems

Finding efficient and satisfactory energy storage systems (ESSs) is one of the main concerns in the industry. Flywheel energy storage system (FESS) is one of the most satisfactory energy storage which has lots of advantages such as high efficiency, long lifetime, scalability, high power density, fast dynamic, deep charging, and discharging capability. The above features are necessary for electric vehicles (EVs), railways, renewable energy systems, and microgrids. Also, electrical machines, power electronics converters, and control systems are the cores of energy transfer in FESS. Therefore, they have a critical role in determining efficiency, power rating, power factor, cost, angular velocity, and volume of FESS. So, in this study, the FESS configuration, including the flywheel (rotor), electrical machine, power electronics converter, control system, and bearing are reviewed, individually and comprehensively. Additionally, the mentioned components have been categorized to be a guide for future research. The investigated electrical machines are compared by Finite Element Analysis (FEA). Subsequently, our laboratory’s measurement results are reviewed experimentally showing the progress in the field of FESS, such as designing robust control algorithms and an Interior Permanent Magnet-Synchronous Reluctance Machine (IPM-SynRM) to use in FESS.

(UPS) [31], [32]. Accordingly, the FESS usages in EVs, UPS, and microgrids AC are shown in Fig. 2. Also, a laboratory model of the FESS which is from our laboratory is shown in Fig. 3. In different areas of grid storage such as power quality, frequency regulation, and balance, voltage sag control, and mitigation of grid voltage reduction, the flywheel system is widely used owing to its fast charging-discharging capability [33], [34]. Also, the system stability is disturbed by the emergence of renewable energy sources such as wind and solar energy. In wind energy systems, a flywheel system is used to correct wind oscillations and system frequency balance, while in solar energy systems, FESS is utilized in combination with batteries, which increases the system's efficiency and battery life [35].
The rotor's size and the rotation velocity affect the FESS's total energy. Moreover, the used electrical machine and its power factor affect the FESS's power rating and active power, respectively. Since FESS is often used in high rotation velocity, the chosen electrical machine must be eligible for this operating condition. One of the significant factors in choosing an ESS is cost; in FESS, the total cost relies on the electrical machine's cost. Moreover, efficiency is the other significant factor of these systems. The efficiency of the electrical machines has the most effect on the efficiency of FESS. Accordingly, the design of an electrical machine with lower losses while having high rotation velocity and cost-effectiveness is the priority in FESS. Hence, the electrical machine has a key role in determining the FESS's property and is the main core of transferring energy, which power electronics converters and control systems can aid it in a better mission. Thus, presenting a review to investigate the used electrical machines, power electronics converters, and control systems in FESS was necessary as a guide for the engineering community. Still, no review study has not been investigated the details of FESS configuration, especially electrical machines. i.e., [36] and [37] gives a general view of FESS but not a detailed study. Reference [38] is a review of the FESS usage in power systems and microgrids. In [39] and [40], compared electrical machines used in FESS are very limited. Thus, the following review study fills a gap in the literature, and the main contributions are: • Review of flywheel energy storage system configuration, separately and comprehensively, including different categories of electrical machines, power electronics converters, control system strategies, and bearings.
• Comparing the finite element analysis of the used electrical machines, regarding the usage metrics of the flywheel energy storage system.
• Review of the laboratory progress in the field of flywheel energy storage systems, including proposing robust control algorithms and designing the IPM-SynRM configuration.
In Fig. 4 a chronological diagram is shown by presenting the most effective and well-known research in each of the surveyed fields including electrical machines, power electronics converters, control systems, and bearings.
In this study, at first, the FESS configuration, including flywheel (rotor), electrical machine (motor/generator), power electronics converters, control systems, and bearings, are investigated comprehensively. Then, compared electrical machines are verified by the Finite Elements Analysis (FEA). Finally, the review of our laboratory works in the FESS field, including the design of a robust control system and Permanent Magnet Assisted Synchronous Reluctance Machine (PMa-SynRM) to use in FESS, is done. An overview of the review methodology and details of each section are displayed in Fig. 5. Fig. 6 demonstrates the main configuration of the FESS including a flywheel (rotor), electrical machine (motor/generator), power electronics converters, control system, and bearing. The FESS has three operating modes, including charging mode (storing energy), standby mode (keeping energy), and discharging mode (releasing energy). FESS receives energy from an electrical source, such as the grid in charging mode. In charging mode, the higher kinetic energy is yielded by increasing the flywheel's angular velocity (rotor), storing the electrical energy in the form of mechanical energy in the flywheel (rotor). In charging mode, the electrical machine acts in motor mode using a power electronics converter and control system, as shown in Fig. 7(a). In standby mode, the flywheel (rotor) must stay at an almost constant angular velocity, as shown in Fig. 7(b). In the discharge mode, the flywheel's angular velocity (rotor) goes down, and kinetic energy converts into electrical energy. In discharge mode, the electrical machine acts as a generator, supplying the grid or loads through the power electronics converters, as shown in Fig. 7(c). Accordingly, the electrical machine and its control system are the main core of transferring energy in FESS. Fig. 8 is displayed for comprehending the FESS operating modes, including charging (storage), standby (keeping), and discharge (releasing) [14]. In the following, the main configuration of the FESS is investigated individually, such as flywheel (rotor), electrical machine (motor/generator), power electronics converters, control system, and bearing. The changes in flywheel angular velocity and energy under different operating modes such as charging, standby, and discharging can be shown in Fig. 9.

A. FLYWHEEL (ROTOR)
A flywheel (rotor) is a rotating disk that stores mechanical energy. The stored energy in the flywheel is as follows: where E is the stored kinetic energy, J is the moment of inertia in the flywheel, and ω is its angular velocity [14], [41]. According to (1), energy can be higher by increasing the angular velocity or increasing the moment of inertia. It brings two choices for FESS, FESS with low rotation velocity (LSFESS) and FESS with high rotation velocity (HSFESS). The LSFESS is made from steel and has more weight with less cost than HSFESS [42], [43], [44]. The flywheel's size, form, and mass (rotor) affect the moment of inertia. In general, the rotor is a solid cylindrical object. J is written as below: where m is mass, r is the radius, a is the length, and ρ is mass density. it is noteworthy that, more energy can be stored by using materials with higher mass density or by increasing the disk radius [14]. The flywheel's function can be simplified as follows: where T em is the electromechanical torque and f is the friction coefficient.

B. ELECTRICAL MACHINE (MOTOR/GENERATOR)
The electrical machine has a key role in FESS. Using a control system appropriately transfers energy between the motor and generator modes. The electrical machine is an electromechanical interface in FESS that converts electrical energy into mechanical energy and vice versa [45]. In storing mode, the electrical machine must act as a motor, and by increasing the flywheel's angular velocity (rotor), the electrical energy will be stored as kinetic energy. In releasing mode, the electrical machine must act as a generator; so, by decreasing the flywheel's angular velocity (rotor), the kinetic energy converts into electrical energy [14], [39], [40], [46], [47]. The electrical machine's size and cost rely on T em . The rotation velocity range of the flywheel (rotor) must be between a minimum and maximum value, which causes a satisfactory value of T em . The capacity of storing or releasing energy by FESS can be shown as follows: While ω min is usually set between one third and one half of ω max . Choosing ω min = 1/3ω max makes the electrical machine 50% larger than when ω min = 1/2ω max , but more energy extraction is yielded from the flywheel (rotor). So, a trade-off must be chosen in the angular velocity range. Keeping this fact in mind that, for low-power machines, as the flywheel (rotor) brings the highest cost to the FESS, a system with lower ω min will be chosen for increasing the usable capacity. Correspondingly, the contrary is true for high-power machines [14]. In HSFESS, both the electrical machine and the flywheel are fully integrated. In LSFESS, they are partially integrated into the housing box or will be employed separately [27]. In Table 1, the LSFESS and HSFESS are investigated against each other [14], [27], [44], [48], [49], [50].

C. POWER ELECTRONICS CONVERTERS AND CONTROL SYSTEMS
Power electronics converters are the link between the electrical machine and the electrical load/supply, which play a significant role in the satisfactory performance of electrical machines. Using control of the electrical machine, energy transfers between the FESS and the electrical load/supply [1], [36], [51]. Increasing the switching frequency in power electronics converters has advantages, such as increasing the rectifier control bandwidth and reducing Current Ripple (I ripple ). On the other hand, increasing the switching losses is often a disadvantage [44], [52]. The electrical machine acts as a generator or a motor based on the situation by controlling power electronics converters. The most common power electronics converters used in the FESS to control electrical machines are AC-DC-AC   [14], [27], [44], [38], [49], [50].
(back-to-back) and AC-AC converters. Back-to-Back (BTB) converters contain a DC link and two converters on both sides. These converters are Machine-Side Converters (MSC) and Grid-Side Converters (GSC) [53], [54], [55], [56], [57]. In BTB converters, the GSC converts AC voltage to DC voltage, and then MSC converts DC voltage to an arbitrary AC voltage and frequency. A two-level BTB is shown in Fig. 10(a). The MSC controls the angular velocity and flux, owing to control of the FESS active power; subsequently, the GSC controls the DC link voltage [36]. In BTB converters, the control of switches is often by the Pulse Width Modulation (PWM) strategy. The PWM strategy uses Rectangular Pulse Waves (RPWV) and adjusts their width. The RPWV controls the converter in generating an AC voltage sinusoidal from the DC link, leading to control of the electrical machines for operating appropriately in the FESS [35], [44], [58]. Adding a DC-DC stage between the AC-DC and DC-AC stages obtains more satisfactory operating conditions in the BTB converters. Accordingly, as shown in Fig. 10(b), a boost converter can be used to adjust and boosts the voltage when the angular velocity of the flywheel is low [59], [60]. Owing to the high voltage challenges in conventional two-level BTB converters, multilevel converters such as Neutral Point Clamped (NPC) can be used, which gives advantages such as higher efficiency, the smaller size of filter elements, lower harmonic orders, lower voltage stress on switches and the lower common-mode voltages. On the contrary, the higher number of switches will bring higher costs. Fig. 10(c) shows the NPC three-level BTB used in FESS. In the NPC three-level BTB converter, besides the DC-link voltage control, the voltage balance control of capacitors is also necessary [61], [62]. The BTB converter topologies have disadvantages owing to the DC-link capacitor, which brings a shorter system lifetime and makes the converter bulky and heavy, increasing maintenance costs. Accordingly, the AC-AC converters or Direct Matrix Converter (DMC), without a DC-link capacitor and consequently no large size, can be regarded to overcome the above challenges. The DMC is an array of nine switches bidirectionally arranged to give the connection capability between output and input, one by one. The matrix converters have disadvantages such as the voltage gain constraint, high Total Harmonic Distortion (THD), and more control complexity [63], [64], [65]. The matrix converters can be categorized into Direct MC (DMC) and Indirect Matrix Converters (IMC). Therefore, to overcome the voltage gain constraint, IMCs can be regarded. The IMC and DMC are shown in Fig. 10(d) and Fig. 10(e), respectively. In IMCs, the voltage gain will be boosted in reverse mode when the generator's output voltage is low, owing to the low angular velocity of the flywheel [66], [67]. Other converter categories have been used to control the drive of electrical machines in FESS, including Z-source Converter (ZSC) [45], [68], [69], boost DC/AC converter [70], and DC-DC Plus NPC (P-NPC) [62]. Whereas the most used converters for electrical VOLUME 11, 2023  machines' drives in FESS are; two-level BTB, two-level BTB Plus Boost (P-Boost), NPC three-level BTB, DMC, and IMC. Comparing the above converters is displayed in Table 2 [39], [40].
In practical operations, the FESS must demonstrate a superior level of performance, including a smooth and rapid charging process, effective dynamic response capability through the discharging process, and anti-interference ability during both the charging and discharging process. Hence, The power electronic converter's control performance plays a critical role in achieving these goals [59]. Generally, most of the Variable Frequency Drive (VFD) control strategies can be employed to apply MSC control in FESS [25], [26], [71], [72], [73], [74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84], [85], [86], [87], [88], [89], [90], [91], [92], [93], [94], [95], [96], [97]. However, designing controllers for FESS requires taking into account some differences and peculiarities compared to VFD control. One of the challenges in FESS control is maintaining the desired voltage for the load during discharging mode, which involves dealing with nonlinearities, disturbances, and uncertain factors. The DC-link controller needs to be robust against these factors to ensure consistent performance. While most literature focuses on charging control strategies for FESS, there is limited work on discharging control techniques. For example, in reference [59], the main contribution of the authors was to study the FESS discharge control of the Buck-boost converter But, considering the wide-range speed variation effect, which is crucial to determine the energy storage capacity and discharge depth of a FESS, has been ignored. In [79], a discharge strategy that can consider the wide range of speed variation and maintain robust discharge performance is presented. However, this strategy does not control the inner current loop. To consider consistent fast dc-link voltage dynamic performance for the HSFESS, a fast dc-link voltage strategy based on a global linearization model and a linear extended state observer is proposed in [78]. In FESS, since the rotation velocity of the flywheel (rotor) is very high, especially in systems with magnetic bearings, sensorless control is more attractive; without the challenges of rotor angle extraction in the higher rotation velocity. Moreover, while fast charging and discharging are out, it is necessary to use smooth transition control between charging and discharging to avoid disturbances in transients. Accordingly, the most common strategies of MSC control are Field-Oriented Control (FOC), Direct Torque Control (DTC), Model Predictive Control (MPC), and intelligent control. Comparing the above control strategies is displayed in Table 3, according to the electrical machine and power electronics converter. Based on Table 3, the FOC control strategy, the two-level BTB, and IM or PMSM have more employed [40].

D. BEARINGS
Bearings are responsible for providing minimum friction drag for the electrical machine and the flywheel (rotor) [98]. Friction, loss, and cost tightly rely on the bearings' design. So, a weak bearing design brings more friction, losses, and more maintenance costs [14], [37], [99], [100]. In general, the bearings can be categorized as mechanical and magnetic. Mechanical bearings are usually employed for LSFESS with low rotation velocity, giving advantages such as lower initial cost and easy mounting. On the contrary, it brings some disadvantages, such as high friction and losses, requiring lubrication and maintenance, and a shorter lifetime [14], [38], [48], [101], [102]. A rolling element bearing is shown in Fig. 11(a) as a mechanical bearing. Although rolling element bearing friction is low over the other mechanical bearings, leading to higher losses. So, rolling element bearings are usually employed in conjunction with magnetic bearings. After the emerging magnetic bearings, the FESS industry has fundamentally changed. Magnetic bearings have advantages such as a longer lifetime, higher rotation velocity, higher load capacity, and higher dynamics. Nevertheless, it brings more complexity to the control system [103], [104]. Magnetic bearings are categorized as Passive Permanent Magnetic Bearings (PMB), Active Magnetic Bearings (AMB), and Superconducting Magnetic Bearings (SMB).

1) PERMANENT PASSIVE MAGNETIC BEARINGS (PMB)
A PMB system has no external controls and consists of PMs, which gives the rotation capability to the flywheel (rotor). On the contrary, it should be employed along with other bearings owing to low stability. PMB has the advantages of high stiffness, low cost, and low losses owing to the lack of a current. Although it has low damping capability and no active control [105], [106], [107], [108]. It can be categorized as Repulsive Passive Permanent Magnetic Bearing (R-PMB) and Attractive Passive Permanent Magnetic Bearing (A-PMB); in the axial or radial installation. A PMB consists of two annular magnets, which are magnetized in a coaxial direction, as shown in Fig. 11(b). It is stable since reducing the gap between them increases the force. However, the bearing is unstable in radial mounting. So, requiring the radial bearings with radial stiffness greater than the anti-stiffness of the axial bearing. Using two rolling element bearings to maintain the radial location of a vertical shaft and use an R-PMB to handle the weight. On the other hand, an A-PMB can eliminate the second magnet that would have to be used on the rotor, in the case of R-PMB, as shown in Fig. 11(c). Although the radial anti-stiffness is also lower, this bearing is unstable, given the increasing axial attractive force with a reducing gap. The A-PMB tends to work best in conjunction with rolling element bearings, which can also take a share of the axial load. The radial PMB, Fig. 11(d), has stacks of annular magnets on both the shaft and the stator, while the fields of all magnets are aligned radially. Accordingly, the annular magnets in the gap keep the shaft centrally aligned. On the other hand, the shaft will move axially without axial constraints, leading to instability [14], [109].

2) ACTIVE MAGNETIC BEARINGS (AMB)
AMB has windings whose currents can adjust the amount of electromagnetic force in the system, reducing rotor vibrations and excluding their losses. Adjusting is through a rotor angle feedback controller. A radial AMB is shown in Fig. 11(e). The bearing restoration force is controlled by changing the current in the stationary coils. Although it has control complexity, using such bearings is attractive, given bearing maintenance can be significantly lower against the rolling element bearings that suffer from wear and fatigue. The axial AMB can also be used, Fig. 11(f), requiring coils on either side of a disc that can only attract the disc. So, it works similarly to the radial one, the attraction of the side that is moving away [14]. AMB has the advantages of control ability, high stiffness, and long VOLUME 11, 2023 lifetime, whereas it has disadvantages such as control system complexity, high cost, and losses owing to the current of the control system [14], [110], [111], [112], [113]. Using AMB and mechanical bearings simultaneously gives lower control complexity and more stability and affordability [39], [114].

3) SUPERCONDUCTING MAGNETIC BEARINGS (SMB)
SMB is based on superconductivity; Fig. 11(g) and 11(h) are for radial and axial mountings. Here, magnetic fields create currents in superconducting coils in the stator, repelling PMs on the rotor [14]. The SMB gives advantages such as high rotation velocity, no friction, long lifetime, less size, and stability. Accordingly, SMBs are eligible for HSFESS. The major disadvantage of the SMB is the necessity of a cryogenic cooling system to avoid the bearings' failure, which brings more cost to the SMB system. High-temperature superconductors can be employed to overcome the cooling system challenge in SMBs. Additionally, SMB and PMB can be used simultaneously to have a lower necessity for the cooling system and be more costeffective [14], [115], [116], [117], [118], [119], [120]. Comparing the bearing categories is displayed in Table 4. Fig. 12 displays the main configuration of the FESS.

III. ELECTRICAL MACHINES CATEGORIES IN FESS
The following section reviews the electrical machines employed in FESSs. Whenever the electrical machine acts as a motor, the energy transferring, charging, and storing in the FESS will be yielded. On the other hand, when the FESS is discharging, the machine acts as a generator [37], [39]. Accordingly, the electrical machine has a key role in electrical and mechanical energy conversion and significantly affects  [39], [40].

A. INDUCTION MACHINE (IM)
Since IM has a rigid configuration, low cost, high T em , strength, and reliability, it can be regarded as the best choice for high-power usages [35], [123], [124]. On the contrary, the main challenges of IMs are rotation velocity constraints, the control drive complexity, and numerous costly maintenances [108]. The IM generally has lower efficiency than the PMSMs [125] and is not eligible for the HSFESS [91]. Accordingly, in [126], a new design model has been suggested to lower losses in FESS usage. Double-Fed Induction Machines (DFIMs) are also suggested to overcome the low rotation velocity disadvantage [117], [127], [128], [129], [130]. Moreover, DFIMs have advantages, including flexible control and low conversion rate, allowing them to have a smaller converter's size [131]. Although the IM has severe constraints, particularly in HSFESS, according to economic efficiency, it is used in LSFESS, and the DFIM is more attractive for HSFESS.

B. PERMANENT MAGNET SYNCHRONOUS MACHINE (PMSM)
PMSMs are satisfactory in FESS owing to their high power density and efficiency [35], [132], [133], [134], without requiring an external excitation current due to the magnetic flux of PMs. Hence, PMSM gives a higher power factor, lower losses, and causes lower the necessity for Volt-Ampere Inverters (VAIs). However, PMSMs are subject to demagnetization due to having magnets, leading to more sensitivity and vulnerability to heat than IMs [24]. In the PMS motor/generator (PMSM/G), the stator core's iron loss is owing to the magnetic field change in the idling mode, which causes slows down the PMSM/G [135]. In general, PMSMs are used in the industry with high rotation velocity necessity [20], [114], [136], [137], [138], [139], [140]. Although it is used at high rotation velocity, the rotor's mechanical behavior has complexity due to the existence of magnets in the rotor configuration. References [124] and [141] are given an overview of PMSM design concerns to use in the FESS, especially for lower iron loss. Seok-Myeong Jang designed an iron lossless PMSM for high-power FESS [143]. In [144], the operating range of a Double-side Permanent Magnet Synchronous Motor/Generator (DPMSM/G) is investigated to use in a FESS. In [145], the study on FESS design and assessment is on reducing motor Torque Ripple (T ripple ) to ensure output power stability. In [146], a DPMSM/G is investigated in the FESS to lower the iron losses in idling mode.

C. SWITCHED RELUCTANCE MACHINE (SRM)
SRM has advantages, such as configuration simplicity and rigidness [147], [148] and less iron loss, as well as the ability to work in harsh environments and operating conditions [149]. Further, it has a wide rotation velocity range, high acceleration capability, high efficiency, simplicity in converter circuit design, and fault-tolerant ability [37]. Although the SRM has no magnet [150], [151], [152], its efficiency is analogous to and even higher than that of the IMs at high rotation velocity. Additionally, SRM is controlled easier at high rotation velocity than IMs [153]. In general, the absence of magnets in the rotor configuration of SRM causes core losses in the stator core, especially at high rotation velocity, much less than in a PM machine. Hence, the efficiency of this machine is relatively high. Contrary to all the SRM's advantages and its limited usage in the FESS [154], [155], [156], it has a high T ripple and is not eligible for this aim.

D. SYNCHRONOUS RELUCTANCE MACHINE (SynRM)
SynRMs are employed as motors/generators in FESS due to zero spinning losses when the machine generates no T em . Moreover, the design of SynRM's rotor can have satisfactory integrity in construction if the rotor saliency is made by alternating layers of magnetic and non-magnetic metals, which are bonded together by a high-strength bonding method [153]. The simplicity of configuration [157], high efficiency, high power capability, and low VAI are the most significant design concerns of motor/generator in FESS [158]. Ordinarily, in SynRM, power factor, rotation velocity range at constant power region, and dynamic of the motor rely directly on the saliency ratio, which can be used for proposing a motor design optimally [159]. Different rotor designs in SynRM have been suggested to earn a satisfactory saliency ratio, including the single barrier rotor, the axially-laminated rotor, and the transversally-laminated rotor. Fig. 13 shows the SynRM rotor configurations. The single barrier type has a low saliency ratio, weakening its operating condition. The axially-laminated type is the best choice regarding the saliency ratio, but due to the layered configuration of the rotor, the eddy current losses will be high. The transversally-laminated type is analogous to the IM in manufacturing and is satisfactory for high rotation velocity owing to the strength of the rotor. In the rotor configurations of the above types of SynRM, using radial and tangential blades ensures mechanical strength at high rotation velocity. On the contrary, increasing the blades' width, especially at high rotation velocity, grows the leakage flux and the magnetic current. Accordingly, in SynRM, a high saliency ratio causes a high power factor, reducing losses and the rated VA of the inverter for the electrical machine [139], [157].

E. BEARINGLESS MACHINE
Using a mechanical bearing in the FESS suffers challenges such as friction, wear, tear, and dust contaminations, limiting the system's rotation velocity and useful life [160], [161]. These challenges can be solved by using magnetic bearings, but the increasing size and cost of the whole system will yield. Therefore, electrical machines without bearings are more attractive. The FESS, which combines two modes, including magnetic gear and magnetic bearing, in one machine, uses a bearingless machine [162], [163], [164]. Conventional bearingless electrical machines such as Bearingless Permanent Magnet Synchronous Machines (BPMSM), Bearingless Switched Reluctance Machine (BSRM), and Bearingless Brushless Direct Current Machine (BBLDCM) are employed in the FESS. However, the rotation velocity regulation of BPMSM is not satisfactory [162], and an external excitation system for BSRM is necessary, which certainly makes the cost and complexity higher [164], [165]. BBLDCM has advantages such as high rotation velocity, high reliability, wide rotation velocity regulation range, no mechanical friction, and no excitation, which make it attractive for FESS [166]. According to the above BBLDCM's advantages, its usage in the FESS is regarded [167]. Accordingly, in [168], an outer rotor BBLDCM is suggested for the FESS.

F. BRUSHLESS DC MACHINE (BLDCM)
BLDCMs are highly efficient, with many advantages such as high power density, high efficiency, relatively wide rotation velocity range, mechanical stability, low maintenance cost, and no Electromagnetic Interference (EMI) [59], [115], [169], [170], [171], [172]. However, BLDCM no-load losses are usually significant at high rotation velocity. Moreover, unbalanced forces during no-load bring high necessity conditions and obligations to the bearing capacity and stiffness of the magnetic bearing system. Ironless BLDCMs can overcome the above disadvantages and, owing to their high operating efficiency and low standby loss, are regarded as one of the best choices for FESS [173].

G. HOMOPOLAR MACHINE (HM)
The ACHM is categorized as a Homopolar Synchronous Machine (HSM) and Homopolar Inductor Alternator (HIA) [174], [175], [176]. The HM can offer advantages such as rigid rotor configuration, low idling losses, and high reliability, which make it significantly attractive for the high rotation velocity range [175], [176], [177], [178], [179], [180], [181]. Using the ACHM technology, the self-discharge will be lower; hence, the efficiency and energy density will be higher [180]. So, due to high efficiency, high rotation velocity range, and high energy density, ACHM is satisfactory to use in FESS [182], [183], [184]. Additionally, in the ACHM, the electromagnetic losses can be mitigated by disconnecting the excitation current when the FESS is idling, which aids the energy storage efficiency of the FESS, especially in long-term operating conditions [174]. The ACHM still suffers from disadvantages, including a unipolar air-gap flux density; thus, its power density is practically half of an electrical machine with a bipolar air-gap flux density [180]. In addition, all magnetic flux passing is axially through the rotor, so the air gap flux will be limited by the rotor diameter owing to the saturation of the iron core [185]. Since the rotor's angular velocity limits the rotor's diameter, the ACHM rotor's length and diameter are limited, limiting the machine's power and energy storage [125], [186]. An outer rotor ACHM with a FESS is suggested in [180] to have better energy storage capacity. Specifically, in [178], the electromagnetic function of a 30 KW ACHM is investigated and tested.
Accordingly, the summary of the above-investigated electrical machines' advantages and disadvantages is given in Table 5. Consequently, in the following section, these electrical machines will be investigated and analyzed comprehensively using the FEA by considering the FESS constraints.

IV. COMPARATIVE STUDY OF THE SUPERIOR EMPLOYED MOTORS IN FESS
Based on the above review and Table 5, 1PMSM, SMPMSM, and SynRM have superiority over the other electrical machines to use in FESS, owing to their advantages over the other categories. Accordingly, in the following, the IPMSM, SMPMSM, and SynRM will be comprehensively investigated. Since the IM will not be satisfactory in HSFESS, and 81238 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE. Restrictions apply.  SRM has a very high T ripple , which makes their usage in the FESS unsatisfactory, IM and SRM will not be regarded in the following investigations.
The topologies of the investigated electrical machines are shown in Fig. 14, which are analogous in Phase number, Pole number, air gap length of 0.48 mm, the materials in the laminations of stators and rotors, and the stators' configurations with the same outer and inner radius and the same number of slots. All the magnetic field analyses are the same in the stator current and the windings' fill factor. The stator VOLUME 11, 2023 81239 Authorized licensed use limited to the terms of the applicable license agreement with IEEE. Restrictions apply.  configurations' data of the investigated electrical machine models are given in Table 6.

A. COMPARING IPMSM, SMPMS, AND SynRM 1) FLUX DENSITY
The magnetic field density in the stator and rotor cores and the air-gap flux density are significant for the examination of the motors. The Permanent Magnet Synchronous Machines configurations are commonly categorized into IPMSM and SMPMSM. The same PM material (N30UH 1 ) with B r of 1125 mT is utilized in the above configurations' investigations [187]. The flux density of the above configurations is not analogous, owing to the dissimilarity in the magnet's arrangement and the leakage PM flux. Fig. 15 shows the flux density in the rotor and stator of IPMSM, SMPMSM, and SynRM. According to Fig. 16, which shows the load air-gap magnetic flux density of the above electrical machines, it can be seen that the stator current is more effective owing to the effective smaller air-gap in the IPMSM, besides the higher 1 It belongs to the group of NdFeB magnets.   maximum flux density in the air-gap than the SMPMSM. Moreover, in the SynRM, the maximum flux density in the air gap is higher than that of the SMPMSM and lower than the IPMSM. Increasing the flux density at the ribs of SynRM and PMSM is shown in Fig. 15, leading to the lower T ripple [188]. According to Table 7, the average flux density in the air gap of the IPMSM is lower than the SMPMSM, which is due to the higher magnet leakage flux of the IPMSM.

2) TORQUE CHARACTERISTICS
According to Fig. 17 and Table 7, the Average Torque (T avg ) of the IPMSM is 3.07% and 31.06% higher than that of the SMPMSM and SynRM, respectively. Considering the analogous constant angular velocity for operating conditions of IPMSM, SMPMSM, and SynRM give the same order of the output power (P out ) as that of the T avg . The lowest and the highest T ripple belong to the SMPMSM and the SynRM, respectively, since SMPMSM has no Reluctance Torque (T reluctance ) and SynRM has only T reluctance .

3) EFFICIENCY AND POWER FACTOR
Determination of the stator and rotor iron losses (P core ) along with the magnet losses (P PM ) and copper winding losses (P cu ) gives total losses in load mode, as displayed in Table 7, and the efficiency of each motor is written as below: η = P out P out + P core + P cu + P PM (5) According to Table 7, the efficiency and power factor of IPMSM and SMPMSM are practically the same. On the contrary, the SynRM has lower efficiency and power factor than the PMS motors. Accordingly, the examinations endeavor to get a higher power factor in SynRM by optimizing the salience factor. Fig. 18 is the efficiency map under the Maximum Torque Per Ampere (MTPA) control strategy, displaying that the SMPMSM has the highest maximum efficiency.

4) FLUX-WEAKENING CAPABILITY
The optimal operation of the electric motor at a high rotation velocity range is one of the essential required features for FESS. Comparing the T em and P out at a wide rotation velocity range of the investigated motors is displayed in Fig. 19. Accordingly, the IPMSM gives more than 30 kW at 60,000 RPM, whereas SMPMSM has less than 10 kW and SynRM of about 22 kW. Thus, SMPMSMs are not satisfactory for FESS, requiring a Constant-Power Wide-Speed Range (CPWSR). Moreover, the SynRM is not inherently satisfactory to use in CPWSR, adding PMs in flux barriers will overcome this challenge [189].
Additionally, the SynRM has weaker magnetic characteristics, higher reliability, and easier maintenance than PMSM [190], [191], owing to the low heat in the winding and bearings and its simplicity, robustness, and lack of PM in rotor configuration. Besides the above merits, the manufacturing cost of a SynRM is much lower than PMS motors [192]. In hot areas and when the use of cooling systems is challenging, using the PMSMs has a high risk of demagnetization. The PMs in SMPMSMs should be fixed on the rotor surface against high centrifugal forces, which makes it unsatisfactory, especially in HSFESS. Consequently, in SMPMSMs, rotor sleeves are designed with high-strength alloys to overcome the above challenge, increasing the manufacturing cost of SMPMSM in HS-FESS. Accordingly, the SMPMSM is not satisfactory to use in HSFESS. Comparing the PMa-SynRM with the IPMSM is given in the following.

B. COMPARING PMa-SynRM AND IPMSM
According to the above analysis, the comparison of the PMa-SynRM with V-type IPMSM and SynRM is shown in the following. Due to the dissimilarity in operating conditions of PMa-SynRM in different PMs materials, two PMs materials, including ferrite and N30UH, have been regarded in this study. The topology of the designed PMa-SynRMs is displayed in Fig. 20.

1) FLUX DENSITY
According to Fig. 21, which displays the on-load air-gap magnetic flux density of IPMSM, SynRM, and two types of PMa-SynRM with ferrite and N30UH PMs materials, owing to the more significant effective air-gap in the PMa-SynRM, it has a lower air-gap flux density than IPMSM with N30UH PM material. Moreover, the flux density in the air gap of the two types of PMa-SynRMs is analogous to the SynRM. Fig. 22 displays the on-load flux density in the rotor and stator of two types of PMa-SynRM. According to Table 8, the average flux density in the air gap of PMa-SynRM with N30UH is higher than that of SynRM.  According to Fig. 23 and Table 8, increasing the T avg and the torque density of both PMa-SynRMs is more significant than the SynRM configuration. Increasing the T avg in the PMa-SynRM with PM N30UH and ferrite is 30% and 9%, respectively, more than T avg of the SynRM, which is owing to the PM's torque term. Moreover, according to Table 8, adding PM makes the motor's torque density higher, which is more in the N30UH PM type than ferrite. By adding ferrite to the SynRM, T ripple is better by 7%, and on the contrary, by adding N30UH, T ripple is worse by 14%. Owing to that, the higher the magnetic flux between the rotor and the stator, the higher the T ripple . Consequently, when the rare earth magnet is utilized in the SynRM, the magnetic flux will grow, and the T ripple will be higher. In general, by adding PM to the SynRM, the T avg of PMa-SynRM will be analogous to IPMSM.

3) EFFICIENCY AND POWER FACTOR
According to Table 8, increasing the efficiency of both PMa-SynRMs can be seen by 4.7% and 1.3% in PMa-SynRM with N30UH and ferrite more than SynRM efficiency, respectively. Moreover, increasing the power factor of PMa-SynRM with N30UH and ferrite is 37.4% and 7.1% more than SynRM, respectively. In Fig. 24, the efficiency maps under the MTPA control strategy are displayed, in which the  PMa-SynRM with N30UH has the highest maximum efficiency.

4) HIGH-SPEED FLUX-WEAKENING CAPABILITY
Comparing the T em and P out of the investigated motors in Fig. 25, including IPMSM, SynRM, and two types of PMa-SynRM with ferrite and N30UH PMs materials, shows that, at high rotation velocity range, both PMa-SynRMs have a better operating point than that of the SynRM. Accordingly, the PMa-SynRM with N30UH PMs materials gives an analogous P out with IPMSM at 60000 RPM. Whereas PMa-SynRM with ferrite has less P out than PMa-SynRM with N30UH and more P out than SynRM. Thus, adding PM to SynRM as PMa-SynRM gives a better operating point in CPWSR than SynRM. Also, the higher value of the  (B × H ) Max for the used PM in PMa-SynRM gives a better P out and T em in HSFESS.

5) COST
Cost is one of the most significant factors in choosing an electrical machine to use in FESS. Therefore, the cost of the used materials in the investigated PM motors, including PMa-SynRM with ferrite, PMa-SynRM N30UH, and IPMSM, is given in Table 9. The unit costs for iron lamination, copper, ferrite, and N30UH magnet are regarded to be 1.45 $/kg [193], 7.5915 $/kg [194], 3.2 $/kg [195], and 126.2 $/kg [195], respectively. Accordingly, the lowest cost is for PMa-SynRM with ferrite, followed by PMa-SynRM with N30UH, and the highest one is for the IPMSM configuration. Generally, according to Fig. 26, PMa-SynRM is better than SynRM and is satisfactorily close to IPMSM. On the hand of economic concerns, SynRM can be a satisfactory item. But based on electrical characteristics,  IPMSM is more satisfactory. Owing to the lower cost and the higher reliability of the PMa-SynRM, it will have a high superiority over the other rivals, if both metrics are regarded.
Given that energy efficiency has a direct and indirect effect on the operating condition of the system and its cost, respectively, it is one of the key metrics for the assessment of an energy storage system [38]. So, high efficiency can be regarded as an attribute of usable capacity, since the former leads to the latter. Accordingly, the main losses that must be regarded in the FESS are bearings' friction losses, wind losses around the rotating elements (rotor and flywheel), windings losses, iron losses in the electric machine (stator and rotor), vacuum leakage losses, serration losses due to magnetic float and losses in power electronics converter [38]. In Fig. 27, a summary overview of efficiency calculation and different losses in the flywheel system is displayed.

V. THE LABORATORY PROGRESS IN THE FIELD OF FESS
Our laboratory investigation and progress in the field of FESS such as introducing robust control strategies [72], [87] and the new design of PMa-SynRM [197] are reviewed in the following section, which can be regarded as a guide for the engineering community.

A. IMPLEMENTING AND DESIGNING OF A CONTROL SYSTEM WITH THE ABILITY TO WITHSTAND MEASUREMENT ERROR FOR THE FESS [72]
Proposing a control system for the FESS with PMa-SynRM, which can withstand measurement error is a big honor. The PMa-SynRM inherently has a T ripple . Moreover, the lack of satisfactory design and switching algorithms can cause more mechanical damage.
Accordingly, in [72] an accurate digital controller is systematically designed by the FOC algorithm, as shown in  It was shown that if no error is in the measurement, enhancing the motor's acceleration and remaining within the nominal operating range of T em will be obtained, through the designed controllers. Thus, it can be argued that the designed controller can be employed optimally in the FESS for continuous charging and discharging in this system. The test bench is displayed in Fig. 29. According to outcomes in [72], it can be seen that although the control system is designed precisely, measurement error significantly affects the system. The dead zone and measurement offset inject the non-ideality in the current individually, leading to the oscillation of rotor velocity. In the worst case, if the offset and dead zone are in the current sensor simultaneously, the motor stability will fail. Hence, the function of the designed controller relies on the accuracy of the analog measurement. Accordingly, the design and use of High Pass Frequency (HPF) and Finite Impulse Response (FIR) filters appropriately in the designed control system can overcome the offset and dead zone challenges. Details of the employed filters are given in Table 10. Moreover, by implementing the designed filters in the drive system, the offset and dead zone errors are mitigated, and it gives lower rotor velocity oscillations. The suggested filters have advantages, such as accessibility and simplicity in design and usage. In Table 11, comparing the suggested strategy with  others is displayed. Technical data of PMa-SynRM and its prototype configuration separately are given and displayed in Table 12 and Fig. 30, respectively.
In order to verify the suggested controller, with 2.4 kW PMa-SynRM and using the DSP TMS320F28335, the measurement results are experimentally investigated. An Omron Encoder E6B2-CWZ6C with a resolution of 2000 is the rotor angle and angular velocity sensor. Moreover, the test bench is shown in Fig. 29. Accordingly, the measurement of the angular velocity and stator currents are shown in Fig. 31. According to Fig. 31(a), the magnitude of the stator current at the first start is high to reach the base angular velocity. VOLUME 11, 2023 [87].
Then, the stator current will lower. Additionally, as shown in Fig. 31(b), the motor's acceleration rate is satisfactory and reaches its steady value with a negligible steady-state error. Given that the T em of the motor is 3.9 Nm, the reference 81246 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE. Restrictions apply.  acceleration rate is written as T e j = dω r dt , which must be about 95 rad s 2 . Fig. 32 shows the measurement of I dq . After reaching the rotor angular velocity to the base rate, I dq lower to near zero.  [87] Owing to the non-zero current in the d-axis, the non-linear dynamic is the main challenge of the PMa-SynRM drive system. Accordingly, in [87], a linear dynamic for the PMa-SynRM is suggested, which solves the complexity of designing non-linear methods. So, the augmented linear model can be employed to design an MPC for the PMa-SynRM drive system [87]. In CCS-MPC, according to the system's dynamic, the control signals are generated. On the other hand, in the FCS-MPC, a switching mode is chosen among the number of switching states. Moreover, a modulator gives a fixed switching frequency, to generate control signals in CCS-MPC. It should be noted that using CCS-MPC leads to less T ripple and lower steady-state error in SRMs. The designed CCS-MPC algorithm to control PMa-SynRM, using an augmented linearized model, is shown in Fig. 33.
Using an MPC will give faster and more accurate dynamics, more acceleration rate, fewer oscillations, more effective torque, less angular velocity dropping, and less recovery time and steady-state error than the Proportional-Integral (PI) VOLUME 11, 2023   [197].
controllers, which makes the MPC more eligible for FESS. In [87], the effectiveness and fast dynamics of the designed PMa-SynRM control strategy for FESS are displayed experimentally against the PI controller and FCS-MPC. Comparing the designed MPC and PI controller is shown in Table 13.
C. DESIGN OF MODIFIED PMa-SynRM FOR IMPROVING ITS TORQUE AND EFFICIENCY IN FESS [197] The design of IPM-SynRM, which is satisfactory for FESS owing to its advantages, is the other progress in the field of FESS. Since placing of PM materials is usually into the flux barriers region of the rotor and PMs' size relies on the available area of the barriers, using the PMs' advantages is not efficient. The PMa-SynRM rotor consists of two segments, including the Interior PM (IPM) and the SynRM. In IPM-  [197].
SynRM the above segments are installed axially on the shaft. Fig. 34 shows the Three-Dimensional (3D) views of the built IPM-SynRM rotor. According to [197], increasing the T em by 15% and decreasing the T ripple by 70% is the merits of the designed IPM-SynRM.
Contrary to the SynRMs and PMa-SynRMs, the T ripple rate of IPM-SynRM will be lower with increasing load level. Moreover, PM flux linkage is significantly more in IPM-SynRM, which is owing to the design of PMs (i.e., sizes and how arrangement) not relying on the SynRM design. So, increasing PM flux linkage per unit PM mass by up to 340% is yielded. The built prototype of the designed IPM-SynRM along with the test bench is shown in Fig. 35. Comparing the power factor, T avg , T ripple , PM volume, electromagnetic losses, and efficiency between IPM-SynRM and PMa-SynRM are displayed in Table 14. Accordingly, IPM-SynRM has more efficiency, higher T avg , higher P out , higher power factor, and lower losses than PMa-SynRM, showing the superiority of IPM-SynRM usage.
Accordingly, in the above section, the laboratory progress in the field of FESS, such as robust control algorithms of PMa-SynRM and IPM-SynRM design were reviewed, which give advantages, especially in the FESS usages, as displayed in Table 15.

VI. CONCLUSION
The FESS has lots of benefits, including long lifetime, few contaminations, fast dynamic, and high charging/discharging capability. The FESS can be used in EVs, railways, and renewable energy systems. The total charges and discharge cycles of FESS are very high, and it can transfer high energy in a short time (seconds to minutes). Accordingly, FESS has superiority in sustainability, especially for high-cycle and high-power storage systems. Since the efficiency, power rating, cost, angular velocity, power factor, weight, and volume of FESS have a high dependency on the electrical machine, power electronics converter, and control system, which have the most significant role in transferring energy, this study comprehensively reviews them according to FESS usage.
Accordingly, the FESS's configurations, including flywheel (rotor), electrical machines (motor/generator), power electronics converters, control systems, and bearings, are investigated in detail. So, the review of electrical machines' control systems and bidirectional converters displayed that the most concentrations are on the FOC strategy in PMSMs and IMs and the two-level BTB converters. On the other hand, using PMa-SynRM, and especially the suggested IPM-SynRM, can be regarded as a satisfactory suggestion to use in the FESS.
The FESS has two main categories, in terms of rotation velocity, including HSFESS and LSFESS. Composite flywheel (rotor) and magnetic bearings are employed in HSFESS. In LSFESS, a steel flywheel and mechanical or mixed (mechanical and magnetic) bearings are employed. Since HSFESSs have more usage, electrical machines with the capability of high rotation velocity should be regarded, such as PMSM, SynRM, and PMa-SynRM.
Conventional electrical machines in FESS are IM, PMSM, SRM, and SynRM. SRM has a high T ripple , and IM is not satisfactory for HSFESS. PMSM has the advantages of high power density, high efficiency, and high power factor, though the iron loss causes constraints on it. SynRM has advantages such as low-cost manufacturing, high efficiency, and high mechanical strength, whereas it has challenges such as low power factor, low torque density, and large T ripple . Since the electrical machines' T avg , T ripple , and power factor have significant effects on FESS, the PMa-SynRM machine is suggested and investigated. PMa-SynRM gives better power factor and torque density, owing to T PM besides T reluctance , which has the advantages of PMSM and SynRM, simultaneously. In addition, PMa-SynRM has the advantages of rotor configuration simplicity, control simplicity, and less power loss owing to the lack of rotor current.
Finally, the outcomes of the laboratory in FESS, including the design of electrical machines and their control algorithms, are investigated in the above review study. Since the measurement errors will cause higher angular velocity and system instability, using HPF and FIR filters will remove the offset and dead zone errors and fix the angular velocity oscillations in FESS. Additionally, using the suggested CCS-MPC gives a faster and more accurate dynamic and less T ripple . On the other hand, the IPM-SynRM is a new configuration of PMa-SynRM, which has advantages such as increasing T avg by 15% while 29% less used PM, reducing T ripple by 70%, improving PM flux linkage per unit PM mass up to 340%, and increasing the load level, power factor, and efficiency.