Hybrid Energy Storage Modeling and Control for Power System Operation Studies: A Survey

Abstract

As the share of variable renewable energy sources in power systems grows, system operators have encountered several challenges, such as renewable generation curtailment, load interruption, voltage regulation problems, and frequency stability threats. This is particularly important for power systems transitioning to net zero. Energy storage systems are considered an effective solution to overcome these challenges. However, with the increasing penetration of renewable energy sources, different requirements have emerged, and a single energy storage solution may not effectively meet all of them. Hybrid energy storage systems have recently been proposed to remedy this problem. Different individual energy storage systems possess complementary characteristics that can enhance the reliability, security, and stability of power systems. However, hybrid energy storage systems often require more intricate modeling approaches and control strategies. Many researchers are currently working on hybrid energy storage systems to address these issues. This paper thoroughly reviews the modeling and control schemes of hybrid energy storage systems for different power system operation studies. It also examines the factors influencing the selection of hybrid energy storage systems for various power system applications. Finally, this paper provides recommendations for future research in this area.

1. Introduction

1.1. Background and Motivation

The global transition from fossil fuel-based energy resources to renewable energy sources (RESs) is crucial in improving energy security and reducing carbon emissions [1,2]. However, the volatile and intermittent nature of renewable energy sources may lead to voltage and frequency stability problems, poor power quality, and reduced system reliability [3]. To address these challenges, energy storage systems (ESSs) are considered one of the most practical and efficient approaches. They offer effective capabilities to offset RES intermittency and optimize energy management to achieve the balance between energy demand and supply [4,5]. However, a single type of ESS might not be sufficient to meet the diversified energy service demands of current renewable energy-integrated power systems due to limited capacity, energy density, flexibility, and life cycles [4]. This is where a hybrid energy storage system (HESS) comes into play, merging the desirable attributes of multiple energy storage technologies to provide higher capacity, performance, and flexibility for the system operators. In general, HESSs comprise high energy storage (HES) and high power storage (HPS) systems where the HES meets the long-term energy demand and the HPS absorbs the transient and peak powers [4]. To maintain power system stability, a combination of ESS technologies with diverse endurances, capacities, and response times are required that can provide crucial ancillary services, such as Frequency Control Ancillary Services (FCAS) and Network Support and Control Ancillary Services (NSCAS), for the system [6].

Some comprehensive literature surveys have been conducted on HESS features, applications, benefits, and future trends. The authors in [7] presented the classification of HESS applications based on the power quality support and power system ancillary service categories, including load profile smoothing, frequency regulation, reactive power control, transient stability enhancement, and providing an uninterruptible power supply. The work in [7] also presented the energy management services and non-technical benefits of HESS applications, as well as their status and challenges. An extensive analysis of recent ESS integration strategies shows that a HESS performs better than a simple battery–ESS (BESS) in enhancing the reliability and stability of a microgrid [8]. An extensive analysis of the HESS-based microgrid control techniques for grid-connected and standalone modes is carried out in [9]. The significance of using different storage technologies with varying endurance and response times to uphold power system stability by delivering FCAS and NSCAS has been examined in [6]. The authors in [10] have reviewed different mechanical, chemical, electrical, and electromechanical ESS technologies and presented HESS combinations to achieve improved performance, extended life cycles, and minimized cost.

A wide range of HESS combinations, such as supercapacitor–BESS (SC–BESS), superconducting magnetic energy storage–BESS (SMES–BESS), hydrogen fuel cell–BESS (HFC–BESS), SC–HFC, flywheel energy system–BESS (FES–BESS), compressed air energy storage–BESS (CAES–BESS), and HFC-FES, can be formed for different power system applications [10,11]. The combination of BESS and SC has been investigated in [12], considering the rate of power constraint in managing the HESS’s function. Optimization methods addressing the HESS’s characteristics and constraints regarding charging/discharging, temperature, durability, cost, and lifespan are assessed in [13]. Extensive research has been conducted on SC- and Fuel Cell (FC)-enabled HESSs for electric vehicles (EVs) in [12,14], highlighting the improvements in power and energy density, operating temperature, and driving range. Stationary batteries and hydrogen-based HESSs are considered suitable options to meet the growing demand for large-scale and long-duration energy storage [6,15,16]. Research works [17,18,19,20] highlight the applications of different battery, SC, SMES, ultracapacitor (UC), and flywheel combinations in hybrid energy systems. There is extensive research on HESSs and their applications. Some review works have been published in this domain [6,7,20,21,22,23,24,25]. However, there is a noticeable gap in the literature regarding a thorough evaluation and comparative analysis of the modeling and control of HESSs for various power system operation applications. This paper addresses this literature gap and comprehensively surveys HESS modeling, control schemes, and selection criteria.

1.2. Contributions

The main contributions of this review paper are outlined below.

• Providing an in-depth and systematic review of the HESS’s role in enhancing power system stability, security, and reliability.
• Extensive analysis of the multi-dimensional decision criteria for HESS selection.
• Reviewing and evaluating the modeling and control schemes applied to HESSs.
• Assessing the crucial role of HESSs in the context of net zero transitioning, a key consideration in today’s changing energy landscape.

To the best of the authors’ knowledge, these contributions are specific to this paper and have not been presented in other review papers.

1.3. Organization

The organization of this paper is as follows: Section 2 elaborates on the methodology used for this literature review. Section 3 discusses energy storage systems with a focus on HESSs. The efficacy of HESSs in improving power system stability and meeting net zero transition goals is discussed in Section 4. Section 5 investigates the multi-faceted selection criteria for achieving an efficient HESS. The conclusions and recommendations for future research on HESSs are provided in Section 6.

2. Literature Review Methodology

This study aims to comprehensively review HESSs and provide research recommendations for their effective implementation in power system applications. To achieve this goal, the authors have developed a step-by-step search and screening methodology to select the most relevant references. Figure 1 depicts this methodology (n indicates the number of selected references in each screening stage).

Articles published in the last ten years are mostly gathered for the final evaluation and in-depth analysis. This particular time period is selected for the following reasons:

• The emerging and advanced nature of HESS technology.
• Changes in power system operation due to the increasing penetration of renewables in the last ten years.
• To collect and evaluate the most recent information in the HESS domain and power system applications of HESSs.

The distribution of articles over the last ten years after the third screening stage (2015–June 2024) is shown in Figure 2. It can be observed that the number of articles published in this area shows an overall increasing trend, particularly in the last three years. Note that the number of articles published in 2024 only includes the first half of the year.

At first, documents related to HESSs and their power system applications were searched for in large and reputed scientific databases, including Scopus, IEEE Xplore, and Google Scholar. These academic databases cover different journals and conference proceedings. IEEE Xplore focuses on electrical engineering and technology, while Scopus covers various disciplines. Both databases can lead to a more comprehensive literature review across different fields. Specific keywords, such as “Hybrid Energy Storage System” and “Electrical Power System”, are used as the primary selection criteria. A total of 613 papers were found using these keywords. In the first screening process, articles written in languages other than ‘English’ are excluded. After the first screening stage, a total number of 548 articles are retained. In the second screening process, the papers are limited to the subject areas relevant to this review work (i.e., engineering and energy). Papers related to other subject areas (such as mathematics, computer science, physics, astronomy, and chemistry) are excluded from the second screening stage. The total number of papers remaining in this stage is 508. In the third screening stage, only journal articles and review papers are considered, and the remaining number of papers becomes 252. In the final stage, an in-depth analysis was conducted to remove any duplication in the obtained data and evaluate the relevancy of the works. Irrelevant papers, such as those using the abbreviation “HESS” to refer to hydrogen energy storage systems, and papers related to HESSs in ships, railways, and aircraft, have been excluded. At this stage, the total number of remaining papers has reached 100.

3. Energy Storage System (ESS)

This section introduces the most relevant ESSs, highlighting their significant benefits, limitations, and comparative features. Moreover, the factors motivating the use of a HESS instead of a single ESS are also discussed. In addition, various viable combinations of ESSs to form a HESS are introduced. Finally, different HESS control mechanisms, including optimization strategies and artificial intelligence-based control schemes, are described in this section.

3.1. Overview and Comparison of Various Energy Storage Systems

Several energy storage technologies have been developed over the years that can be utilized in power systems [9]. However, in this survey, the ESSs used to form the HESS in the selected research papers are considered.

Table 1. Some large-scale practical ESS projects.

Storage Technology	Name of the Project	Country	Power (MW)	Energy (MWh)
PS [6]	Pioneer-Burdekin pumped storage	Australia	5000	120,000
PS [26]	Fengning pumped storage	China	3600	40,000
FES [27]	Hazle spindle FES by Beacon Power	USA	20	-
CAES [28]	Hubei Yingchang	China	300	1500
BESS [29]	Edwards & Sanborn solar-plus-storage project	USA	-	3287
SC [30]	-	China	5	-

provides a list of the most significant large-scale practical ESSs. A brief overview of ESSs is given below.

3.1.1. Pumped Storage (PS)

PS uses two interconnected reservoirs at two different heights. The water is pumped to be stored in the upper reservoir during the surplus generation period and released during peak hours to the lower reservoir to generate electricity [31]. PS is developed based on the hydro generation principle. It is a mature storage technology with benefits, such as a long life and an enormous number of lifecycles [32]. At the same time, PS outperforms all other types of storage in terms of capacity, mainly due to its large storage reservoir. A total of 125 GW of PS capacity accounts for 96% of the global electricity storage capacity [32]. The stored mechanical energy is directly proportional to the height of the head and the volume of the upper reservoir. Therefore, PS can be established only in geographically feasible locations where high heads and large reservoir volumes are available [31]. Moreover, PS exhibits a lower efficiency and can be harmful to aquatic life [31]. However, due to the rapid integration of RESs, some large and expensive PS systems have been built and commissioned around the globe. Fengning pumped storage, with a power capacity of 3600 MW and an energy capacity of 40,000 MWh, is the largest existing PS in the world [26]. Pioneer-Burdekin pumped storage, with a power capacity of 5 GW and an energy capacity of 120 GWh, is the most extensive PS to be constructed in Queensland, Australia [6].

3.1.2. Flywheel Energy Storage (FES)

In FES, a large rotating cylinder, whose bearings are lifted by magnetic levitation, is coupled with a motor/generator [32]. While charging, the motor rotates the flywheel at a high speed to store the mechanical energy as kinetic energy [32]. While discharging, the flywheel rotates due to the stored kinetic energy, and the machine generates electrical energy [33]. The stored kinetic energy is directly proportional to the mass moment of inertia and the square of the rotational speed [31]. FES is an exceedingly popular storage technology due to its high power density and fast response capability. These qualities make it suitable for power system applications, such as inertial response [6]. Flywheels have long lifetimes and low maintenance requirements. FES can withstand high power fluctuations. The limitations of flywheels are the high initial cost and the high leakage [31]. The 20 MW flywheel being used for frequency regulation in Hazle Township, Pennsylvania, USA, is developed by Beacon Power [27].

3.1.3. Compressed Air Energy Storage (CAES)

In this technology, energy is stored by using compressed air under pressure. Motors compress air and store it in chambers. The compressed air is extracted during peak times and passed through the turbine with natural gas to generate electricity [31]. Similar to PS, CAES is a large-scale energy storage system. This high-capacity energy storage technology has a widespread presence due to its smaller size compared to PS. Among different possible techniques, supercritical CAES has prominent characteristics of high energy density and higher thermal efficiency [32]. CAES exhibits low leakage and long lifetimes. However, its efficiency is not significantly high [31].

3.1.4. Hydrogen Fuel Cell (HFC)

HFCs are the most significant form of hydrogen storage system (HSS). They convert chemical energy into electrical energy, producing water and hydrogen as byproducts. A fuel cell can store energy by using off-peak electricity to produce hydrogen via electrolysis. During peak demand, it generates electricity by combining stored hydrogen with oxygen from the atmosphere. As shown in Figure 3, hydrogen is continuously supplied to the anode, where it is split into H+ ions and electrons. The electrolyte, positioned between the anode and cathode, allows only H+ ions to pass through, blocking electrons. The electrons are instead directed through an external load, creating an electric current, before reaching the cathode. At the cathode, the electrons recombine with H+ ions and oxygen to form water [11,32]. Due to the high energy density, fuel cells are used in power systems to improve power quality. Proton exchange membrane fuel cells (PEMFCs) are becoming increasingly popular due to their high efficiency and ability to produce clean energy [32]. Hydrogen is the most common fuel in fuel cells. The electrolyte defines the category of a fuel cell, such as PEMFCs, with an efficiency of 58%, alkaline fuel cells, with an efficiency of 60%, and solid oxide fuel cells (SOFCs), with an efficiency of 60% [11]. PEMFCs are preferable for transportation due to their low temperature, fast start-up, and high current [11]. SOFCs are highly efficient and capable of working in extreme atmospheres. However, they need a temperature control device, leading to cost elevation [34].

3.1.5. Battery Energy Storage System (BESS)

A battery stores energy in chemical form and releases it in the form of electrical energy. Batteries are classified as non-rechargeable or primary batteries and rechargeable or secondary batteries. However, primary batteries are not used in HESS applications. Chargeable batteries are considered to be the oldest ESS used in electric grids [35]. Rechargeable batteries have an anode, cathode, separator, and electrolyte, the last of which transfers electrons between the cathode and anode [11]. A large BESS, belonging to California’s Edwards & Sanborn solar-plus-storage project, has a capacity of 3287 MWh [29]. The surveyed literature revealed that lithium-ion, lead–acid, and vanadium redox are the most common battery technologies. The batteries are further discussed below.

• Lithium-Ion Battery (LIB)

The electrolyte of this battery contains lithium ions [11]. The capacity of these batteries does not deteriorate from frequent charging without being fully discharged [32]. They are particularly suited for EV applications due to possessing a higher energy density than other batteries, such as lead–acid batteries and VRFBs. In addition, LIBs are useful in enhancing grid stability in electric power systems and providing peak shaving and frequency control capabilities [35].

2.

Lead–Acid Battery

This is an advanced storage technology with a set of desirable characteristics. It does not experience capacity reduction even when charged repeatedly after partial discharge with an incredibly low self-discharge [32]. Among all rechargeable batteries, this is the oldest technology [35]. It should be worth noting that 70% of lead–acid batteries are used in automobiles, 21% are used in communication applications, and only 4% are used in other applications [35].

3.

Vanadium Redox Flow Battery (VRFB)

This is the most mature technology among all flow batteries available today. One of the key features of this battery is its ability to provide constant voltage under various operating conditions and an instant recharge facility [32]. Additionally, it has a quick response time, high efficiency, low maintenance, and a tolerance for deep discharge [11].

The material compositions of the three most relevant batteries for this survey are provided in Figure 4 [36,37]. Figure 4 shows that the vanadium redox flow battery and lead–acid battery require significant water for manufacturing. However, the vanadium redox battery requires more water than the lead–acid battery. Both of these technologies use polypropylene significantly. Despite the significant water use, the lead–acid battery requires lead and lead oxides (together, they compose 60% of the material used for this battery). Lead and lead oxides cause significant environmental pollution. Significant lead pollution has been observed in several countries in the soil, crops, and water due to the anthropogenic activities of lead–acid batteries. The environmental impacts of these technologies are given in

Table 2. Environmental impact of different batteries [38].

Storage Technology	Ranking Points 1
LIB	3
VRFB	4
Lead–acid Battery	1

¹ Storage technologies are ranked from 1 (worst) to 6 (best).

[38]. The table shows that the VRFB has a reduced environmental impact compared to the lead–acid battery and LIB.

Figure 5 displays the global share of different battery technologies for power system applications. The global installed energy storage capacity in 2023 is given in Figure 6 [39]. From the figure, it can be seen that the hydro pump is still the most prevalent storage type in power systems, followed by utility-scale batteries and behind-the-meter batteries. In 2017–18, Australia saw 6440 tons of BESS and EV battery sales (270 million equivalent battery units), as shown in Figure 7. By 2050, lithium-ion battery sales for these applications are projected to exceed 600,000 tons annually [36]. Lead–acid batteries have largely been replaced by lithium-ion chemistries, while nickel–metal hydride batteries in EVs will phase out as Toyota transitions to lithium-ion [40]. The increase in battery demand is also prominent through the rise of demand for battery materials. Figure 8 presents the global demand for the materials of batteries used in EVs from 2016 to 2022 [41]. For instance, the figure illustrates a tenfold increase in global lithium demand between the years of 2016 and 2022.

3.1.6. Supercapacitor (SC)

Supercapacitors are also referred to as ultracapacitors in some of the literature [35]. They store energy as electrochemical energy instead of the electrostatic energy stored in conventional capacitors. Their life cycles are much longer than those of conventional capacitors, with a very high power density and efficiency [32]. However, SCs suffer from high costs, a low energy density, and a high self-discharge [11,42]. A SC comprises two conductors separated by a porous membrane separator and an electrolyte. The stored energy is directly proportional to the capacitance and the square of the electrode’s voltage [11]. SCs are appropriate for frequent switching and high charge/discharge applications [43]. They can operate in a wide temperature range of −40 °C to 70 °C [42].

3.1.7. Superconducting Magnetic Energy Storage (SMES)

SMES stores electrical energy in the form of a magnetic field [32]. Its high power density and fast response time make it suitable for power system applications [32]. SMES consists of a coil made of a superconductor with zero resistance [10]. It stores energy in the magnetic field created by direct current flowing through it, experiencing zero energy loss. The stored electrical energy is directly proportional to the coil’s inductance and the square of the current flowing through it [11]. However, it has a high cost and high volume, as well as specific temperature requirements [10,44].

Table 3. Characteristics of different ESSs [9,21,32].

Storage Technology	Charge Time	Discharge Time	Leakage (%/Day)	Response Time	Lifetime (Years)
PS	h–month	h–days	very small	min	40–60
FES	s–min	s–min	20%/h	ms–s	15–20
CAES	h–month	h–days	0.5–1	1–15 min	20–40
HFC	h–month	s–days	0.003–0.03	ms–min	20–30
LIB	min–day	min–24 h	0.1–5	ms–s	5–16
VRFB	h–month	s–h	very small	s	5–20
Lead-acid Battery	h–month	s–h	0.1–0.3	s	5–15
SC	s–h	s–1.2 h	5–40	ms	10–30
SMES	min–h	ms–30 min	10–15	ms	20–30

compares certain characteristics of these ESSs, while

Table 4. Types and ratings of different ESSs [6,21,28,30,32].

Storage Technology	Stored Energy	Power Range (MW)	Energy Range (MWh)	Power Density (W/L)	Energy Density (Wh/L)
PS	Mechanical	10–5000	180–120,000	0.5–1.5	0.5–1.5
FES	Mechanical	0.25–20	0.0052–5	1000–2000	20–80
CAES	Thermal	5–300	580–1500	0.5–2	3–6
HFC	Chemical	0–58.5	0.312–39	0.2–20	500–3000
LIB	Electro-chemical	0–100	0.004–10	500–2000	200–480
VRFB	Electro-chemical	0.3–3	<60	0.5–2	16–33
Lead-Acid Battery	Electro-chemical	0–40	0.001–40	10–400	50–80
SC	Electro-chemical	0–5	0.0005	500–5000	2.5–15
SMES	Electrical	0.1–10	0.0008–0.015	1000–4000	0.5–15

features their types and ratings. Moreover, Figure 9 illustrates the evolution of various energy storage technologies.

3.2. Why Use Hybrid Energy Storage Systems?

Renewable energy generation typically suffers from fluctuations caused by variations in the weather and climatic conditions [46]. A solution to cope with these fluctuations is to integrate and coordinate ESSs with renewable generation. Lead–acid batteries appear to be an option for this purpose. However, they may suffer from premature failure when used under high power and abruptly changing conditions due to their low power density [21,46]. In addition, the reduced lifetime of the battery significantly increases the replacement cost [24]. Therefore, an ESS with a high power density is required to protect the high-energy battery from transients, allowing the battery to only handle an average level of power. As a result, a HESS may be used to fulfill the load requirements while preventing the high-energy source from deteriorating over time [46].

There are many practical cases where a single ESS may be unable to meet the user’s requirements. A combination of two or more ESSs, forming a HESS, can be utilized in such instances. There are two operation modes for ESSs. One of them is a high-power mode, which is also called the sprinter mode [47]. The other one is a high-energy mode, which is also called the marathon mode [47]. For instance, SC, SMES, and FES can respond quickly, but only for a short time (ms-s/min), and are called high-power ESSs. Conversely, PS, CAES, and BESS have a slow response but can operate for a long time, and are called high-energy ESSs [47]. Therefore, a single ESS may not be able to meet both high power and high energy demands, leading to the need for the hybridization of ESSs [11]. This is highlighted in the context of today’s power systems which typically have both high power and energy demands [23]. Therefore, the integration of two or more ESSs to form a HESS that can simultaneously provide high-power and high-energy features is essential for power system applications. It is worthwhile to mention that while LIBs possess both high power and high energy density, their life cycle is affected by temperature, and they are not recommended for deep-discharging applications [11,32]. This further highlights the need for HESSs in modern power systems.

There are several benefits to using a HESS instead of a single ESS. A HESS is primarily designed to enhance the efficiency and capability of energy sources [13]. A well-designed HESS shows satisfactory performance against short-term and long-term intermittencies, thereby reducing the investment cost and maximizing the reliability of the system [18]. A HESS, by decoupling power and energy demands, provides a reduction in initial investment as compared to using a single ESS [22]. A HESS brings an improvement in the performance and efficiency of ESSs and increases their lifetime [48]. Furthermore, a HESS can serve a broader range of applications than a single ESS [49].

3.3. Overview of Some Common HESSs

This section discusses the most frequently utilized HESSs as reported in the selected research papers surveyed in this work.

3.3.1. SC–BESS

In 2002, R. A. Dougal and colleagues introduced a HESS consisting of a battery and an ultracapacitor. They also provided theoretical evidence that implementing this HESS can enhance performance, reduce battery self-discharge, and extend the overall system’s life cycle [50]. A battery, being a high-energy device, is the primary power source for low-frequency loads. A SC, being a high-power device, provides power to the transients and handles high-frequency demands [7]. This HESS comes with several advantages over individual ESSs. Firstly, this combination manages the depth of discharge (DOD) problem by charging each individual ESS during its low state-of-charge (SOC) state [51]. Secondly, the lifetime of the battery improves as a SC can handle high-frequency demands and eliminate fluctuations for the battery. Therefore, thermal stress, as well as battery costs, is reduced [52]. A battery’s lifetime is dependent on the charge and discharge patterns, whereas a SC can withstand extra charge–discharge cycles [53]. The deep discharge of the battery can be prevented under this HESS scheme. Therefore, the life expectancy of the battery could be significantly improved [54]. Thirdly, as the battery has to supply only the base load, its size and weight requirements are reduced [54].

3.3.2. FES–BESS

In a HESS, the attributes of different energy storage devices, such as power and energy density, are combined to obtain the desired benefits. For example, an electrical energy storage system such as a BESS has low power density but high energy density, whereas a mechanical energy storage system such as FES could meet high power demands [9]. Therefore, FES and BESS can be combined to leverage the benefits of both mechanical and electrical storage systems. However, FES, a mechanical energy storage technology, has high standby losses due to the aerodynamic drag and friction of the bearings in the flywheel rotor [33]. Moreover, the conversion between mechanical and electrical energy via a motor generator incurs losses such as hysteresis, eddy currents, and copper losses. On the other hand, BESS–BESS, an electrical–electrical HESS, is not suitable for frequent, fast, and deep-discharge operations that are typically required in frequency regulation services [33]. In contrast, an electrical–electrical hybrid storage system can merge the attributes and benefits of different electrical storage systems, although it experiences challenges in terms of inertial responses compared to a mechanical–electrical HESS (i.e., FC–BESS) [9]. Table 5 compares the advantages, disadvantages, and applications of different types of HESSs.

The FES–BESS is useful for supporting frequency and improving power quality; however, the high leakage of FES is the bottleneck for this scheme [9]. The high power density of FES and high energy density of a BESS complement each other in this HESS. The battery smooths the low-frequency component of the power, while the flywheel compensates for the high-power component [55]. This scheme is popular due to its low environmental impact and cost. Moreover, it can provide grid frequency regulation services [56] and improve the voltage stability margin of DC microgrids [57,58,59]. This scheme is also used for AGC control in power systems with thermal generating units [58].

3.3.3. SMES–BESS

SMES is combined with a BESS to leverage the high power density of SMES and high energy density of a BESS [21,32]. The integration of SMES with a BESS saves the battery from frequent charging and discharging by supplying high-frequency loads using SMES, thereby improving the battery’s lifetime [60,61]. The size of both the battery and SMES decrease due to the hybridization, although the cost reduction in this HESS mainly happens due to the improved battery lifetime [18]. This combination helps to enhance the power system stability margin as it can be used as a virtual inertia source [62,63]. This HESS can stabilize the output power of wind-based AC grids [64]. The DC-link voltage of the grid-connected PV systems can be stabilized using this scheme [62]. Moreover, incorporating this HESS can improve DC-bus voltage stability in a DC microgrid [38,60]. Additionally, it can be utilized to enhance the frequency stability of a wind-based microgrid [59], including load frequency control [65,66].

3.3.4. BESS–BESS

This HESS scheme is formed by combining high-energy and high-power batteries. This combination has also been referred to as a hybrid battery energy storage system (HBESS) in the literature [59]. This HESS provides more flexibility when providing grid services than using a single BESS. Generally, one of the batteries is a high-energy battery, such as lithium iron phosphate, while the second battery is a high-power battery, such as lithium nickel manganese cobalt oxide. This type of combination increases the lifetime of the HESS [67]. Furthermore, the low internal resistance of the high-power battery reduces the size of the storage system. Therefore, this HESS can provide a storage solution with a low cost, low size, and high lifetime [59]. However, it has limitations regarding the dependency of the battery’s lifetime on charge–discharge patterns and depth of discharge [9].

3.3.5. HSS–BESS

An HSS is technically suitable for long-term and large-scale energy storage. A BESS can be used for high power requirements to complement an HSS. However, the payback period for an HSS is typically long due to its low efficiency. The efficacy of hydrogen storage is the main cause of the low efficiency [15]. Alkaline hydrogen-based HSSs and LIB-based BESSs are used in remote areas to ensure an uninterrupted power supply [68]. In an integrated energy hub (IEH), an HSS–BESS system can provide a significant cost reduction and improvement in environmental effects compared to a BESS-only system [69]. It should be worth noting that biomass is used to produce hydrogen under this scheme (HSS–BESS). The loss of power supply probability (LPSP) is equal to or less than 1% in the embedded microgrid [70]. An HSS–BESS–SC system can be incorporated into a grid-integrated PV system for stabilizing the DC-link voltage [71]. The advantages, limitations, and applications of HESSs are summarized in Table 5.

Table 5. Advantages, limitations, and applications of different HESSs.

HESS Type	Advantages	Limitations	Applications
SC–BESS	High-frequency demands are met by the SC, thus extending the battery’s lifecycle and reducing battery size, weight, and thermal stress.	Limited energy storage capacity in the SC	Regulating power output [60] Improving voltage stability [61]
FES–BESS	Low environmental impact Cost-effective Providing grid frequency regulation services Improving voltage stability	High leakage in FES decreases overall efficiency [9].	Grid frequency regulation services [56] Improving voltage stability margin of DC microgrids [55,57] AGC control [58]
SMES–BESS	Increasing battery lifetime by reducing charging/discharging cycles [60,61] Can be used as a virtual inertia source [63]	High cost of SMES. Lack of HESS control [10]	Microgrid frequency stabilization [65] Load frequency control [66] Power system frequency stabilization [63]
BESS–BESS (HBESS [59])	Increased flexibility for grid services High lifetime Low size and cost [59]	Lifetime is limited by charge–discharge cycles and depth of discharge [9].	Improvement of frequency stability [67]
HSS–BESS	Suitable for long-term and large-scale energy storage Cost-effective in energy hubs [69] Environmental benefits [69] Low loss of power supply probability (LPSP ≤ 1%) [70]	Low efficiency [15]	Ensuring uninterrupted power supply in remote areas [68] DC-link voltage stabilization in PV systems [71]

Table 5. Advantages, limitations, and applications of different HESSs.

HESS Type	Advantages	Limitations	Applications
SC–BESS	High-frequency demands are met by the SC, thus extending the battery’s lifecycle and reducing battery size, weight, and thermal stress.	Limited energy storage capacity in the SC	Regulating power output [60] Improving voltage stability [61]
FES–BESS	Low environmental impact Cost-effective Providing grid frequency regulation services Improving voltage stability	High leakage in FES decreases overall efficiency [9].	Grid frequency regulation services [56] Improving voltage stability margin of DC microgrids [55,57] AGC control [58]
SMES–BESS	Increasing battery lifetime by reducing charging/discharging cycles [60,61] Can be used as a virtual inertia source [63]	High cost of SMES. Lack of HESS control [10]	Microgrid frequency stabilization [65] Load frequency control [66] Power system frequency stabilization [63]
BESS–BESS (HBESS [59])	Increased flexibility for grid services High lifetime Low size and cost [59]	Lifetime is limited by charge–discharge cycles and depth of discharge [9].	Improvement of frequency stability [67]
HSS–BESS	Suitable for long-term and large-scale energy storage Cost-effective in energy hubs [69] Environmental benefits [69] Low loss of power supply probability (LPSP ≤ 1%) [70]	Low efficiency [15]	Ensuring uninterrupted power supply in remote areas [68] DC-link voltage stabilization in PV systems [71]

3.4. HESS Modeling and Control

This section discusses the modeling, hybridization, and control schemes used for HESSs. The optimization and AI-based control of HESSs are discussed in Section 3.5.

3.4.1. HESS Modeling Overview

The modeling of the storage system has significant impacts on the simulation process of the power system. Three different models of energy storage systems, including HESSs, can be found in the literature [72]. These are:

(1)

Physics-based model or white box model.

(2)

Circuit-based model or grey box model.

(3)

Data-driven model or black box model.

The physics-based model is developed based on the physical theories of energy conversion, which involve a large volume of partial differential equations [72]. For example, the physical model of a lithium-ion battery (LIB), one of the most widely used ESSs, depends on specific physical parameters that affect its performance. Therefore, the electrochemical model of an LIB functions as a white box model. In this regard, a pseudo-two-dimensional method was developed by Doyle, Fuller, and Newman, which describes the behavior of lithium ions during the battery’s charging and discharging states. However, its drawback is the involvement of partial differential equations and nonlinear algebra, which require considerable computation burden to solve [72]. Although such models accurately represent the behavior of energy storage at a microscopic level, the computational complexity acts as a barrier to using these models and encourages researchers to consider simplified models [72].

The circuit-based model uses a series of circuit elements to match the physical phenomenon of the physical model. This is a simplified representation of the physics-based model [73]. The circuit-based model is commonly used for the ESS with electrochemical characteristics. The generic representation of this model comprises a voltage source with internal resistance and an RC network. The corresponding model of a battery, SC, and FC can be expressed as given in (1) [73].

$$V_{EL}\left(t\right)=V_O+V_a\left(t\right)+I_{EL}{R}_C$$

(1)

In (1), $V_{EL}$ is the instantaneous voltage per cell, $I_{EL}$ is the instantaneous current per cell, $V_a$ presents the voltage across the RC circuit, and $R_C$ is the internal resistance of the voltage source. The power rating of the battery, SC, or FC can be increased by stacking the number of cells. In the circuit-based model, the stacking voltage, current, and power (i.e., (2)–(4)) are usually used to present the model for high-power applications [74].

$$V_{EL}\left(stack\right)=N_s{V}_{EL}$$

(2)

$$I_{EL}\left(stack\right)=N_p{I}_{EL}$$

(3)

$$P_{EL}\left(stack\right)=V_{EL}\left(stack\right)I_{EL}\left(stack\right)$$

(4)

In (2)–(4), $N_s$ is the number of cells connected in series in the stack, $N_p$ is the number of cells connected in parallel in the stack, and $P_{EL}$ is the active power of the stack.

For power system studies, the average model of a BESS is usually used [75]. The BESS can be modeled using differential algebraic equations (DAEs) (5)–(14) as below [75]:

$${\displaystyle \frac{{d\overline{V}}_{in}}{dt}}=-{\displaystyle \frac{1}{C_2{R}_{bat}}}{\overline{V}}_{in}-{\displaystyle \frac{1}{C_2}}{\overline{I}}_L+{\displaystyle \frac{1}{C_2{R}_{bat}}}{V}_{bat}$$

(5)

$${\displaystyle \frac{{d\overline{V}}_{dc}}{dt}}={\displaystyle \frac{1-d}{C_1}}{\overline{I}}_L-{\displaystyle \frac{1}{C_1}}{I}_{dc}$$

(6)

$${\displaystyle \frac{{d\overline{I}}_L}{dt}}={\displaystyle \frac{1}{L}}{\overline{V}}_{in}-{\displaystyle \frac{1-d}{L}}{\overline{V}}_{dc}$$

(7)

$$I_{dc}=\left({\displaystyle \frac{{3V}_{sd}}{{2M}_c{\overline{V}}_{dc}}}\right)I_d$$

(8)

$$V_{td}={\displaystyle \frac{1}{2}}{m}_d{\overline{V}}_{dc}$$

(9)

$$V_{tq}={\displaystyle \frac{1}{2}}{m}_q{\overline{V}}_{dc}$$

(10)

$$P_s={\displaystyle \frac{3}{2}}{I}_d{V}_{sd}{N}_c$$

(11)

$$Q_s=-{\displaystyle \frac{3}{2}}{I}_q{V}_{sd}{N}_c$$

(12)

$$L_f{\displaystyle \frac{{dI}_d}{dt}}=V_{td}-V_{sd}-R_f{I}_d+{\omega }_o{L}_f{I}_q$$

(13)

$$L_f{\displaystyle \frac{{dI}_q}{dt}}=V_{tq}-V_{sq}-R_f{I}_q-{\omega }_o{L}_f{I}_d$$

(14)

In (5)–(14), $V_{in}$ is the voltage at the battery bank terminals (the bar superscript over the variables represents average values), $V_{bat}$ is the equivalent battery bank voltage, $R_{bat}$ is the equivalent resistance of the battery bank, $C_1{,C}_2$ are the capacitances of the DC–DC converter capacitors, $L$ is the DC–DC converter inductance, $I_L$ is the inductor current, $V_{dc}$ is the voltage at the terminals of the dc-link, $I_{dc}$ is the dc-link current, $d$ is the duty cycle of the DC–DC converter boost switch, $M_c$ is the number of DC converters in parallel per voltage source converter (VSC), $m_d$ and $m_q$ are the d- and q-axis modulation ratios, $V_{sd}$ and $V_{sq}$ are the d- and q-axis grid voltages, $I_d$ and $I_q$ are the d- and q-axis currents of the VSC, $V_{td}$ and $V_{tq}$ are the d- and q-axis voltages at the terminal of the BESS, $L_f$ is the inductance of the filter and the transformer, $R_f$ is the resistance of the filter and the transformer, $N_c$ is the number of VSCs in parallel per BESS, $P_s$ and $Q_s$ represent the active and reactive powers provided by the BESS for the system, and ${\omega }_o$ is the angular frequency. The SOC logic of the battery pack is also used in the comparator of the current controller to compare the measured and reference currents of the d and q axes. The error signal is passed through PI controllers to generate control signals. Feed-forward compensation is added to the control signals for calculating the desired VSC terminal voltage, which is used to calculate the modulation ratios $m_d$ and $m_q$ of the VSC. These modulation ratios control the actual output voltage of the VSC, such that the BESS’s injected and reference powers are equal [75].

The supercapacitor can be modeled as follows [76]:

$$C_{sc}=C_o+K_v{U}_{sc}\left(t\right)$$

(15)

$${ C}_k^s={\displaystyle \frac{1}{2}}{C}_{sc}\hspace{1em}k\in \left\{1\dots n\right\}$$

(16)

In (15) and (16), $C_o$ represents capacitance at 0 V, $K_v$ is the constant, $U_{SC}$ is the SC voltage, and $C_{sc}$ represents the capacitance of the SC. In addition, $C_k^s$ is the infinite sum capacitance of the parallel branches and $n$ indicates the number of parallel RC groups in one branch of the SC model. The value of $n$ up to 5 gives satisfactory results [76].

The supercapacitor bank voltage and current model can be expressed as given in (17) and (18) [76]:

$$U_{SC}^s=n_s{U}_{sc}$$

(17)

$$i_{sc}^m=n_p{i}_{sc}$$

(18)

In (17) and (18), $n_s$ represents the number of series cells to form a string and $n_p$ represents the number of parallel strings in the SC bank. In addition, $U_{SC}^s$ is the SC string voltage and $i_{sc}^m$ represents the SC module current, whereas $i_{sc}$ represents the individual cell current. Equations (15)–(18) are used by the SC model block to provide $U_{sc}$ and $U_{SC}^s$ to the charge control block. The PQ control processes the error signal from the active and reactive power inputs and provides d-axis and q-axis currents to the charge control block. The currents generated by the PQ control represent the system demand without accounting for the state-of-voltage of the SC. The charge control block ensures that the SC’s charging and discharging remain within safe limits, as the SC is highly sensitive to overvoltage and an overcharge exceeding 5% of the rated voltage can damage a cell [76]. Additionally, the charge control block determines the final setpoint d-axis and q-axis currents for the inverter while considering the voltage state of the SC. The inverter uses these currents, along with an input from the PLL regarding the voltage reference frame, to provide the desired active and reactive powers.

The basic SMES model can be represented by an RL circuit. The charging current of the SMES can be expressed as [77]:

$$I_c\left(t\right)={\displaystyle \frac{U}{R_e}}\left(1-exp{\displaystyle \frac{R_{e}t}{L}}\right)$$

(19)

In (19), $U$ is the voltage of the SMES coil, $R_e$ is the SMES coil resistance, $L$ is the SMES coil reactance, and $I_c$ is the charging current at time $t.$ The discharging current can be expressed as below:

$$I_{Dis}(t)=I_{o}exp\left({\displaystyle \frac{R_{e}t}{L}}\right)$$

(20)

In (20), $I_o$ is the initial coil current and $I_{Dis}$ is the discharging current at time $t$. The formulations (19) and (20) are used in the control loop to estimate the inputs for the PQ control of the SMES converter [77].

The black box model can be developed based on empirical or data-driven modeling. This modeling does not rely on physical phenomena [78]. However, it uses large amounts of experimental data to obtain the non-linear model of the storage [78]. The data-driven model is more adaptive and highly generic [78]. It is suitable for modeling different types of storage systems and, therefore, suitable for the modeling of HESSs. As per the investigation in [79], artificial neural networks (ANNs) can provide satisfactorily accurate results in modeling the battery energy storage systems; however, extreme gradient boosting and fuzzy logic algorithms can be considered to provide improved results. Black box models, also known as mathematical models, can be either analytical or stochastic. Nevertheless, the accuracy of these models is influenced by the efficiency of the training methods and the size of the datasets. Additionally, in these methods, underfitting and overfitting can lead to numerical errors [80].

It should be worth noting that the circuit-based model can be obtained from both the physics-based and data-driven models. Figure 10 demonstrates the comparison of various ESS/HESS modeling approaches.

3.4.2. Hybridization Approaches and Architectures

In a hybrid energy storage system (HESS) design, the hybridization architecture significantly influences the control, energy management strategies, and various features such as modularity, flexibility, efficiency, and cost. More flexible architectures allow for greater control and energy management options, leading to better performance but often at the cost of increased complexity and expenses. Most hybridization architectures are tailored to specific control and energy management requirements, ranging from simple, low-cost designs to more complex, high-cost options. These architectures are generally classified into three main types: active (i.e., active parallel), passive (i.e., passive parallel), and semi-active (i.e., cascade) [81]. Power electronic devices (PEDs) are used within these architectures, each offering unique advantages and disadvantages [81]. We have presented each of these topologies below, followed by a comparison table (Table 6).

Active Topology

An active HESS architecture requires active control to maintain optimal performance [82]. In this design, the power management system (PMS) manages the charging and discharging of energy storage components based on load demands and system conditions. This ensures maximum efficiency while minimizing the degradation of the components [83]. Active architectures, as given in Figure 11, use a DC/DC converter, a DC/AC converter, and advanced control algorithms, making them highly versatile and adaptable to varying load profiles and energy demands [84]. While active systems offer enhanced control, better system management, and higher efficiency, they are also more complex and costly compared to passive architecture. This topology requires regular maintenance and monitoring to ensure long-term performance. Active designs are ideal for applications with fluctuating or unpredictable loads and high energy demands, where performance, efficiency, and reliability are the key priorities [82].

Passive Topology

A passive HESS design connects energy storage components in a way that allows the system to operate automatically without active control. As illustrated in Figure 12, the components are configured to enable load sharing and charge balancing without requiring complex control algorithms and a DC/DC converter. The system adapts automatically to load changes, making it simpler and more cost-effective than active architectures. However, passive systems are less flexible when it comes to handling varying energy storage requirements and have limited control over the charging and discharging processes. Passive designs are ideal for applications with stable and predictable loads and lower energy storage requirements. Additionally, the lower cost makes this topology suitable for situations where budget constraints are a priority [85,86,87,88].

Semi-Active Topology

A semi-active HESS architecture combines elements of both passive and active technologies. In this setup, energy storage components are interconnected to allow some automatic operation while still requiring human control for optimal performance. As shown in Figure 13, the PMS in a semi-active architecture provides partial active control over the charging and discharging of the components, using advanced management algorithms and a DC/DC converter [89,90,91]. Additionally, passive balancing devices may be employed to ensure the efficient operation of the system. Semi-active HESS architectures offer a balance between passive and active systems, making them versatile enough to handle a broader range of load profiles and energy storage requirements. While this is generally less complex and more cost-effective than fully active systems, this topology still provides greater control over system operations than passive designs. It should be worth noting that a semi-active topology does not provide the same level of control, flexibility, or adaptability as an active topology. Like active architectures, semi-active designs may require regular maintenance and monitoring to ensure long-term performance. A semi-active topology is well-suited for applications with moderate energy storage requirements. Furthermore, this covers the balance between automation and control [82,91].

Table 6. Comparison of different HESS architectures.

Strategy	Control Complexity	Fault Tolerance	Space Requirement	Flexibility	Cost	Ref.
Active	High	Yes	High	Full	High	[83,84]
Semi-active	Moderate	Only HPS	Moderate	Partial	Moderate	[89,90,91]
Passive	Low	No	Low	No	Low	[85,86,87,88]

Table 6. Comparison of different HESS architectures.

Strategy	Control Complexity	Fault Tolerance	Space Requirement	Flexibility	Cost	Ref.
Active	High	Yes	High	Full	High	[83,84]
Semi-active	Moderate	Only HPS	Moderate	Partial	Moderate	[89,90,91]
Passive	Low	No	Low	No	Low	[85,86,87,88]

3.4.3. Control Design

P. K. Behera et al. [92] proposed an improved filtration-based (FB) control scheme for low-voltage DC (LVDC) microgrids with a SC–BESS and EVs. This control scheme successfully stabilized the grid voltage. L. Liu et al. [93] proposed an opposite vector modulation-based control for a grid-connected AC microgrid with a LIB–SC to improve the power quality (voltage waveform). However, the results showed a 20% increase in power allocation performance at the cost of a higher total harmonic distortion (THD) and lower efficiency compared to conventional space vector modulation control [93]. A modular multi-level converter-based HESS has been proposed in [62]. SMES and BESS are adopted in that work to form the HESS. The SMES is controlled through proportional-integral (PI) and feed-forward control, whereas a single closed-loop control is used for the BESS. A conventional FB control with a low-pass filter and a FB feed-forward control are shown in Figure 14.

The primary objective of a FB feed-forward control (shown in Figure 14b) is to minimize the charging and discharging stresses on the battery, thereby extending its lifespan. Throughout this process, it is assumed that the state-of-charge (SOC) of all energy storage components remains within acceptable limits. In this algorithm, the average value of the DC grid voltage ($V_o$) is compared with a reference voltage ($V_{ref}$), and the resulting error is fed into a proportional-integral (PI) controller. This PI controller calculates the total current required by the energy storage system ($I_{tot\_ref}$). The calculated $I_{tot\_ref}$ is then divided into low-frequency ($I_{LFC\_ref}$) and high-frequency ($I_{HFC\_ref}$) components.

The low-frequency component is given by [94]:

$$I_{LFC\_ref}=f_{LPF}\left(I_{tot\_ref}\right)$$

(21)

where $f_{LPF} (·)$ represents the low-pass filter function. The component $I_{LFC\_ref}$ is further processed through a rate limiter to regulate the charge/discharge rates of the battery, yielding the battery reference current as shown [94]:

$$I_{B\_ref}=f_{RL}\left(I_{LFC\_ref}\right)$$

(22)

where $f_{RL}(·)$ is the rate limiter function. Afterward, $I_{B\_ref}$ is compared with the actual battery current ($I_B$), and the resulting error ($I_{B\_err}$) is sent to a PI controller. This controller generates the duty cycle ($D_B$), which is used by the pulse width modulation (PWM) generator to produce switching pulses for the battery switches ($S_{w2}$ and $S_{w3}$), as shown in Figure 14b.

The high-frequency component of the total current (i.e., $I_{HFC\_ref}$) is given by [94]:

$$I_{HFC\_ref}=I_{tot\_ref}-I_{B\_ref}$$

(23)

Due to the battery’s slow dynamic response, it may not track $I_{B\_ref}$ instantly. Consequently, the uncompensated battery power ($P_{B\_uncomp}$) can be expressed as [94]:

$$P_{B\_uncomp}=(I_{HFC\_ref}+I_{B\_err})·V_B$$

(24)

where $V_B$ is the battery voltage. This uncompensated power must be compensated by the supercapacitor (SC). Thus, the reference current for the SC ($I_{S\_ref}$) is determined by [94]:

$$I_{S\_ref}={\displaystyle \frac{P_{B\_uncomp}}{V_S}}={\displaystyle \frac{(I_{HFC\_ref}+I_{B\_err})·V_B}{V_S}}$$

(25)

where $V_S$ is the SC voltage. The $I_{S\_ref}$ is then compared with the actual SC current ($I_S$), and the error is processed by another PI controller. This controller generates the duty cycle ($D_B$), which is sent to the PWM generator to produce switching pulses for the SC switches ($S_{w4}$ and $S_{w5}$), as shown in Figure 14b.

In [95], a modified FB scheme is utilized for the distribution of power to a HESS, consisting of a SC and BESS. The structure of the FB control with a high-pass filter is shown in Figure 15. The control design has considered the orchestration of the active power of the load and renewable generation [96]. The HESS has been applied to an isolated wind-dominated microgrid. The results indicate a 78.08% reduced peak overshoot of voltage and a 5.85 s reduction in the settling time for voltage in the case of load increase compared with the conventional control. A fast composite backstepping control (CBC), which is based on a higher-order sliding mode observer (HOSMO), has been used to control the SC–BESS in a microgrid [97]. This robust control has improved voltage stability and handled uncertainties [97]. In [98], full vehicle-to-grid (V2G) support is provided along with a two-stage grid following a converter. It includes voltage, frequency, reactive power, and harmonic support for the grid. A scaled-down version of a 2.5 kW system with a DC link capacitor and BESS has been used in this work for experiments.

In [99], a HESS, composed of a SC and BESS, has been controlled by a reference power modulation method using PI control. This control is applied to a standalone nano-grid system. Demand forecasting error minimization is the goal of the proposed control. In [100], an enhanced energy management algorithm, which is a combination of an average current mode controller-based frequency power-sharing scheme and an interleaved boost converter (IBC), has been used for an UC–BESS in EV applications. The regulation of DC-bus voltage has been achieved via this method. In [101], a multi-timescale equivalent model of dynamic interaction stabilization (DIS) is introduced for EV charging stations, with a SC and BESS as the HESS. Dynamic voltage stabilization is achieved using DIS. In [102], virtual resistance and capacitor-based control has been used for a hybrid energy storage converter (HESC) working in the off-grid mode. A SC has been added to the BESS to protect the battery against short-time high-power conditions, thereby increasing the lifetime of the battery and decreasing the overall operation cost.

An adaptive nonlinear controller with disturbance estimation was tested in a non-recursive manner for a HESS (SC–BESS) working in a DC microgrid [103]. The constant power loads (CPLs) have been modeled and simulated with this control scheme, with a clear indication of large-disturbance stability improvement. Z. Li et al. [104] applied decentralized active disturbance rejection control (ADRC), where virtual resistance and capacitance are integrated into a DC microgrid with SC–BESS. A reduction in the number of current sensors of the local controller was observed. They have also considered a CPL model. Their study provides guidelines for the industrial application of the controller. Improved dynamic performance compared to a PI controller and model predictive control was presented.

In [105], a novel autonomous finite-time backstepping control has been used for a SC–BESS in a DC microgrid. The CPL model, uncertainty modeling, and voltage regulation have been considered. In [106], a four-port PV-based HESS is proposed to improve power system reliability and flexibility. A 250-watt experimental setup has been used to validate the efficacy of the proposed scheme and comparable performance with the state-of-the-art schemes. Moreover, it is the only scheme with four bi-directional ports, four continuous current ports, and common ground, providing three control variables, which results in a higher degree of freedom. In [64], a novel power control scheme with a low-pass filter to control the HESS (SMES–BESS) in the context of doubly fed induction generator (DFIG)-based wind turbines and distribution level loads has been presented. The battery lifetime has been improved, along with reducing power fluctuations and stabilizing DC-link voltage.

The work in [107] has proposed a HESS (SC–VRFB) for power smoothing and voltage stability enhancement of a large-scale hybrid wind/PV farm (HWPF) in a multi-machine power system. A probability-based power-sharing control scheme has been applied to the HESS. Eigenvalues and root-loci plots for small-signal stability analysis with and without the HESS have been obtained. Dynamic and transient time domain simulations, considering sudden changes in PV and wind outputs, for large-signal stability have also been performed.

In [46], a modified virtual capacitance droop (VCD) control for SC and a virtual resistance droop (VRD) control for the battery were proposed. The proposed control has been implemented and tested for a HESS integrated in a grid-connected AC–DC microgrid. Fast voltage recovery and harmonic mitigation have been achieved, as indicated by the MATLAB simulation followed by the experimental validation. The cascaded PI controllers with traditional V-P droop are given in Figure 16. The work in [108] has used an improved linear ADRC (iLADRC) for SC–BESS control. The obtained results have been compared with the results of another iLADRC method and found to be superior in terms of reducing voltage ripple and computational burden. Experimental validation has also been performed.

In [110], a multioutput multilevel converter with model predictive control (MPC) has been used to control a HESS (SC–BESS). The MPC-based control topology is shown in Figure 17. The voltage from the common DC-bus has been used to develop the MPD control. The reference current and load current were generated from the output of the PI controller. This scheme allows a sensor-less balancing of voltage and energy with low THD. Moreover, the HESS can simultaneously connect multiple AC systems with lower filter requirements. Simulation and experimental validation have been provided to prove these advantages. In [111], a HESS (SC–BESS) was used to improve the quality of voltage for a DC nano-grid in isolated mode using filtration-based (FB) MPC and a grid-forming (GFM) community HESS. The work in [112] conducts a study related to a HESS (SC–BESS) in a DC microgrid, considering various faults under different control schemes. Actual battery current compensation and reference battery current compensation are used in that work. Filtration and PI-based control are utilized for the HESS. In [113], delay-compensating stabilizing feedback control has been proposed for a grid-connected PV–HESS system. Power balance and stability have been achieved through the proposed control strategy.

In [115], rule-based power management is applied to a HESS (UC–BESS) in residential microgrids with EVs. The frequency decoupling internal model control (IMC), a type of predictive control, has been used. The control is applied to the DC–DC converter system. In [116], a BESS–SC with a variable speed pump storage (VSPS) unit is proposed as the HESS. A modified modular multilevel cascaded converter based on a triple-star bridge cell is used. Feedback control for the cluster capacitor voltage elements and vector control for the current in the grid side are proposed. For validation, real-time power variation data are used. In addition, a weighted median filter is used to generate appropriate references for the BESS and SCs. Interconnection damping assessment–passivity-based controller (IDA–PBC) has been used in the outer voltage control of a HESS (SC–BESS) [117]. The inner current control loop has been developed based on the MPC scheme. IDA–PBC differs from other nonlinear control methods as it considers the energy relations of the system rather than focusing on signals. MPC is easily comprehensible and can handle nonlinearities and multiple variables at the same time. MATLAB/Simulink is used for numerical simulations.

In [118], a HESS (SC-BESS) has been applied in EV-dominated DC microgrids. Adaptive FB current allocation effectively adjusts the filter’s bandwidth to ensure that the high-frequency variations of RESs or loads are only absorbed by the internal SC of the EV. Similar HESSs have also been used in wind farms with capacity optimization [119,120].

In [121], a SC and BESS formed a HESS for a wind farm. A first-order filter and PI controller are employed to acquire the control signals for the HESS [121]. New concepts regarding the inertia constant and wind turbine rotational speed, as seen from the grid side, have been introduced. In [122], a HESS, formed by SC–BESS, has been deployed in a grid-connected microgrid. MPC is employed to control the grid-connected inverters. It requires more computational capability than PI-based control. Maximizing the renewable power output and minimizing fluctuations are the main objectives of that work. In [123], a SC and BESS are combined for a wide range of microgrid applications. Ramp-rate control is used in [123]. Smoothing variations and system efficiency are two main objectives of that work.

Figure 18 outlines the key points of this section. This figure shows that most of the HESSs reviewed in this work have been modeled and used in AC microgrids, followed by low-voltage DC microgrids. HESSs are mostly used in the microgrid compared to bulk power systems. In addition, the most widely used HESS is the SC–BESS, and model predictive control is the most commonly used method for controlling HESSs.

3.5. Application of Optimization and AI in Modeling and Control of HESSs

3.5.1. Optimization-Based Modeling and Control

In [124], a two-stage optimization strategy has been adopted to minimize the deviation of voltage in reactive power control for a HESS (SC–BESS). The HESS has been used in conjunction with a wind farm. An online gradient projection-based iterative algorithm has been applied to solve the optimization problem. In [125], a multi-objective optimization approach has been adopted for the energy management of a DC microgrid with SC–BESS. That work uses an adaptive Kalman filter to estimate the current instead of relying on load and generation data. The results indicate that the proposed scheme can achieve the required voltage regulation. It should be worth noting that the proposed method is superior compared to droop control and fuzzy logic-based control.

A significant reduction of 34.3% in the size of the HESS compared to the battery-only option has been achieved. In [126], a HESS (i.e., UC–BESS) has been modeled for a grid-connected ocean wave-based energy system. A genetic algorithm (GA) has been used to optimize the required number of cells of the BESS and UC for an actual 625 kW wave energy system. The results indicate a significantly reduced HESS size compared to meeting the demand with individual ESSs. Moreover, the lifetime of the HESS has been improved as the UC saves the BESS from deep discharge.

In [67], high-power and high-energy BESSs have been combined together to form a HESS. An energy management system (EMS) using machine learning and GA-based optimization is proposed. The proposed method is deemed suitable compared to the rule-based method. In [127], thermal and electrical energy storage have been combined to form a HESS for multi-energy microgrids. Day-ahead energy cost minimization has been achieved using distributed robust optimization. The proposed strategy proves to be fast and cost-effective, achieving energy balance under various uncertainties. In [69], an Information Gap Decision Theory-based normalized weighted-sum approach that considers uncertainties in generation and demand has been applied to perform multi-objective optimization in an IEH. The multi-objective optimization includes three objectives, including operational costs, carbon emissions, and the energy export index (EEI) [69]. By integrating the HSS with the BESS, costs are reduced by 35.29%, carbon emissions by 33.37%, and the EEI by 71.6% compared to using a BESS alone.

The authors in [128] have proposed a HESS (HSS-BESS) to be used in a future microgrid at Khalifa University. Mixed logical dynamical modeling has been used. An MPC with optimization in MATLAB with the YALMIP tool and GUROBI has been performed to achieve savings in HESS maintenance costs and enhancement of the battery’s lifetime. The zero-phase controlled auto-regressive integrated moving average filter algorithm is used for HESS (UC–BESS) sizing in a wind storage system [129]. Nonlinear programming is used for scheduling, considering battery degradation, to flatten wind power fluctuations and reduce costs [129]. The comparative results show a 50% reduction in battery size and a 25% reduction in SC size compared with wavelet packet decomposition.

The work in [63] has proposed Golden Eagle Optimization-based PID control for a HESS (SMES–VRFB) for LFC of a multi-area power system. An improved settling time for the frequency, compared with some other metaheuristic optimization methods, has been observed with this algorithm. In [130], a two-phase decision-making algorithm has been utilized for a HESS made of portable and stationary batteries (BESS–BESS). The objective of the optimization problem in [129] is to maximize the revenue through arbitrage. For a practical grid in California, a 6.4% increase in revenue has been achieved with this algorithm.

SC–BESS is used for wind farms in [131]. Particle swarm optimization (PSO) is used to minimize the lifecycle cost of the BESS based on the DOD and low-pass filter (LPF). This method (PSO) is used to allocate power between the battery and SC. Actual data from the Oak Ridge National Laboratory are used in this work to demonstrate the efficacy of the proposed optimization approach [131]. In [132], HESS (BESS-SC) lifecycle costs are minimized by finding the optimal SOC algorithm, filter time constant, and DOD for the wave energy converter. It is also revealed that the use of HESSs is more cost-effective than using the battery alone.

In [71], a SC–BESS (redox flow battery) has been used in grid-connected PV systems to improve the stability of the DC-bus voltage. For the HESS control, the optimal tilt integral derivative with a filter current tracker has been used to track the current reference rapidly. A Butterworth low-pass filter is used to split the frequency components of the effective power demand [71]. MATLAB simulation and hardware validation under solar irradiation variations, sudden pulse power loading, and random grid conditions have been performed. In [133], a HESS (SC–BESS) is proposed for a wind farm. A stochastic framework utilizing mixed-integer linear programming to optimize day-ahead market profits for the wind–HESS is presented. This approach considers wind power and market price uncertainties, generating scenarios using the Monte Carlo method.

In [58], a FES and BESS are combined for improved AGC of thermal generators. The scenario tree generation method is used to devise a multi-scenario optimization model. Based on this model, the HESS output is optimized by formulating a quadratic programming problem. Actual historical data of a 330 MW thermal generator in north China are used for the simulation in MATLAB [58].

In [70], an HSS–BESS is applied for off-grid operation. PEMFC is used for a hybrid renewable energy system in an embedded microgrid. NSGA II has been used for optimizing three objectives, including LPSP, annualized cost of the system, and potential energy waste, in a multi-objective optimization model. In [134], SC–BESS is used for solar PV-integrated power systems. The proposed control method at the higher level optimizes the operation of the HESS in a PV plant to limit peak power exchange at the grid’s point of common coupling (PCC), thereby reducing overvoltage and easing grid integration. The lower-level controller is associated with the ramp rate control of PV. In [135], SC–BESS–heat storage has been used to achieve power smoothing. Two-stage frequency division with wavelet packet decomposition and discrete Fourier transform is used to decompose the net load power data. A GA is used to optimize energy system planning. A regional integrated energy system with PV, wind, a combined heat and power unit, battery, an electric boiler, and a SC is utilized as the case study [135,136].

In [137], a SC–BESS has been used to improve transient stability. Frequency decomposition and power distribution are achieved through discrete wavelet transform. Droop control is employed as a supplementary loop in [137,138]. The HESS’s capacity is optimized by using a linear weighted method and generalized Benders’ decomposition. In [120], an HSS–SC is used for a microgrid in remote islands. The Kalman filter algorithm is utilized to forecast the HESS’s output. Financial cost and transient stability are two metrics considered in [120].

The work in [139] presents a SC–BESS–CAES system for a microgrid on a remote island. A quadratic moving average filtering method is deployed in this work for power division while a GA is utilized to optimize capacity allocation and minimize the total cost. In [140], a SC–BESS is employed to regulate power system demand response. A sizing method for the HESS is formulated to achieve an expected dynamic characteristic. The paper explores a variant of a cascade fractional order two-degrees-of-freedom controller, integrating the advantages of a fractional order controller (FOC) and tilt integral derivative controllers for frequency regulation services. Quasi-oppositional Harris Hawks Optimization is used to optimize the controller’s coefficients. In [43], a SC–BESS is used to achieve power smoothing of grid-connected PV systems. Self-adaptive VMD is used for power allocation. The HESS is connected to the grid through a bidirectional DC–DC converter to stabilize the bus voltage and decrease voltage swings. Real-life operation data from a PV power station are used to test the developed approach. A power-based and energy-based combination of ESSs is discussed in [141]. Cycle limits of the HESS are restricted to increase the storage lifetime, and the stored energy is also constrained to ensure proper scheduling. Thus, a network-constrained unit commitment problem is formulated. The problem is solved by integrating progressive hedging and dual decomposition algorithms. A six-bus test system and a modified IEEE 118-bus test system are used for the simulation studies.

In [142], a SC–BESS is used for the demand response of an island microgrid. Optimization was achieved with two objectives: comprehensive operating cost and flexibility insufficiency rate. The m-II algorithm was utilized to solve the optimization problem. That work introduces demand-side management and flexibility for HESS capacity optimization. In [66], a BESS–SMES is used for optimizing cost and tackling grid contingencies. The controller’s optimal parameters are determined by PSO.

In [143], a SC–BESS is used to reliably operate an off-grid power system. A modified moth–flame optimization algorithm is utilized to minimize the total cost of electricity (TCE). The system consists of a PV generator, wind turbine, and HESS. In [144], hybridization of the power battery and SC is achieved, where the control strategy is based on subtractive clustering and an adaptive fuzzy neural network. In [145], a HESS (SC–BESS) is used to enhance the stability of a weak grid. The proposed method compensates for disturbances in the DC voltage loop by using power feed-forward techniques along with a voltage feed-forward phase-locked loop (PLL) compensation method. Moreover, an adaptive bandwidth low-pass filter is used for power allocation. The HESS is connected to a PV array through a boost converter with a maximum power point tracking (MPPT) control algorithm. In [146] for a SC–BESS, Pontryagin’s minimum principle optimization algorithm is employed to formulate a rule-based control that considers battery degradation. The battery degradation model is formulated from experiments considering multiple degradation factors.

3.5.2. AI-Based Modeling and Control

P. Wang et al. [61] have proposed a two-layer control strategy, where the upper layer uses a fuzzy-controlled low-pass filter while the lower control layer uses a fast MPC. The HESS is formed using a SMES–BESS for a PV-connected power system to reduce power fluctuations. The proposed control has successfully improved the battery’s lifetime and enhanced the system’s voltage stability. An ADP-based control strategy has been proposed in [64] for a HESS (SMES–BESS) to improve the frequency stability of a wind-dominated microgrid. Simulations performed on an island microgrid have been reported to reveal the efficacy of the control strategy for frequency stability enhancement. In [147], the variational mode decomposition (VMD) method is used to control a HESS (SC–BESS) in a wind farm by separating different frequency components. Multi-agent deep reinforcement learning is also used for optimal power sharing in the secondary control. Semi-physical experimental validation is performed after the simulation. The authors in [148] have utilized Adaptive Dynamic Programming (ADP), based on reinforcement learning, to control a HESS (LIB–SMES). Power system reliability enhancement and power fluctuation reduction are achieved.

Fuzzy logic has been used for power sharing in a PV-dominated microgrid with a SC–BESS [149]. Power smoothing and voltage stability are the two key objectives used in the study. The work in [150] uses a combination of a non-dominating sorting genetic algorithm (NSGA) and fuzzy logic-based control for the control of EV with SC–LIB. The strategy is useful, increasing the battery lifetime by 76.38%.

In [151], a SC–BESS is used for frequency regulation of power systems in the UK. A fuzzy logic controller with dynamic filtering and a variable voltage control strategy are used. A scaled-down HESS experimental platform is used for hardware verification. The utilization rate of the SC is improved, and the cost of the converter is reduced by controlling the DC voltage [151]. In [152], a SC–BESS is used to improve voltage stability. A multi-mode control strategy with two modules, including mode selection and rule-based power distribution, has been implemented. Furthermore, the gray wolf optimization algorithm is utilized to determine the battery output threshold and SC upper charge limit. In [153], the HSS–BESS is used for island microgrids using fuzzy logic control. DC-bus voltage stability is ensured through the employment of the outer voltage loop. An island DC microgrid is used as the test case.

In [136], a SC–BESS is used to stabilize the voltage and frequency. A robust cascaded control with MPC is the primary control scheme, supplemented by an intelligent neural network (INN)-based secondary control. Both hardware and software experiments are performed on a microgrid based on a real-world electrical distribution network. In [154], a SC and BESS are combined to be used in islanded microgrids. The primary controller uses PSO and fuzzy logic whereas the secondary controller uses an INN, with the objective being precise and efficient voltage and frequency control. A type 2 fuzzy strategy has been used in [155] to enhance asymptotic stability and fast-track DC-bus voltage regulation. This scheme uses the SC–BESS method as the HESS.

Table 7. Summary of optimization techniques used for HESSs.

Reference	HESS	Solving Technique	Objectives/Advantages
[124]	SC–BESS	An online gradient projection-based iterative algorithm	Minimize voltage deviation in reactive power control
[125]	SC–BESS	Adaptive Kalman filter	Voltage regulation
[150]	SC–BESS	Non-dominating sorting genetic algorithm III (NSGA-III)	Battery life improvement, size reduction
[126]	UC–BESS	GA	Battery life improvement, size reduction
[127]	Thermal and electrical	Combination of column-and-constraint generation and analytical target cascading algorithm	Minimization of day-ahead energy cost
[69]	BESS–HSS	Information Gap Decision Theory-based normalized weighted-sum approach	Reduction of financial cost, carbon emissions, EEI
[128]	BESS–HSS	GA	HESS maintenance, cost-saving, and enhanced battery lifetime
[129]	UC–BESS	The zero-phase controlled Auto-Regressive Integrated Moving Average filter algorithm	Reduction in size
[63]	SMES–VRFB	Golden Eagle optimization	Improved settling time for frequency, reduced frequency overshoot and undershoot (objective function is Integral Time Absolute Error)
[130]	BESS–BESS	PSO	Maximize revenue through arbitrage
[131]	SC–BESS	PSO	Minimize lifecycle cost of BESS
[132]	SC–BESS	PSO	Minimize lifecycle cost of BESS
[133]	SC–BESS	Mixed integer linear programming	Optimizing day-ahead market profit
[58]	FES–BESS	Quadratic programming	Improved AGC of thermal generators
[70]	HSS–BESS	NSGA-II	Annualized cost of system, LPSP, and potential energy waste probability
[152]	SC–BESS	Gray wolf optimization	Battery life extension and energy loss minimization
[135]	SC–BESS–heat	GA	Power smoothing, reducing financial cost
[137]	SC–BESS	Linear weighted method and generalized Benders’ decomposition	HESS capacity
[138]	SC–HSS	Kalman filter	Transient stability, financial cost
[139]	SC–BESS–CAES	GA	Optimal capacity allocation and cost
[140]	SC–BESS	Quasi-oppositional Harris Hawks optimization	Power system demand response regulation
[43]	SC–BESS	Self-adaptive VMD	Reduction in HESS lifecycle cost
[141]	Generic model of power-based ESS (SC) and energy-based ESS (BESS)	Integrating progressive hedging and dual decomposition algorithms	Increase storage lifetime and ensure proper scheduling
[154]	SC–BESS	PSO in primary controller	Precise and efficient voltage and frequency control
[142]	SC–BESS	m-II	Comprehensive operating cost and flexibility insufficiency rate
[66]	BESS–SMES	PSO	Optimizing economic cost
[143]	SC–BESS	Modified moth–flame optimization algorithm	Minimizing total cost of electricity
[144]	Power battery–SC	Subtractive clustering	Improve energy storage performance of hybrid electric vehicles
[146]	SC–BESS	Pontryagin’s minimum principle optimization	Reduced energy usage rate and slower battery degradation

summarizes the HESS optimization techniques. It is clear from this table that the majority of the HESSs considered in the surveyed literature are based on electrical and electrochemical technologies. Only one study considered both thermal and electrical options for the HESS [126]. System security/efficiency enhancement, battery life extension, and cost/size reduction are the dominant objective functions used in these studies.

Table 8. Summary of AI algorithms used in HESSs.

Reference	HESS	AI Algorithm	Objectives/Advantages
[126]	BESS–SMES	ADP based on reinforcement learning	Power system reliability improvement and reduction in power fluctuations
[64]	BESS–SMES	ADP based on reinforcement learning	Reduction in power fluctuations
[149]	BESS–SC	Fuzzy logic rule-based power sharing strategy	Power smoothing and voltage stability
[150]	SC–BESS	Sugeno-type fuzzy logic controller	Battery life improvement, size reduction
[61]	BESS–SMES	Two-layer control, upper layer is fuzzy logic-controlled	Battery life improvement, enhanced system voltage stability
[147]	SC–BESS	Multi-agent deep reinforcement learning	Optimal power allocation
[151]	SC–BESS	Fuzzy logic combined with a dynamic filtering method to devise a power management strategy	Enhanced SC utilization and low converter cost
[155]	SC–BESS	Type 2 fuzzy strategy	Enhanced asymptotic stability, fast tracking DC-bus voltage regulation, and signal noise reduction
[153]	HSS–BESS	Fuzzy logic controller for power allocation	DC-bus voltage stability
[136]	SC–BESS	INN (for secondary control)	Voltage and frequency stability
[154]	SC–BESS	Fuzzy logic in primary controller and INN in secondary controller	Precise and efficient voltage and frequency control
[144]	Power battery–SC	Adaptive fuzzy neural network	Improve energy storage performance of hybrid electric vehicles

summarizes the works related to AI algorithms in HESS. It is seen from this table that the AI methods have been used primarily to achieve power smoothing, allocate power, and stabilize the DC voltage.

4. Application of HESSs in Power System Operation

This section briefly explains the impacts of HESSs on power system stability and net zero transition.

4.1. Role of HESSs in Addressing the Challenges in Power System Stability

The contributions of HESSs in improving power system stability are discussed in this subsection.

The work in [101] has discussed low-frequency oscillation issues due to the interactions between the voltage and inertia loops of VIDC of a DC microgrid. Control loops of different timescales have been modeled to form a multi-timescale impedance model. The RLC circuit has been used to explain low-frequency oscillations in the voltage loop and inertia loop. The positive-damping reshaping loops have been suggested to compensate for negative damping to improve the transient stability margin.

A HESS (SC–BESS) has been proposed in [156] to provide primary frequency responses. The researchers have used a comprehensive control scheme combining virtual inertia and virtual droop controls. Large-disturbance and small-disturbance frequency responses have been simulated in MATLAB to verify the improvement in long-term frequency stability. The authors in [145] have proposed a dual-loop feed-forward control scheme for improving the stability of a grid-connected microgrid. The negative parallel impedance caused by PLL may result in voltage instability and harmonics in weak grids. A poorly tuned DC voltage loop has a similar impact. The proposed control scheme for the HESS (SC–BESS) can reduce the frequency range of the negative impedance and improve the small-signal converter-driven stability of the microgrid.

The work in [157] proposed a SC–BESS along with an additional damping control based on a dynamic power compensation control scheme. It was observed that the conventional dual closed-loop control of the grid-connected PV contributes to the inter-area oscillations in the range of 0.1–0.7 Hz. A small-signal model and the torque method have been used for the assessment of the damping increment.

A PSO–fuzzy algorithm in the primary control and a Lyapunov function-based INN in the secondary control is proposed in [154] for a HESS (SC and BESS) to improve voltage and frequency stability in an islanded microgrid. The authors in [140] have analyzed the combined impact of demand response (DR) and the HESS (SC–BESS) on the improvement of frequency stability in an interconnected power system. Bode plot-based stability evaluation along with hardware experiments were used in the work to test the efficacy of the proposed method.

The work in [137] has proposed an optimal sizing for a SC–BESS-based HESS for estimating the transient frequency regulation capability (TFRC) for arresting frequency excursions so that the operation of under-frequency load shedding can be avoided. Multi-objective optimization has been performed to find the minimum investment cost and minimum power fluctuations while satisfying the TFRC constraint. An extended system frequency response model has been developed to determine the required TFRC using the HESS. Electromechanical transient simulations have shown an improvement in the frequency nadir attained by the proposed scheme compared to the no-storage and single-storage cases.

Y. Wang et. al., [135] have proposed a HESS (SC–BESS) planning and optimization model for regional integrated energy systems to enhance frequency stability. Their proposed optimization model uses a GA with elitist preservation to obtain a minimum-cost solution. The work in [153] has proposed a HSS–BESS-based HESS for improving the DC-bus voltage stability of an isolated microgrid. A fuzzy power allocation strategy has been adopted.

The authors in [102] have used machine learning-based EMS for the control of a battery–battery HESS. This scheme has been used to improve the frequency stability of the pan-European microgrid, including EV charging stations.

M. M. Mohamed et. al., [158] have proposed PSO for designing a virtual synchronous generator (VSG), integrated with droop control, for a microgrid consisting of a HESS (SC–BESS), PV-based generation, diesel-based generation, and load. The small-signal-based design of VSG is based on objectives including minimization of frequency ITAE, frequency nadir, and ROCOF. A MATLAB simulation was performed to validate the proposed scheme.

In [56], an FES–BESS-based HESS is proposed for a wind farm. Virtual droop control for the BESS and virtual inertia control for the flywheel have been used. Grid frequency regulation has been achieved. The SC–BESS has been used for improving short-term voltage stability in [111]. The authors in [115] proposed a rule-based power management system for an UC–BESS for stability improvement in a microgrid. The frequency decoupling cascade IMC has been proposed in this work.

Delay-compensating stabilizing feedback control for bilinear systems with a SC–BESS-based HESS has been applied in [114]. Global asymptotic stability has been achieved using Lyapunov criteria. Steady-state and large-disturbance voltage stability are assessed in this work.

The work in [57] has applied an improved hierarchical control scheme for a FES–BESS-based HESS in a PV- and wind-based DC microgrid. This research has studied the impact of DC microgrid grounding in the presence of the HESS. A protection scheme has been designed, and it has been validated that this topology and control scheme can improve large-signal voltage stability. M. Shaban et. al., [112] have applied FB PI control for a HESS (SC–BESS). This setup has improved large-disturbance voltage stability status.

S. A. Ghorashi et. al., [111] combined a community SC and a community BESS to form a HESS for a nano-grid application. FB–MPC and GFM-based HESSs have been used for improving small-disturbance voltage stability. A MATLAB-based simulation validated the topology. In [110], MPC-based control has been applied to the multi-level converter of a HESS (capacitor and battery). A stable region has been defined for the system’s operation, and sensor-less control has been implemented to attain the system’s voltage stability. An improved linear active disturbance rejection control, incorporating a dynamic event-triggered mechanism, has been reported in [108] to control the HESS (SC-BESS) in EVs. An experimental setup has also illustrated the efficacy of the proposed HESS control scheme. The large-signal voltage stability is the main focus of [44]. MPC-based control has been proposed for the HESS (SMES–BESS). A time domain simulation was performed to validate the proposed control scheme.

The work in [159] has introduced a HESS (UC–BESS) with a PEMFC for a PV-based grid-connected system. The voltage stability has been improved by applying PI control optimized by the Path Finder Algorithm. In [46], small-signal voltage stability and frequency stability have been improved by incorporating a HESS (SC–BESS) in a hybrid AC–DC microgrid. A modified VCD control for SC and VRD control for the battery has been applied in that work.

The work in [107] has applied a HESS (SC–BESS) for a multi-machine power system connected to PV and wind farms. The FB control strategy has been applied to improve the system’s voltage stability. Detailed stability analyses have been conducted using eigenvalues, root loci, and time domain simulations. Similarly, the work in [64] has used a HESS (BESS–SMES) to enhance the voltage stability of DFIG-dominated power systems.

A HOSMO-based CBC to control a HESS (SC–BESS) is proposed in [97]. This control scheme is based on the Lyapunov function and improves large-signal voltage stability in a DC microgrid. MATLAB simulation and experimental validation for the proposed control scheme have been presented in [97]. A modified FB power distribution control scheme for a HESS, composed of a SC and batteries, has improved voltage stability in a wind farm-based isolated DC microgrid [71]. The small-disturbance voltage stability of the DC microgrid with the HESS has been assessed using MATLAB simulation and experimental work.

Zhang et al. [62] have utilized a HESS (SMES–BESS) to enhance the short-term small-signal voltage stability of a PV-based microgrid. The changes in PV generation due to changes in radiation and temperature are compensated for by the HESS. The PI and feed-forward control have been effectively applied to improve the system’s voltage stability. Behera et al. [92] have utilized a new FB control scheme to control a HESS (SC–battery) in a DC microgrid. The short-term small-signal voltage stability has been improved through this arrangement.

The work in [149] demonstrated that the voltage stability of a PV-based microgrid can be improved using a HESS constituted by a SC and BESS. The Model Predictive Current Control (MPCC) strategy has been used for the HESS, where a power allocation scheme determines the required current injections for maintaining the voltage. Both software and real-time-based assessments for the control scheme were presented in [149].

ADP-based reinforcement learning is used in [148] for a HESS (SMES–BESS). The main focus of this work is to improve the power system’s voltage stability. A sudden load change is simulated to demonstrate how the HESS can help maintain voltage stability. Peng et. al. [124] have used a distributed SC and BESS to form a HESS to be used in a wind farm. They have also used two-stage optimization to minimize the node voltage deviation and maintain voltage stability, while reactive power varies, using the HESS. In addition, formulations have been presented in [124] to define the stability boundary for the given control scheme.

4.2. Role of HESSs in Meeting Net Zero Transitioning Requirements

The Paris Agreement was signed by 190 countries with the common goal of reducing global warming, along with establishing resilience to climate change by achieving net zero emissions by 2050 [160]. Eighty-six percent of CO₂ emissions are caused by energy production and consumption. However, the global progress in achieving net zero targets is currently falling short of requirements. Hence, there is an urgent requirement for increased efforts to achieve a net zero transition [161].

It is a well-established fact that renewable generation instead of fossil fuel generation is the key to achieving net zero emissions [23]. For instance, according to the research conducted by the International Renewable Energy Agency, to achieve the net zero targets, the share of renewable generation in global power generation must increase to 60% by 2030 [162]. Among sources of renewable energy worldwide, PV and wind are the most significant options. However, PV generation is only available in the daytime. Therefore, it requires ESSs as its backup, which needs to be efficient and cost-effective [163]. A HESS, being capable of providing a more efficient and economical solution than individual ESSs, can be utilized to achieve the net zero goals. The role of HESSs in providing the required support for achieving net zero emissions is discussed next.

4.2.1. Extending Storage Lifetime

When it comes to energy storage, batteries are the most common form of ESS. However, their lifetime may decrease if they are exposed to frequent charging and discharging. This premature failure can limit their effectiveness. However, a HESS instead of a BESS can be utilized to protect the BESS from frequent charging and discharging, thereby increasing its lifetime [164,165], which is effective for net zero transitioning. The critical aspect of battery lifetime is the requirement of metals to manufacture batteries. The metals used in battery manufacturing (such as the lithium, cobalt, and nickel used in LIBs) come from mining, and extensive mining causes emissions [162,163]. A longer battery lifespan would decrease emissions from the battery production lifecycle.

4.2.2. Reducing Storage Cost

A key barrier to the incorporation of long-duration energy storage is its high cost, despite BESSs (LIBs) offering a lower levelized cost as compared to other long-duration ESSs [162]. The effectiveness of HESSs in meeting net zero ambitions is substantial as it can further decrease the levelized cost compared to the use of LIBs alone for storage. It is demonstrated in [139] that for a PV- and wind-based microgrid, the integration of a SC and CAES with a LIB to form a PV–wind–HESS combination provides a lower levelized cost than the PV–wind–BESS combination. Furthermore, due to a single ESS’s limitations in terms of being either a HES or a HPS, its use for meeting both may become uneconomical. Therefore, a HESS can offer a more cost-effective solution by integrating one high-energy ESS with another high-power ESS [92]. Furthermore, as demonstrated in [128], employing a HESS can result in higher energy arbitrage revenue compared to the exclusive use of a BESS.

4.2.3. Overcoming Intermittencies in RESs

Being climate-dependent resources, PV and wind generation possess inherent fluctuations [20,166]. A HESS can be used to overcome these fluctuations or intermittent energy generation [97]. A HESS bridges the gap between renewable energy generation and demand, particularly during peak load or periods of low generation, thereby reducing the need for fossil fuel-based generation and supporting the achievement of net zero goals [162].

5. Selection Criteria for Hybrid Energy Storage Systems

Installation of a HESS instead of a single ESS does not always guarantee enhanced system performance. An inappropriate design and combination could compromise the overall performance of the system. Therefore, the selection of suitable ESSs for the formation of HESSs is an important design consideration [167]. Several factors influence the choice of combining individual ESSs to form a HESS. The availability of ESSs for a certain location, the nature of the load to be supplied, energy efficiency, financial considerations, utilization limits (such as ramp rate constraints), and short-term and long-term applications are some of the factors that affect the optimum choice of a HESS. Some HESSs are recommended for short-term applications, such as SC–BESS, BESS–HFC, BESS–SMES, HFC–SC, HFC–BESS, and BESS–BESS [167,168]. On the other hand, SC–CAES, HFC–FES, HFC–SC, and BESS–PS are considered appropriate for long-term applications [168]. The key selection criteria are briefly discussed below.

5.1. Power and Energy Density

Power and energy requirements are the two basic criteria for selecting or combining multiple ESSs to form a HESS. A higher density corresponds to a reduced land requirement for the HESS [92]. Ideally, a HESS should be based on a combination of ESSs that provide both high power and high energy. Therefore, ESSs possessing complementary characteristics in terms of power and energy density are usually integrated for HESS formation. For example, in [113], a HESS made of a SC and BESS has been proposed due to the high power density of the SC and the high energy density of the BESS. This combination lowers the stress on the battery and increases its lifetime [92,112,113]. The SC and BESS have been combined to take advantage of both high power and high energy densities [65,92,93,110].

5.2. Lifetime Enhancement

Extending the lifetime of a HESS is another selection criterion. The lifespan of a BESS depends on the charge and discharge patterns [53]. For example, batteries suffer from degradation if they go through irregular and sudden charge and discharge cycles. To increase their lifetime, it is recommended to use an ESS that can handle power spikes and allow the batteries to operate based on the average demand [46]. The work of [146] integrated an UC with a BESS to form a HESS. The experimental results reported a 40% improvement in the BESS’s lifespan [146]. In [61], a stationary BESS is combined with a portable BESS to reduce the degradation of batteries. Furthermore, a SC has been used with the BESS to avoid the premature failure of the BESS [61] and protect the battery from overvoltage [71].

5.3. Geographical Limitations

Geographical location and climate conditions should be considered before selecting the candidate ESSs for a HESS. For example, a HESS composed of a PS–BESS may not be constructed in highly populated cities where the ground conditions do not support the feasibility of PS, which requires certain geographical conditions. Similarly, extreme temperature regions limit the use of certain types of HESSs. For instance, HFCs and BESSs have been combined due to the temperature-weighted capability of LIBs [68]. Likewise, a HSS–BESS has been proposed for a remote island in Japan due to its efficacy in a land use-restricted area [138]. Moreover, alkaline–hydrogen batteries and LIBs have been integrated to form a HESS to benefit from the LIB’s capability to withstand rainy conditions and high temperatures in Africa [68].

5.4. Financial Factors

Costs associated with various technologies are always a key factor for power system planning. Thus, selecting a HESS that can meet the required services at minimal cost is a key issue. A HESS using a SC–BESS has been selected in [92] due to its low cost and other considerations such as high power density and technology readiness. Cost is also considered as a key selection criterion for selecting a HESS (UC –BESS) in EV applications [69,169]. The work in [69] shows that the cost of operating an integrated energy hub can be reduced by 35.29% by using a HSS–BESS instead of using a BESS alone. The work in [131] demonstrates that the levelized energy cost of a wind farm can be reduced with the integration of a SC–BESS (2.5 c/kWh) compared to a BESS alone (2.7 c/kWh) and a SC alone (5.9 c/kWh). The work in [149] demonstrates a reduction in the total cost of energy through integrating a HESS (SC–BESS) with a PV–wind off-grid system. Furthermore, a real-world case study in California, reported in [130], illustrates an increase of 6.4% in the revenue generated by the energy arbitrage through the deployment of a HESS instead of a BESS alone.

5.5. Ramping Capability

Ramping capability is an important factor in selecting the candidate ESSs for a HESS. ESSs with a slow response time cannot be useful for rapidly changing loads. We need, at least, one ESS with a fast response time in the HESS used for such applications. In [131], a SC and BESS have been combined due to the high ramp capability of the SC, making it a suitable solution to cope with pulsed loads. Other studies [92,110] have used a SC with a BESS due to the fast dynamic response of this HESS. Similarly, the fast dynamic response of a SMES–BESS makes this combination a suitable HESS for improving DC-bus voltage stability under disturbances [44]. Moreover, an UC and BESS have been used as a HESS, due to the quick response time of the UC, for a residential microgrid in [115] and for a PHEV-V2G application in [120]. In [43], a SC-BESS has been used for smoothing power variations due to the SC’s fast response time.

5.6. Efficiency and Energy Loss

High efficiency and minimum losses are desirable characteristics for HESSs. The choice of a HESS is greatly influenced by how much energy loss it will cause. Therefore, complementary ESSs are integrated to overcome each other’s limitations and improve the HESS’s efficiency. For example, the authors in [112] have combined a SC and BESS to enhance the storage efficiency of a DC microgrid. A similar combination has been used in [146] for hybrid electric vehicles. A SMES and a BESS have been integrated in [44] to benefit from the high efficiency and low energy loss of the combination. In [112], to increase the efficiency of PHEVs, a HESS (SC–LIB) has been used.

5.7. Suitability for Intended Objective

The intended objective of a HESS is another key factor for the selection of candidate ESSs for HESSs. A variable speed pump storage (VSPS) unit is integrated with a BESS and SC to provide peak load shaving and frequency modulation capabilities in a power system [116]. In [156], after a comprehensive comparison of the performance of several ESSs regarding frequency regulation, the SC–BESS (LIB) system is selected due to its effectiveness in improving the primary frequency modulation of a power system. Similarly, a SC–BESS is used in the UK’s national grid due to the suitability of this combination for providing a firm frequency response [151]. A SC–BESS has been used as a VSG in an isolated AC microgrid in [158]. A HESS (FES–BESS) has been used in [58] to improve the frequency response of a grid-connected wind farm. Furthermore, a HESS consisting of a SC–BESS–PEMFC has been formed in [44] to combine the diverse characteristics of the individual ESSs to handle short-term and long-term disturbances in the power system.

Figure 19 compares the frequency of use of each HESS selection criterion in the selected papers for this survey. The figure reveals that lifetime enhancement, followed by power and energy density, is the most widely used HESS selection criterion.

6. Conclusions and Future Research Directions

This paper has reviewed the modeling and control of HESSs for power system operation studies. Various HESS control strategies, including artificial intelligence and optimization methods, have been reviewed and reported. This paper has also discussed the effectiveness of HESSs in meeting the demands of net zero transitioning. It has been extensively discussed that HESSs can offer significant benefits for enhancing power system stability and achieving net zero targets. To fully leverage the advantages of HESSs, its ESS components should be carefully selected, considering all relevant criteria. Based on the comprehensive review presented, the future research directions are outlined below.

(a)

A proper HESS design is important to make it more efficient than a single ESS. In this regard, the selection of an appropriate combination of ESSs is crucial. The HESS selection criteria have been presented and discussed in this paper. Sizing the individual ESSs of a HESS should be carefully performed using appropriate optimization methods. State-of-the-art relevant optimization examples have been provided in this review.

(b)

A BESS is mostly integrated with a SMES and SC to form a HESS. Very few works have considered the formation of a HESS with thermal storage. However, with the extension of energy hubs and multi-carrier energy systems, this can be a suitable research direction for future studies.

(c)

HESSs formed by combining multiple (more than two) ESSs need further investigation on their optimum design and efficacy in providing ancillary services. While these HESSs may be able to provide higher storage capability and flexibility, their design and implementation may encounter more complexities. Thus, evaluating the benefits of these HESSs requires further study.

(d)

HESSs can be effective in improving the voltage and frequency stability status of a power system, as can be seen from the literature works. However, the impacts of HESSs on other stability issues (such as control interactions and sub-synchronous and super-synchronous oscillations), as well as other types of stability emerging in renewable energy-integrated power systems (such as converter-driven stability), need further study.

(e)

HESSs can play a particularly useful role in net zero transitioning by providing a flexible storage capacity to overcome the intermittency of renewable generation and providing cost-effective grid services for the stable operation of the power system. However, although ESSs are integrated for advancing towards net zero emission targets, the hidden environmental impacts of the manufacturing process of ESSs, such as batteries, should be considered and analyzed while selecting the optimum combinations of ESSs for a HESS. Alternative storage systems should replace ESSs that result in high emissions during their manufacturing process.

(f)

There are very few grid codes for HESSs in practical power systems, while it is a key issue for the extensive use of HESSs in practice. Further research needs to be conducted in this area.

(g)

Simplified degradation models for ESSs have been used in many studies. In the future, instead of using simplified degradation models for ESSs, more accurate and dynamic models that can cover the relevant constraints and factors involved in ESS lifetime degradation and ESS replacement costs should be considered to determine the most economical combination of ESSs for forming a HESS or to perform the cost-benefit analysis of a HESS.

(h)

Multi-objective optimization and multi-criteria decision-making for HESSs can be considered in future research to simultaneously meet the technical, economic, and environmental objectives of HESS planning.