A Review on Hybrid Circuit Breakers for DC Applications

Direct current (dc) power system protection presents a major challenge due to its unique characteristics, namely the absence of natural zero current crossing points. Thanks to their features, dc protection devices allow the safe commutation of the fault current. Selective protection and fast fault isolation are key features that any dc circuit breaker should feature, assuring minimal power outages and the effective protection of sensitive electronic components. To overcome the liabilities of mechanical circuit breakers (MCBs) and solid-state circuit breakers (SSCBs), several novel concepts of hybrid circuit breakers (HCBs) have been proposed over the years, to take advantage of the benefits of both the MCBs and SSCBs. This article presents an up-to-date state-of-the-art on the technologies applied to dc HCBs and describes novel HCB concepts. Design considerations, challenges, and future trends are also discussed.


I. INTRODUCTION
Direct current (dc) systems have been considered as either an alternative to conventional ac distribution systems or as a complement to ac, in the form of low voltage dc (LVDC) microgrids, medium voltage dc (MVDC), or high voltage dc (HVDC) transmission.The world's first commercial HVDC transmission link in operation was the Gotland HVDC Link, commissioned in 1954 in the Baltic Sea [1].Since then, a multitude of dc projects have been successfully put into operation, such as the Zhoushan 200 kV multiterminal HVDC system [2].To assure protection and selectivity, circuit breakers (CBs) integrated in HVDC are placed at both ends of all transmission lines.The way the terminals are interconnected influences the power transmitted by generators under faulty conditions, demonstrating the importance of using dc circuit breakers (DCCBs) [3].
Dc power systems are characterized by their low impedance, compared with peer ac systems, resulting in fast propagation of dc faults [4].In addition, the rate of rise of the fault current in dc systems is 5 to 10 times faster than in ac systems [5], and conventional ac protection devices are unable to deal with such demanding fault conditions.With that in mind, DCCBs should meet a few requirements: fast response and fault interruption, low conduction losses under normal operation, reduced electric arc energy and duration, along with the ability to support significant overcurrents and overvoltages [6].
Early elementary DCCBs started being used in the early 1900s [7].Further research on the design of DCCBs dates back to the 1970s [8], [9].The first DCCBs were based on a mechanical switch aided by a capacitor and an inductor [10].The interruption time was around the hundreds of milliseconds.Nowadays, DCCB technologies evolved to faster, more efficient, and cost-effective devices, capable of attaining effective fault cancellation within a few microseconds.
DCCBs can be grouped into three main categories, according to their constitution and working principles: mechanical, solid-state, and hybrid CBs.Mechanical circuit breakers (MCBs) are characterized by low conduction losses under normal operation and by the integration of galvanic isolation.In turn, their poor response time in case of a fault, along with the occurrence of a long-duration electric arc caused by the separation of its contacts are considered the main drawbacks of mechanical CBs [11], [12], [13].Solid-state circuit breakers (SSCBs) totally rely on power electronic devices and, therefore, have no moving parts, thus providing fast current interruption at the cost of significant conduction losses.Such losses induce heating, meaning the thermal capacity of the employed semiconductors has to be taken into account [14].
To avoid the drawbacks of SSCBs with regards to on-state losses, hybrid circuit breakers (HCBs) were proposed.With regards to their physical constitution, HCBs are distinguished from other DCCB technologies by the integration of both mechanical parts and solid-state components.Thanks to such combination, HCBs meet the requirements of fast operation and low conduction losses, while demonstrating the ability to effectively limit the fault current [15], [16], [17], [18], [19].The combination of dc-compatible mechanical switches with solid-state components, by itself, is sufficient to make some HCB designs fully functional in dc systems.To reduce arcing and wear of mechanical switches, some HCB designs for dc applications integrate passive components or other auxiliary mechanisms that force the current to zero in an artificial manner.
Thanks to their features, HCBs are today deemed as the DCCBs showing the best performance regarding efficiency, response time, current and voltage ratings [20].
Apart from excellent fault-breaking capability, HCBs must fulfill a few requirements to be suitable for dc applications, such as the following [21].
r Fast response to interrupt the current, preventing the fault current from reaching damaging values to the equipment.
r Low conduction losses, to assure maximum efficiency and minimum thermal stress on the HCB.
r Reduced arcing between the mechanical contacts, to minimize the erosion of the contacts.
r Full controllability of the HCB system, allowing com- plete configuration of the HCB conducting and breaking states.HCBs can be characterized according to the voltage ratings they stand.Essentially, they can be grouped in three main categories -HVDC, MVDC, and LVDC.HCBs designed for HVDC applications (higher than 33 kV) are meant for high power applications such as substations, dc buses, offshore windfarms [22], large hydroelectric dams [23], among others.HCBs with features that make them applicable for MVDC (1-33 kV) are used in applications that include dc rail systems [24], shipboard, submarine systems [25], [26], aircraft or solar arrays [27], [28].HCBs designed for LVDC systems are frequently applied to protect datacentres [29], dc microgrids [30], [31], [32] and its sensitive components, especially in the domestic environment, such as loads (electronic components, electric vehicles), energy storage systems and distributed energy resources.For niche applications implying bidirectional power flow, as it is the case of energy storage systems or electric vehicle (EV) charging, HCBs integrating bidirectional breaking capability are usually preferred [33].
Topics like system architecture, interactions between distinctive grids configurations, and adopted converter topologies are critical considerations that must be taken into account when projecting a dc protection system.With that in mind, HCBs have been extensively analyzed lately, in both industry [34] and scientific community.Over the years, substantial developments have been made in terms of response time, power losses, reliability, efficiency, lifetime, voltage and current ratings, and used components/materials.For instance, the replacement of traditional Silicon-based semiconductors by novel wide bandgap (WBG) devices has proven effective to further improve the efficiency and support more demanding operating conditions [35], [36].
The literature clearly lacks a study that compiles relevant information regarding the most recent HCBs and their features, while conducting a critical overview about them.Indeed, the most recent and most cited paper fully dedicated to HCB technologies was published back in 2015 [14].In [14], HCB technologies for ac and dc systems are presented, without establishing a distinction between them.Besides, the authors of [14] do not provide detailed HCB design guidelines neither a parametric comparison among HCBs suitable for dc applications.
Ever since, many advances have been made in this area of interest.Other review papers published in the meantime focus on dc protection technologies other than hybrid breakers [37], [38], [39], [40], including solid-state, mechanical, and Z-source devices.Accordingly, this article presents a literature review with some background information related to HCBs for dc systems, their appearance, the basic configuration of HCBs, the working principles, constitution, overview of the recent developments and emerging research, and identification of future trends.Most prominent developments, made since the last review paper on HCBs, addressed in this article include the optimization of the commutation operation [41], [42], the adoption of alternative, more efficient semiconductor devices [43], [44] such as WBG, the reduction in the HCB cost [45], the integration of bidirectional breaking capability [46], the improvement in performance of each component, simpler control of the HCB, ratings it can achieve, faster operation, size and cost optimization.Altogether, such developments led to HCBs with enhanced breaking capability, higher efficiency and reliability, and shorter response times.
The rest of this article is organized as follows.In Section II a timeframe relative to the history of HCBs is presented, a description of the basic constitution and working principles of a conventional HCB is discussed in Section III.Section IV discusses the possibilities for the main components' selection of an HCB, as reported in the literature.Section V identifies solutions developed with the intent of reducing the costs of implementation.Section VI performs a comparative analysis of some of the most prominent HCB topologies, taking into account aspects like target applications, cost of implementation and volume.Section VII provides insights on the challenges and future trends for this field.Finally, Section VIII concludes this article.
Within the last decade, particularly prominent and relevant achievements in the field of HCBs in real-world applications were attained, as indeed demonstrated in the timeline shown in Fig. 1.Such achievements translate into enhanced current and voltage interrupting capacity and shorter response time.
In 2014, Alstom Grid finished tests for the 120 kV HVDC breaker prototype named ultrafast mechatronic circuitbreaker.It consists of a main branch with IGBTs and an ultrafast mechanical disconnector, paralleled with an auxiliary branch composed of thyristors and an energy dissipation branch [56].This design demonstrated the ability to withstand a peak current of 7.6 kA and interrupt it in 2.5 ms.The Global Energy Interconnection Research Institute proposed, in 2014, a modular solution consisting of a 200 kV, 15 kA HCB [3], which was subsequently upgraded to a 500 kV, 25 kA version in 2020 [57].
Also in 2014, SCiBreak developed an HCB that uses the concept of a VSC-assisted resonant current (VARC).Instead of using the classical resonant method using a precharged capacitor, a VSC generates high amplitude pulses resulting in very fast current interruption [58], [59].
The State Grid Smart Grid Research Institute achieved, in 2015, the manufacture of a full-bridge-based hybrid DCCB, with a rated voltage of 200 kV, opening time of 3 ms, and current interrupting capacity of 15 kA, with the intent of protecting Zhoushan MTDC grid [2], [60].The authors of [61] proposed an HCB based on a WBG ETO and an ultrafast mechanical switch in 2017.
In 2018, an HCB with a 500 kV modular cascaded structure with an interrupting current capability of 25 kA, was applied to the Zhangbei MTDC grid [62], [63].Also, in 2018, an HCB that resorts to the molten metal bridge of the contacts of the mechanical switch for current commutation is proposed in [64].In 2018, an HCB prototype developed using superconductive materials proved its effectiveness [65].
In 2020, SCiBreak successfully tested the prototype of the VARC HCB for 120 kV/15 kA at the KEMA labs, in The Netherlands [66].Fast forward to 2022, the VARC is applied in the protection of railway systems of Sweden [24], [67].

III. ARCHITECTURE AND PRINCIPLES OF OPERATION OF HCBS A. CLASSICAL CONFIGURATION
HCBs, which result from the combination of MCBs and SS-CBs, are capable of delivering fast arcless current interruption [68].MCBs are characterized by their low contact resistance (of a few micro-ohm) in the closed position and galvanic isolation in the open position.Nevertheless, MCBs require a significant amount of time to interrupt the fault current and need for an electric arc to form which, in turn, causes contact erosion and may even result in injury to personnel [55].On the other hand, SSCBs only use power semiconductors and are known for their fast operation and reliability.In turn, the major disadvantages of SSCBs consist of high on-resistance and sensitivity to transient overvoltages and overcurrents, resulting in conduction losses and heat [17], [69].
Despite the great diversity of possible designs for an HCB, they all consist of the following three common branches.
1) A primary branch, where current flows under normal condition; it contains, at least, a mechanical switch.2) A main breaker, that uses semiconductors to conduct the current for short periods of time, while the mechanical breaker opens.3) An energy dissipation branch that consists of surge arresters (usually varistors), to absorb the energy stored during the fault.Fig. 2 shows the simplest configuration of an HCB.Current flows through the mechanical switch under normal operating conditions.Once a fault occurs, the mechanical switch opens, creating an arc.The arc voltage increases until it forces the current to commutate to the semiconductors, which are turned OFF once the mechanical switch is able to withstand the overvoltage defined by the varistor.When the semiconductors are turned OFF, the varistor rapidly extinguishes the fault current.
Fig. 3 shows the current waveforms at the different components of the HCB during a current interruption event.After the fault detection, the mechanical switch contacts separate at instant T open and the main breaker begins its conduction.At instant T com , the current commutation to the main breaker is completed and at instant T of f the semiconductors of the main breaker are turned OFF, inducing the appearance of an overvoltage across the varistor that establishes its conduction.Once the system voltage drops below the varistor operating voltage, the varistor impedance increases and the current approaches zero.The varistor is also responsible for demagnetizing the system inductance.
HCBs can be damaged due to overvoltage surges, namely the transient recovery voltage.Therefore, it is important to limit the voltage applied to the HCB, particularly to assure the integrity of the power electronic components.To protect HCBs from overvoltage spikes at turn-OFF, additional elements are used to dissipate the fault energy, such as capacitors, resistors, freewheeling diodes, and metal oxide varistors (MOV) [70], [71].
The semiconductors of the main breaker must conduct for hundreds of microseconds, for the mechanical switch to regain its dielectric strength.Otherwise, the mechanical switch might not withstand the transient recovery voltage, resulting in arc reignition [72].
HCBs can either have a bidirectional or unidirectional breaking capability [73].In the case of bidirectional interruption capability, the breaker can interrupt current in any direction, i.e., regardless of the transmission line configuration and fault location.To accomplish bidirectional breaking capability, a few additional components shall be integrated in the conventional HCB to allow interruption capability from both directions [46], [74], [75].Bidirectional breaking capability can be introduced through the utilization of two back-to-back connected IGBTs as main breakers [46], through a set of three thyristors and two diodes [74] or through two commoncathode SCRs with antiparallel diodes [75].

B. TYPE OF COMMUTATION
The commutation mechanisms of HCBs differ from those adopted in MCBs and SSCBs, given their inherently distinct physical structure and operation principles.
Based on the type of HCB, current commutation to the semiconductors' branch can be accomplished either through the voltage of the electric arc formed in the mechanical switch -natural commutation -or by using additional componentsforced commutation.

1) NATURAL COMMUTATION
Usually, HCBs adopting natural commutation rely on the arc voltage that builds up across the contacts of the mechanical switch, to commutate the current to the semiconductor's branch [76].Nevertheless, this structure is not recommended for HVDC systems.Despite the simplicity intrinsic to the physical implementation of this approach, this structure is not recommended for HVDC systems.As the sum of the on-state voltage of the semiconductors used in the commutation branch is much higher than the arc voltage created in the primary branch (typically in the range of several tens of volts), natural commutation becomes unfeasible for HVDC [77].Moreover, the risk of arc restriking is increased.On that basis, HCBs employing natural commutation are mostly considered for low-to medium-voltage applications.
Natural commutation can be, in certain HCB designs, assisted by modified mechanical switches.In [78], a high-speed mechanical switch uses a vacuum mechanism and External Transverse Magnetic Field (ETMF) to eliminate the use of a load commutating switch (LCS), minimizing the conduction losses associated with the use of this auxiliary element.ETMF rapidly increases the arc voltage, reducing the period required for current commutation.A transverse magnetic field is obtained using permanent magnets in the contacts of the mechanical switch.An increase in the magnetic flux density is associated with the reduction of the breaking arc duration [79], allowing fast recovery of the dielectric characteristics.Despite the gains attained with this modified mechanical switch in terms of arc duration, this design overlooks the risk of contact erosion.Therefore, natural commutation supported on this modified design might not reveal suitable for applications expecting recurring breaking operations.

2) FORCED COMMUTATION
ABB was the first to present a proactive HCB for HVDC, eliminating the significant on-state losses related to the use of SSCBs, while also limiting the fault current [15], [80].In the proposed solution, an auxiliary breaker, composed of semiconductors, is used in the primary branch to force the current to commutate to the main breaker.The auxiliary breaker, frequently referred to as LCS, provides a low-loss current path under normal operation.The LCS uses fewer semiconductors than the main breaker [81].By opting for a type of semiconductor known for low on-state voltage in the LCS, losses can be reduced.Under fault-free operation, the LCS is closed, allowing current to flow through it.Once a fault occurs, the LCS is turned OFF, commutating the current to the main breaker so that the mechanical switch can open.Afterward, the current commutates to the varistor, where it is dissipated, and the voltage is limited.Considering that the LCS is responsible for commutating the current, the wear caused by the formation of an electric arc in the mechanical switch is limited.The mechanical switch is responsible for disconnecting the LCS, preventing it from being subjected to the peak voltage generated during the breaking operation.
Since the current only circulates in the main breaker in case of a fault event, the main breaker is not exposed to thermal overload, thus not requiring an active cooling system.This is indeed a major advantage of HCBs over SSCBs, which typically require cooling mechanisms.When IGBTs are used in LCS, the commutation time is expected to be in the range of tens of microseconds [82].
A major disadvantage of HCBs relying on an LCS structure is that the main breaker has to hold its conduction until the mechanical switch has enough voltage withstand capability to tolerate the counter-voltage generated by the MOV, to turn-OFF and commutate the fault current to the energy absorbing branch [83].This way, two delay periods are verified: 1) the commutation from the primary to the main breaker; and 2) the commutation from the main breaker to the MOV.These delays translate into increased fault current and fault interrupting times.To overcome this issue, recent studies suggested sequential tripping strategies [84], [85].
The selection of the HCB components should consider the peak current that flows in the LCS, in the main breaker, and the maximum energy absorbed by the MOV, in the worst fault scenario [4].Typically, the LCS has lower voltage rating than the rating of the HCB system, which reduces the total power losses of the breaker.Nevertheless, the voltage rating of LCS must still exceed the on-state voltage of the main breaker [86].
Even though the adoption of an LCS is the most common strategy to achieve forced commutation in an HCB, alternative methods have also been proposed.An alternative approach for attaining forced commutation, presented in [73], relies on the use of an active resonant circuit.Consisting of a three-winding coupled inductor and a charged capacitor to inject inverse current, it forces the appearance of zero-crossing points.As a result of such characteristic, the mechanical switch can open without arc formation [14].Also, this approach allows reduction in the conduction losses and improvement of the HCB reliability, thanks to the obviation of semiconductors.Semiconductor devices are among the elements that limit the most the lifetime of power electronics systems.In turn, the adoption of coupled inductors increases both the cost of implementation of the solution and the response time.
Another solution for the current commutation relies on two magnetically coupled inductors [87], as shown in Fig. 4.Under normal operation, current flows through both inductor L1 and the mechanical switch.When a fault occurs, the secondary winding (L2) is turned ON by closing the main breaker switch (consisting of semiconductors) and the mechanical switch is opened, thus forming an electric arc that generates a counter-voltage.Both the counter-voltage and the magnetic coupling force the current to flow through L2, since L2 has lower inductance than L1.This approach allows savings in the number of semiconductors and reduction in the conduction losses, however, it can be slow and hard to control, causing undesired oscillations.
Some HCB designs include the use of an oscillating circuit to force the current to zero, allowing the opening of the mechanical switch with nearly arcless condition.
The scheme depicted in [88] claims arc-less breaking operation for the mechanical switch and the maintenance of the dc bus voltage, using precharged capacitors in both the converter side and the transmission lines.The authors of [89] and [90] propose an identical approach, based on the use of a transient commutation current injector (TCCI) in replacement of the LCS.This device is also placed in the main conduction branch, providing negligible power losses.It generates a current pulse with an amplitude close to the fault's current, enabling the current commutation.It also provides near-zero voltage, allowing the mechanical contacts to separate without the formation of an electric arc.According to results presented in [88], the time required to commutate the current to the auxiliary branch is 7.2 μs, while the total fault elimination process takes about 310 μs to complete.
A TCCI corresponds to a switched-mode power electronics circuit, whose design is similar to the one of high-current high-precision pulsed power sources used for controlling magnetic fields.The TCCI basically discharges precharged capacitors in a controlled way, in order to generate the current pulse that enables effective current commutation.Two MOS-FETs are controlled through a dual-band hysteretic control to produce a fast initial current rise.The drawback of this design lies on the introduction of an ac high-frequency ripple current, derived from eventual inaccuracies in the TCCI's control.Besides, the hardware and control complexity may compromise the effective breaking operation.Concerns regarding the reliability of the system and cost of implementation are also present.Fig. 5 shows the schematic representation of an HCB containing a TCCI.
The authors of [60] replace the LCS with a structure composed of a converter with a full-bridge configuration and a capacitor.During normal operation, the current passes through the upper arm IGBT and the lower arm diode.When a trip signal is received, the IGBTs are turned OFF and the current charges the capacitor.The counter-voltage generated commutates the current to the parallel branch, bringing the current of the normal path to zero.This way, the contacts of the mechanical switch open without forming an arc.
These structures combine the benefits of semiconductors and passive components to provide fast and reliable breaking operation.Their only drawback is the complexity inherent to them.

IV. MAIN HCB COMPONENTS' SELECTION A. MECHANICAL SWITCHES
Mechanical switches are characterized by their low contact resistance, which can be as low as 10 μ [91].To make mechanical switches suitable for dc protection applications, it is necessary not only to integrate auxiliary mechanisms that condition the arc voltage and thus force the current to zero [27], but also to improve their speed.
The selection of mechanical switches for HCBs needs to be supported on a careful evaluation of the following aspects.
An important parameter to consider in the choice of the mechanical switch is the system's recovery voltage withstand capability, a metric helpful for defining the chances for avoidance of arc restriking.This parameter is greatly influenced by the arc quenching fluid pressure, temperature, and electric field at the contacts [27].Another parameter to consider is the arc's resistance, which is impacted by the degree of ionization.The arc resistance increases with the decrease in the number of ionized particles between the contacts.The distance between the contacts and cross-section of the contacts also need to be considered (arc resistance increases with the decrease in the contact cross-section) [92].
Currently, there is not a unique and universal criterion for classification of mechanical switches employed in HCBs.The authors of [93] classify mechanical switches according to their current interruption capability as disconnecting switches and CBs.Disconnecting switches are only considered when an artificial zero-current is accomplished.In cases where the artificial zero-current is not feasible, CBs are typically considered in the implementation.For such conditions, it is the arc voltage that ensures the current commutation to the parallel branch.
Within the group of CBs, it is possible to establish classifications based on the actuating mechanism, the medium used to quench the electric arc [94], or the materials of the arcing contacts.
Regarding the actuating mechanisms, it is common to rely on the Thomson Coil and piezoelectric actuators, since they can be activated in hundreds of microseconds.The Thomson Coil is, perhaps, the most popular mechanism in literature for ultrafast switching [93].Indeed, several studies reported the use of a mechanical actuator based on a Thomson coil [55], [58], [61], [95].It consists of an electrically conductive disk, placed near a planar coil, that repels the disk once current flows through it, accomplishing ultrafast operation.Apart from the technologies based on electromagnetic repulsion, like series coil switches [96], double-sided coil switches [97], or moving coil actuators [98]; piezoelectric actuators convert electrical energy in mechanical displacement, allowing fast operation and high precision positioning.
Commonly used mechanical switches, termed ultrafast disconnectors (UFDs), can be realized in different conceptual ways.The more common architectures are based on highspeed electromagnetic repulsion mechanism or gas-insulated switchgear technology [99].Regardless of the adopted mechanism, any UFD should allow an opening time of no more than 2 ms [100].
To attain arc quenching, the arc is redirected to be cooled and subsequently extinguished.Vacuum CBs have a vacuum chamber that provides a low pressure, fast insulation recovery medium to extinguish the arc and provides quick recovery of the dielectric strength.The arc formation leads to the ionization of the evaporated metal particles of the contacts, creating an electric field.In turn, the current circulation creates a magnetic field.Since there are very few gas molecules, when the current crosses zero, the arc will spread diffusely under the influence of the magnetic field to cool down, allowing the arc to recondense and recover the dielectric strength [101].The arc's extinction is greatly influenced by the contacts' geometry and material.
Other CBs use sulfur hexafluoride (SF6) gas under pressure as an arc extinguishing medium.Authors of [102] proposed an HCB that combines the use of an SF6 switch and a vacuum switch connected in series, to form the mechanical switch, aiming to enhance the dielectric recovery ability.When a fault is detected, the contacts are separated and, simultaneously, SF6 under pressure is injected in the arc quenching chamber.
Other quenching methods include the use of oil, air blast, or CO 2 .Some of these strategies are being avoided because of the environmental and safety issues that they imply: the consequences of oil spills on the environment and the consequent risk of explosion or the effect of SF6 on the global warming are some of the negative impacts commonly related to such materials [103], [104].
Fig. 6 presents a scheme that classifies the mechanical switches according to three criteria: 1) the need/absence of zero crossing points for operation; 2) the actuating mechanism; and 3) medium to quench the arc.
Table 1 further elaborates on the actuating mechanisms and arc quenching mechanisms, pointing out some of the advantages and disadvantages of each one of them.
The selection of the materials for arcing contacts is a challenging task that usually results from a trade-off between the arc resistance properties and the target end-use where the mechanical switch is to be integrated [105].Among the desirable properties of the contact material, the following are particularly relevant.
1) Materials with neither too low nor too high vapor pressure.Low vapor pressure materials ensure that the interrupter is more likely to chop the current, extinguishing the arc at lower current.High vapor pressure allows vapor to remain even at zero current, making arc interruption more demanding [106].
2) Good electrical conductivity to minimize losses during continuous operation.This criterion should be prioritized in applications where frequent breaking operations are anticipated [107].3) High thermal conductivity to dissipate generated heat after current interruption [107].4) Good dielectric properties to ensure quick recovery capability.5) High current interruption capability.6) Low weld strength: contacts in the vacuum invariably weld during the closure of the contacts, due to the decrease of the dielectric strength.At some point the dielectric is not sufficient, occurring the formation of an arc.The contacts melt at local points, before being pressed together.To facilitate opening, easily fractured welds are important.This criterion should be particularly weighted in applications whose fault current is anticipated to be high, as it is the case of capacitive loads [108].Pure element materials are unable to cumulative fulfill such requirements.This is the reason why alloys are created to hybridize the advantages of each material.Aluminum-, silverand copper-based materials reveal as the most prominent to meet most of the aforementioned criteria, reason why they are usually considered for most implementations [107], [109].Generally, as the proportion of conductive metal (silver, copper, or aluminum) increases, contact resistance decreases and electrical and thermal conductivity increase, at the cost of increasing the probability of contact erosion and contact welding.In contrast, as the refractory metal (tungsten, chromium, niobium, etc.) content increases, contact wear decreases and there is less likelihood of contact welding.Adjusting material particle sizes, choosing additives, and altering furnace temperatures all play a role in the final properties of the selected contact material [110].

B. SEMICONDUCTORS
In general, semiconductors have the following beneficial characteristics over mechanical switching devices: fast, arcless and soundless operation, long lifespan, and high reliability.Since power semiconductors are a key component in the design of an HCB, the semiconductor technology considered in the design of HCBs plays a major role in the performance and speed of an HCB.The adoption of a certain semiconductor technology will depend on the purpose of the HCB and the characteristics of the dc energy system, such as rated current, expected overcurrent and overvoltage limits and desirable interruption time.The number of semiconductors connected in series is determined by the system voltage, whereas the amount of parallel connected semiconductors establishes the current interruption capability of the HCB [111].
Various semiconductors technologies can be selected for the implementation of HCBs, such as gate turn-off thyristors (GTOs), insulated gate bipolar transistors (IGBTs), integrated gate-commutated thyristors (IGCTs), metal-oxide semiconductor field-effect transistors (MOSFETs).Silicon controlled rectifiers (SCRs) are often used as auxiliary commutation devices [112].In high-power applications, preferred power electronic devices include IGBTs or injection-enhanced gate transistors (IEGTs), given their high current breaking capability.IGBTs are more suitable than IGCTs for the HCB application because of their much higher ability for singletime turn-OFFs.In turn, IEGTs are an optimized type of IGBT with low on-state voltage drop [57].On the other hand, IGCTs provide low loss operation, which can speed up the commutation in natural commutation HCBs [20].The dominant characteristic of IGCTs is their speed, with operating times lower than 100 μs.However, they might require heat sinks if used in high power applications.The main breaker may not require the use of a cooling system since current does not flow through it during normal condition.On the other hand, the employment of an active cooling system is recommended for HCBs implemented in HVDC systems that employ an LCS.A novel active cooling system named "Cothex" [113], based on passive two-phase evaporation/condensation concept with air cooling, can be considered for stand-alone installations as it is the case of HCBs for HVDC systems.
In [114], a combination of IGBTs and IGCTs is used to obtain the best compromise between turn-OFF capability from IGBTs and cost reduction, characteristic of IGCTs.The authors claim that this scheme costs less than half of what a regular main breaker would cost, even though this topology uses an additional branch composed of a resonant circuit.For higher voltage ratings, the cost savings are limited because of the considerably larger capacitors.A similar approach is proposed in [115], resorting to both thyristors and IGBTs in the main breaker.The use of half-bridges built from IGBTs allows the charging of parallel capacitors.Once charged, these capacitors provide a negative voltage across the thyristors' branch, suitable to turn them OFF and direct the fault current to the varistor branch.The architecture of this HCB is complex and demands the charging of one capacitor at a time, sequentially.
In comparison with fully controlled semiconductors, thyristors stand out both in current carrying capability and costeffectiveness.A thyristor could conduct 15 times the rated current briefly, and the price could be as low as 1/10 of an IGBT in the same power rating.The only drawback of thyristors lies on the turn-OFF characteristic, which implies zero current crossing [116].
As an alternative to silicon-based semiconductors, the use of WBG semiconductors brings great advantages in terms of volume and response time.WBG semiconductors also attain superior performance metrics when compared with silicon peers in terms of power density, switching frequency, and losses [117].WBG materials are framed in the third generation of power semiconductors and include materials like silicon carbide (SiC) and gallium nitride (GaN).Such features make WBG semiconductors highly suitable for high-power, high-temperature applications.Despite the significant advantages that WBG semiconductors provide, their cost and the lack of maturity of this technology remain the main obstacles to their widespread use [118].
The authors of [94] performed a comparison between three distinctive HCBs: the first uses only a vacuum mechanical CB; the second uses a vacuum mechanical switch and a Si MOSFET; and the third uses a vacuum mechanical switch and a SiC MESFET.Results showed the arc quenching is faster in the situation where the SiC MESFET is used, and slower when only the vacuum breaker is applied.Given these results and others that demonstrate that the maximum fault current is lower, the MESFET is the preferred technology, ensuring, also, a faster current interruption.In [119], different semiconductor devices, including Si MOSFET, SiC MOSFET, Si CoolMOS, SiC MOSFET, and SiC JFET, are compared with assess their turn-OFF capability, for a rated breakdown voltage, using the double pulse test (DPT).The SiC JFET showed the highest turn-OFF capability, the maximum current density, and the highest peak power density, followed by SiC MOSFET and Si CoolMOS, respectively.On the other hand, Si MOS-FET demonstrated the poorest figures of merit.
Table 2 highlights the characteristics of the semiconductor technologies most deployed in HCBs.

C. SNUBBER CIRCUIT
To protect the power electronic devices from overvoltage spikes at turn-OFF, some HCB configurations also integrate a snubber circuit, connected in parallel with the sensitive components of the HCB like semiconductors of the main breaker.Snubbers enhance breaking performance and limit the stress imposed to the HCB components, absorbing the MOV voltage/current spikes and reducing the losses during turn-OFF of the main breaker semiconductors.Snubbers usually consist of an RC, RCD circuit, or a varistor.The authors of [120] compared the advantages and disadvantages of different snubbers: the varistor, capacitor snubber, RC snubber, and RCD snubber.They concluded that RCD snubber presented the best performance, being able to control the oscillation caused by C and L using a diode to clamp the oscillating voltage.

TABLE 3. Snubbers Used in HCBs
As alternative to conventional snubber structures, the HCB presented in [121] uses thyristors as snubber.When a fault occurs, and once the current is commutated, the charging of the capacitor allows to demagnetize the system inductance through the varistor.The decrease in the current also lowers the voltage, forcing the current to circulate again in the capacitor's branch, generating a negative voltage and current, resulting in current oscillation.The oscillation prevents the current from being interrupted.The use of thyristors prevents such inconvenience, since thyristors turn-OFF when the current is zero, avoiding the consequent change in polarity of current [122].
More elaborated structures, like the combination of RCD and MOV technologies, gathers the merits of both in a single device [123].

D. FAULT CURRENT LIMITER
Apart from the main elements required for any HCB, it is possible to integrate additional functions like current limiters.
Fault current limiters (FCLs) can essentially be divided into FCLs based on power electronics, on inductors, or superconducting fault current limiters (SFCL).FCLs based on power electronics and inductors are reliable and provide a cheap solution to current limitation.In turn, SCFCLs are complex, costly, and might reveal unreliable.
FCLs based on power electronics use semiconductors devices to introduce an energy absorbing impedance to limit the fault current.
Meanwhile, FCLs based on inductors employ a dc reactor as an effective way to limit the fault current rising rate and its peak value.As a negative aspect, the inclusion of such dc reactor may affect the system stability and the fast operation of the HCB [125], [126].
The impact of traditional current limiting reactors on the HCB is discussed in detail in [127].The use of a limiting reactor connected in series with the CB causes the line impedance to increase, thus resulting in a lower peak fault current.Results shown in [128] confirm such trend, by measuring the current in different locations of an HVDC system, for different inductance values.On the other hand, its implementation delays the time instant at which the fault current is transferred to the energy dissipation branch and limits the peak current.Fig. 7   demonstrates how the use of a limiting reactor influences the evolution of the fault current.
The SFCLs can be grouped in resistive or reactive ones.Resistive SFCLs take advantage of the resistive quenching characteristics of superconducting materials using a superconductive coil [129], [130], [131].Reactive SFCLs are more elaborated and consist of a superconductive transformer: a superconductive coil powered by a dc source, wrapped around the iron core and two conventional coils connected in series with the circuit [132].
Under normal operation, the current flowing through the SFCL is below the critical value, promoting a superconductive state, in which a very small value of resistance -near zero -is registered.Once the current increases, caused by the occurrence of a fault, the current density J exceeds its critical value J c , leading to a rapid rise in the electric field E of the SFCL [98], defined as follows: where n denotes the number of superconductive coils.The current growth causes an increase in the temperature and, consequently, in the impedance of the SFCL.This will, subsequently, result in a descending current trend.
To reduce the cost associated with the use of SFCL and the loss of system stability, an improved HCB with self-adaptive fault current limiting capability is proposed in [125], using a parallel branch including a limiting inductor and resistor.Fig. 8 shows a schematic of the self-adaptive FCL.
When operating in a current limiting state, the dc reactor L generates a reverse voltage, v L = L • di/dt, while resistor R limits the peak current value.

E. SURGE ARRESTERS
Generally, surge arresters like MOVs are used to absorb the energy stored during the fault and to reduce the current to zero, while re-establishing the system operating voltage.A rule of thumb often adopted consists of opting for an arrester that has a protection level of 1.5 times the system's rated voltage [133], [134].The parameter design of MOV is a trade-off of margin, which has no optimization space for costs.
The use of MOVs as surge arresters provides a well-defined overvoltage protection level [135].The energy absorbed by the MOV can be expressed as follows [4]: where i is the rated current under normal operation, V is the system voltage, V cell is the voltage rating of the individual semiconductors of the main breaker.t0 is the time instant at which the fault occurs, t 1 corresponds to the main breaker's turn-OFF instant and the completion of the mechanical switch opening.The interval between t 1 and t 2 comprises the energy dissipation period in the MOV.High-energy MOVs present enhanced risk of explosion.The modular placement of MOVs reduces that risk [136].Meanwhile, alternatives to MOV banks have been brought up.It is the case of the works developed in [128], where thyristors and inductors are connected to the ground as a part of the energy dissipation process.
In [137], the MOV is not removed, but SFCLs are plugged in series with the HCB, between the dc HCB and the dc bus terminal.The use of this scheme intends to significantly reduce the size of the required surge arrester.
The complete elimination of the surge arrester is proposed in [138], where an LVDC HCB uses a capacitor and a diode as a dissipation branch.Fig. 9 shows the scheme of the proposed HCB.
Under normal operation, both low speed (LS) and high speed (HS) mechanical switches are closed and the IGBT Q is on.Once a fault is detected, switch HS receives a command signal to open.Given that IGBT Q is already on, switch HS is able to safely turn-OFF, with minimum electric arc that triggers the current commutation.IGBT Q conducts the current for sufficient time to prevent restriking of the arc and is then turned OFF.At that moment, the energy stored in the leakage inductance charges capacitor C located in the parallel branch.

TABLE 4. Preferred Technologies for the Main HCBs
Switch LS isolates the entire HCB from the grid, preventing the capacitor charging during the OFF state.
The capacitor branch forms an oscillating branch with an initial current equal to the fault current.To dissipate the energy stored in the capacitor, before the occurrence of another fault, a resistive circuit is also employed.It consists of a bidirectional current switch, composed of two IGBTs in series with a resistor Rd.The value of Rd must be estimated to have the ability of discharging the capacitor, within the established time, making sure that the fault clearing time is never exceeded.The discharging unit is represented with red color, as shown in Fig. 10.
Table 4 summarizes the typical device technologies considered for the implementation of each HCB component.Please note that the list is not exhaustive, focusing on the most commonly adopted device technologies.

V. COST REDUCTION STRATEGIES
Among other challenges, the cost of protection devices poses a significant hurdle toward the deployment of dc grids.The high cost of HCBs is greatly associated to the thorough specifications of the required power electronic devices.
Different approaches have been explored to reduce the costs associated with the implementation of HCBs.These approaches include the adoption of simpler technologies, rearrangement of components within the HCB, use of unidirectional semiconductors instead of bidirectional ones [139], [140], integration in H-bridge structures [141], or the cooperation of converters [86].Recently, the deployment of integrated CBs instead of multiple HCBs, or the use of a multiport or multiline structure, aiming to attain component sharing, reducing the number of semiconductors used in the main branch and in the LCS [142], [143] has attracted significant attraction.In [144], an HCB with an ac commercially available mechanical three-pole CB was tested, to assess its breaking capability.When its contacts open, the arc voltage feeds a power supply unit, charging a capacitor up to a specific voltage.Once such voltage is reached, a power electronic switch is turned ON and the arc is extinguished.The electronic switch stays on until the gap between the contacts of the mechanical switch reaches a sufficient dielectric strength.A gate driver with low output resistance is implemented to achieve maximum switching speed of the IGBT.Tests showed a successful operation for a current of 200 A.
Associated to the reduction of semiconductors, the works of [145] suggest the use of two double-throw UFDs in the primary branch.This way, only unidirectional power electronic devices are used in the main breaker, halving the number and cost of required semiconductors.UFDs have two positions and are used to accomplish bidirectional breaking capability.This topology was tested for a 900 V, 500 A circuit and compared with an HCB with a bidirectional main breaker.Results showed that it is possible to attain similar system-level performance, while resorting to only half of the semiconductors.
The idea of adopting a circuit arrangement to replace several components of an HCB at a dc node by a single one, termed the multiline Breaker (MLB), was presented in a patent [146].The main goal consists of reducing the number of energy absorption branches and main breakers.
Apart cost reduction, this design minimizes the stray inductance related to the HCB components.One of the most noticeable differences between the MLB concept and existing HCBs is that MLBs allow the simultaneous protection of multiple lines [147].
The works developed in [4] suggest an alternative approach to reduce the high number of semiconductors used in the main branch of a HVDC HCB, to achieve the desired interruption capability.The proposed solution consists of an interlink breaker with a Y-connected design, aiming to reduce by 25% the number of semiconductors and MOVs, by sharing the main breaker between two or more HCBs.Fig. 11 shows the configuration of the HCB proposed in [4].
Configurations like the latter one have particular interest in HVDC applications, including multiterminal dc (MTDC) systems.MTDC is becoming a promising technology for future power systems, drawing more attention from researchers and industries these days, given their ability to integrate largescale offshore power sources such as wind turbines and solar systems [148].
In MTDC systems, a dc bus is usually connected with multiple transmission lines, meaning that three or more HCBs need to be employed for adequate fault protection.To address such constraint, different solutions have been proposed, including the use of multiport HCBs, that share one main breaker for different lines and employ thyristors, diodes, and mechanical switches to commutate the fault current to the main breaker [142], [150].Many of these designs adopt a bridge-type configuration that uses 50% to 70% less semiconductor switches, thanks to the unidirectionality of diodes, which dictate the fault current direction.In practice, this means that a single unidirectional main breaker is required, independent of the fault current direction [151].The multiport structures proposed in [147], [151], and [152] also use additional thyristors, diodes, and mechanical switches to commutate the current into the shared main breaker.
The works presented in [153] suggest the coordination of MMC with the HCB operation, aiming to simplify the structure of the breaker, eliminating the main breaker and replacing it with Energy Absorbing Branches (EABs) as a cost saving solution and a way to reduce the volume occupied by the bulky components.It requires, though, some additional components like capacitors and switching devices.When a fault occurs, the current is directly transferred from the mechanical switch to the EABs.Two EABs are employed: converter-side EAB, constituted by a thyristor and a capacitor bank, and line-side EAB, composed of a diode and a capacitor bank.Likewise, the authors of [149] propose an HCB structure that uses a stack of thyristors and energy absorbing capacitors, as a replacement for the semiconductor main breaker and the MOV.
In comparison, the solution presented in [149] uses less components than [153] and the LCS only needs to withstand 5% of the dc bus voltage.The diode creates a freewheeling path when the HCB opens, and resistors connected in parallel to the capacitors dissipate the energy.This topology is specifically destined to HVDC that use MMCs, in such way that the MMCs are programmed to create zero-voltage in all arms of the converter, isolating the ac-side from the dc-side to prevent fault feeding.Fig. 12 shows the general constitution of the HCB that operates jointly with the MMC.
This article performs a comparison between the conventional HCB (composed of LCS, UFD, main breaker, and MOV), the proposed HCB and the topology presented in [153].Results showed that the current is interrupted within 4 ms in the proposed solution and in about 10 ms for the conventional structure, resulting in increased peak fault current for the conventional structure.On the other hand, the voltage reached in the LCS is considerably larger in the works of [153] and [149], when compared with the conventional structure, which reflects in increased power losses and relevant voltage  surges, emphasizing the major drawback of the structures proposed in [153] and [149].
An integrated commutation compound power-electronicbased HCB (IC-HCB) is presented in [154], arguing a reduction in the volume and cost by 42.4%, in comparison to a conventional HCB.It consists of a primary branch for normal current flow, a resonant branch, and a transfer branch.The transfer branch contains an oscillation switch and a main compound switch, that combines the joint action of IGBTs and IGCTs.The oscillation switch performs two actions: generating oscillating currents, through the action of a precharged capacitor; and serving as commutating switch.Some diodes are added to provide a current path in the discharge stage of the capacitor.A prototype was built and capable of effectively interrupting 10 kV/40 kA in 2.87 ms.
One of the most expensive and noble elements of HCBs is the surge arrester.Typical HCB configuration either employ a single bulky MOV or multiple series-connected devices, with enormous voltage rating.Modularity provides advantage with regards to cost.A design of HCB based on modular discrete MOVs shows improved static voltage balancing capability and effective voltage limitation during breaking, when compared with conventional series connected MOVs [136].Apart from the lower cost, lower residual voltage, lower volume, and diminished steep front effect are attainable.
While selecting semiconductor technologies to build HCBs, there is a broad range of solutions available in the market.Table 5 establishes a cost comparison between most common semiconductor technologies suitable for low and medium voltage HCBs.Similar voltage and current (continuous) ratings are considered for all devices.
It is interesting to note the ratio between price and current ratings.The price per Ampère of GTOs may be 15 times lower than the recent SiC MOSFETs.

VI. COMPARATIVE ANALYSIS
This section presents an overview of the different HCB's topologies presented in this article, in terms of their performance characteristics, ratings, cost, and size.
Table 6 compares the most relevant HCB topologies in terms of their advantages and disadvantages, operation speed, expected breaking capability, and corresponding target applications.
Fig. 13 provides a normalized evaluation of the costs associated to the HCB technologies listed in Table 6.Cost distribution is made by components category.
Overall, the mechanical switches and power semiconductors usually represent the most expensive components of the  HCB.Within the category of power semiconductors, fully controlled semiconductors (IGBTs, IGCTs) and particularly WBG technologies (SiC, GaN) increase the cost the most.As far as mechanical switches are concern, the adoption of actuating mechanisms that prioritize very high switching speed, such as those based on electromagnetic repulsion like Thomson coil, will reflect in an increase in the cost of the HCB; the use of superconducting materials in the mechanical switch has an even greater impact in costs.Despite their contribution to increase the volume, passive components have a reduced or even minimal impact in the cost of the HCB, unless very high ratings components are considered.
Along with the cost evaluation, the evaluation of the volume is an important performance metric.Information about the estimated volume of each HCB topology is presented in Fig. 14, which establishes a comparison between HCB topologies with regards to volume.At the top of each bar, the total volume (in dm 3 ) of each topology is provided.
From the results shown in Fig. 14, it is stated that there are quite distinctive trends with regards to volume breakdown by components category.While semiconductors occupy the majority of volume in topologies dominated by power electronics, mechanical switches represent important shares of volume occupancy in topologies involving the adoption of passive components.Semiconductors and mechanical switches are the main contributors for space occupancy, but snubber circuits are also important contributors.
It is important to recall that the volume comparison is not normalized to the voltage and current ratings of each HCB topology.

VII. CHALLENGES AND FUTURE TRENDS
Dc CBs are more complex to design than ac equivalents, due to the absence of natural occurring zero-current crossing points, in which the current can be interrupted.To solve this issue, the current must be forced to zero using different mechanisms and additional components, which translate in increased costs [133].
The widespread use of distributed energy resources (DERs) and dc loads are pointed out as the main driving factors of the DCCBs market [160].From the point of view of energy generation in DERs, DCCBs are fundamental to isolate faulty sections in wind energy [22] and photovoltaic [161] applications, for example.From the point of view of energy storage in DERs, DCCBs are important to protect uninterrupted power supplies [162] and energy storage systems [163].
HCBs combine the best features of MCBs and SSCBs to overcome the drawbacks of previous generations of DCCBs: the lack of operation speed, significant on-state losses, short lifetime, and the need for bulky cooling systems.While HCBs reveal potential to overcome all aforementioned drawbacks, it is also known that HCBs face a couple of challenges, like the relevant manufacturing costs, the reduced lifetime and efficiency, the un-optimized commutation process, the lack of design guidelines and standards, and the difficulty in operating at high speeds, to name a few.
Despite the relevant research work carried out so far and the effort put into manufacturing HCBs, the cost of implementation of a single HCB is still expensive, which results in poor economic value.This might reflect into lack of interest from the points of view of engineering research and practical implementation.Hence, future works should focus even more on ensuring the economic feasibility of HCBs.Such goal can be accomplished, for instance, through the adoption of cheaper, less noble components or the reduction on the number of components.
On the other hand, it is clear that the protecting capabilities and features of HCB should also be further promoted.Faster fault interruption, lower on-state losses, and enhanced thermal management capability are some of the key features that remain underdeveloped within the context of HCBs.To accomplish such goals, the adoption of WBG semiconductor devices opens new possibilities to future HCBs, given their favorable characteristics: lower power losses, higher voltage ratings (related to higher breakdown fields) and temperature capabilities, and extended avalanche breakdown capability, enabling the reduction of the number and size of energy absorbing elements.The fault interruption speed of HCBs may also benefit from enhanced knowledge about the arc behavior.Further investigations on this regard should help optimizing the switching procedure and the mechanical breaker performance.
Given the pivotal relevance that HCBs play in the effective protection of dc systems, it is also important that HCBs themselves show high robustness and reliability levels.Indeed, HCBs heavily rely on sensitive components which may fail as well, putting at risk the effective operation of the HCB.Apart mechanical switches, studies reported that power semiconductors [164] and MOVs [165], [166] bring relevant reliability issues to their applications.At the same time, these components play critical roles inside HCBs, meaning that any failure on them fully compromises the protection capability of the DCCB.On that basis, the development of fault diagnostics and fault tolerance techniques specifically directed to the HCB building components reveals as a promising research topic as well.
Assuring low levels of Electromagnetic Interference is also a crucial aspect to consider when designing high voltage HCBs.Research showed that the magnetic field around the HCB registered a peak value of 59 kA/m near the adjacent secondary equipment [167].Possible mitigation measures, which worth further research attention, include the improvement of the HCB shielding enclosure or the placement of the secondary HCB equipment (gate driver, sensors, etc.) far apart from the remaining HCB hardware.
Research aiming developments on HCBs mostly focus on their breaking capability, neglecting the fault recovery phase.If not handled properly, the reclosing operation of HCBs may introduce severe secondary impact on the dc grid [168].Hence, it is relevant to ensure integration of smooth reclosing characteristics into HCBs.
Apart from the adoption of WBG devices, future trends point toward the development of HCBs with enhanced breaking capability, capable of meeting the requirements of HVDC energy systems.Within such context, high operating voltage and high breaking current capability are a must.

VIII. CONCLUSION
This article described the solutions for HCBs available in the literature, making an evaluation of their performance.The main topics discussed in this article focused on: r origin of HCBs; r basic constitution and operation principles of HCBs; r the promising advantages of WBG semiconductors, especially attractive to more demanding applications, where features like high-speed and high-current handling capability are important; r alternative solutions to varistors, with the intent to reduce costs and improve reliability; r approaches directed to the overall reduction in cost and volume of an HCB.
It is clear that HCBs represent a proper solution for the protection of dc power systems, given their speed and low on-state resistance.Further investigation leading toward cost reduction and efficiency increase needs to be carried out, to consolidate the benefits that HCBs may play on dc systems.

FIGURE 1 .
FIGURE 1. Timeline of some remarkable advances in the field of HCBs in the past decade.

FIGURE 2 .
FIGURE 2. Basic structure of a conventional HCB.

FIGURE 3 .
FIGURE 3. Current waveforms for different HCB components, observed during current interruption.

FIGURE 6 .
FIGURE 6. Classification of the mechanical switches for HCBs.

FIGURE 7 .
FIGURE 7. Influence of a current limiting reactor on the current interruption process.

FIGURE 8 .
FIGURE 8. Structure of the self-adaptive FCL applied to an HCB [125].

FIGURE 9 .
FIGURE 9. General structure of the HCB proposed in[138], which replaces the varistor by a capacitor and diode for power dissipation purposes.

FIGURE 10 .
FIGURE 10.Modified HCB, integrating an additional resistive circuit (in red) to discharge the capacitor after fault clearance [138].

FIGURE 12 .
FIGURE 12. Constitution of the HCB integrated with an MMC, as presented in [149].The need for main breaker and MOVs is eliminated.

FIGURE 13 .
FIGURE 13.Cost distribution by components category, for the main topologies of HCBs.

FIGURE 14 .
FIGURE 14. Volume distribution by components category and total estimated volume, for the main topologies of HCBs.