Electronics Guide

The Reactive Limit on Alternating Current

An alternating-current line is a distributed capacitance as well as a distributed inductance. Charging current flows into that capacitance whether or not any load is connected, and it consumes conductor ampacity that the useful current would otherwise use. On overhead lines the effect is modest and can be managed with shunt reactors. In a cable it is severe, because a cable places its conductor a few centimeters from a grounded screen and therefore has roughly twenty to thirty times the capacitance per kilometer of an overhead line of the same voltage. Beyond a length on the order of fifty to one hundred kilometers, a high-voltage alternating-current cable spends most of its rating charging itself, and compensating the charging current requires reactors at intervals that a submarine route cannot provide.

Direct current has no charging current in the steady state. A direct-current cable is charged once at energization and thereafter carries only load current. This is the reason that essentially every long submarine transmission link built in the past three decades is a direct-current link, and it is why the export of power from distant offshore wind farms has become an HVDC problem rather than a cable-engineering problem.

Break-Even Distance and Cost Structure

The economics of HVDC follow a simple structure. Converter stations are expensive and their cost is nearly independent of distance. Direct-current lines and cables are cheaper per kilometer than their alternating-current equivalents, because a bipolar direct-current line carries the same power on two conductors that an alternating-current line carries on three, insulation is stressed by a steady voltage rather than a peak, and losses are lower for the same conductor cross section. Total cost as a function of distance therefore starts higher for direct current and rises more slowly, and the two curves cross.

For overhead lines the crossover falls in the range of roughly five hundred to eight hundred kilometers, depending on terrain, right-of-way cost, and the loss valuation used in the study. For submarine and long underground cables the crossover is far shorter, commonly quoted between fifty and one hundred kilometers, because the alternating-current alternative runs into the charging-current limit rather than merely becoming expensive. Below the crossover, alternating current normally wins on cost. Above it, direct current wins, and the margin widens with every additional kilometer.

Controllability and Asynchronous Interconnection

Distance is not the only reason to build an HVDC link. Power flow on an alternating-current line is set by the network, not by the operator: it follows the phase angle difference across the line and the impedances of every parallel path. Power flow on a direct-current link is set by the converter controls and can be ordered to any value within the rating, reversed, ramped at a chosen rate, or modulated to damp an inter-area oscillation. A direct-current link also does not transmit frequency, so it can join two networks that are not synchronized and cannot be, and it does not propagate a cascading disturbance from one side to the other.

Back-to-back stations exploit exactly this property with no transmission line at all. The three major interconnections of the contiguous United States, the Eastern Interconnection, the Western Interconnection, and the Texas system, exchange power through back-to-back HVDC stations because they run asynchronously. Japan uses frequency converter stations for the same reason, joining its 50 Hz and 60 Hz systems. A direct-current link also contributes no short-circuit current to the receiving network beyond the converter rating, which can be an advantage where switchgear fault ratings are already exhausted.

The Twelve-Pulse Bridge

The line-commutated converter, abbreviated LCC and sometimes called the current-source converter, is a thyristor bridge. Its basic unit is the six-pulse Graetz bridge, and practical stations use two such bridges in series, fed by transformers with a star and a delta secondary winding so that their alternating-current voltages differ by thirty degrees. The resulting twelve-pulse group cancels the fifth and seventh harmonics on the alternating-current side and the sixth on the direct-current side, leaving characteristic harmonics at orders twelve times an integer plus or minus one on the alternating side and twelve times an integer on the direct side.

A thyristor conducts once gated and stops only when the external circuit drives its current to zero, which is why the topology is called line commutated: the alternating-current system voltage performs the turn-off. The direct current through the smoothing reactor is held nearly constant, and the converter behaves as a current source toward the direct-current line and as a current sink toward the alternating-current network. Power direction reverses by reversing the polarity of the direct-current voltage, because the current can only flow one way through a thyristor. That constraint is the origin of several of the technology's characteristic limits.

Firing Angle, Extinction Angle, and Reactive Demand

The direct voltage produced by a line-commutated bridge varies with the cosine of the firing angle. A rectifier is operated at a small firing angle, typically around fifteen degrees, to keep the direct voltage high while retaining enough control margin. An inverter is operated at a large firing angle, near one hundred and forty to one hundred and fifty degrees, which is more usefully expressed as an extinction angle of roughly seventeen to eighteen degrees, the electrical angle remaining after the current transfers and before the valve voltage becomes forward again. That margin must exceed the thyristor turn-off time or the valve will fail to regain its blocking capability.

Because the current in each valve lags the commutating voltage by the firing angle plus roughly half the commutation overlap, a line-commutated converter always absorbs reactive power, in both rectification and inversion, and the absorption is substantial: on the order of fifty to sixty percent of the transmitted active power at full load. Converter stations therefore carry large banks of shunt capacitors and tuned harmonic filters, switched in steps as the load changes, and those banks occupy a large share of the station footprint. The filters serve double duty, absorbing the characteristic and non-characteristic harmonic currents while supplying reactive power at fundamental frequency.

Commutation Failure and System Strength

A line-commutated inverter depends on the alternating-current system to commutate. If the alternating voltage sags or shifts in phase during a nearby fault, the outgoing valve may not recover its blocking capability before its voltage turns forward, and the current returns to the valve that was supposed to stop conducting. This is a commutation failure. A single event is a brief direct-current interruption from which the controls recover within one or two cycles, but repeated failures can trip the link, and the loss of a multi-gigawatt link is a serious contingency for the receiving system.

The vulnerability scales with the weakness of the connected network, expressed as the short-circuit ratio between the network fault level at the converter bus and the converter rating. Line-commutated schemes are generally comfortable above a short-circuit ratio of about three and require careful engineering, synchronous condensers, or capacitor-commutated converter designs below that. A line-commutated converter also cannot energize a dead network, because it has nothing to commutate against, so it cannot black start and cannot on its own feed an isolated load such as an offshore collector system. Those two limitations define the boundary between the LCC and voltage-source domains more sharply than any cost comparison.

Thyristor Valves and Ultra-High Voltage

A valve at transmission voltage is a series string of hundreds of thyristors, each with its own grading network and its own voltage-sharing arrangement, suspended in a valve hall as a stack or hung from the ceiling to manage seismic and electrical clearances. Modern devices are typically light-triggered or electrically triggered wafers of one hundred to one hundred and fifty millimeters in diameter, mounted in press-pack housings so that a failed device shorts rather than opens. Redundant levels are built into every string, and the valve continues to operate with several failed devices until the next maintenance outage.

Line-commutated technology holds the records for scale. China's Changji to Guquan link, also known as the Zhundong to Wannan link, entered service in 2019 and carries 12,000 MW at plus and minus 1,100 kV over about 3,324 km, from Xinjiang in the northwest to southern Anhui. It remains the highest direct voltage, the largest capacity, and the longest transmission distance yet built. Series connection of two twelve-pulse groups per pole, each with its own converter transformers, allows such schemes to be operated at half pole voltage when one group is out of service. For point-to-point bulk transfer over land at this scale, no other technology is close on cost or losses.

From Two-Level Bridges to the MMC

A voltage-source converter, abbreviated VSC, uses devices that can be turned off by their gates, in practice insulated-gate bipolar transistors with antiparallel diodes, and maintains a direct voltage of fixed polarity across a capacitive direct-current bus. The first commercial installations, beginning with a pilot link in Sweden in 1997 and the Gotland Light scheme in 1999, used two-level and three-level bridges with series-connected devices switched together at a kilohertz or more. They worked, but the switching losses were high, roughly one and a half percent of throughput per station, and the output required substantial filtering.

The modular multilevel converter, proposed in the early 2000s and first deployed commercially on the Trans Bay Cable project in San Francisco in 2010 at 400 MW and plus and minus 200 kV, changed the economics. Each arm of the converter is a chain of independently controlled submodules, each containing a capacitor and a small switch cell. The arm synthesizes its voltage by inserting or bypassing submodules one at a time, so the output moves in steps of a few kilovolts and approximates a sine wave closely enough that alternating-current filters are often unnecessary. Each device switches only a few times per fundamental cycle, which cuts switching losses to a fraction of their two-level value, and station losses fall to roughly one percent.

Submodules, Balancing, and Arm Reactors

The half-bridge submodule, two switches and a capacitor, is the cheapest and most common cell. It can insert its capacitor or bypass it, so an arm can produce voltages from zero to the sum of its capacitor voltages, but never a negative voltage. The full-bridge submodule uses four switches and can produce plus, zero, or minus its capacitor voltage. That extra capability costs roughly a third more semiconductors and higher conduction losses, and it buys two things that matter: the ability to operate at reduced direct voltage, useful on overhead lines and during faults, and the ability to drive the direct-current fault current to zero without any breaker at all.

Control of an MMC is largely a bookkeeping problem. Several hundred submodule capacitors per arm must all stay near their nominal voltage while the arm current charges and discharges whichever ones are inserted. The controller sorts the capacitors by measured voltage every switching decision and chooses which to insert according to the direction of the arm current, so that the least charged are charged and the most charged are discharged. Arm reactors limit the circulating current that flows between phase legs because of the ripple in the arm capacitor voltages, and they limit the rate of rise of current into a direct-current fault, which buys time for protection to act.

Independent Power Control, Black Start, and Weak Grids

A voltage-source converter controls the magnitude and angle of its alternating-current terminal voltage, and therefore controls active and reactive power independently within its rating. It can absorb or supply reactive power at zero active power, functioning as a static compensator, and it can hold the voltage of the network it faces. Power reversal is achieved by reversing the direction of direct current while the voltage polarity stays fixed, which is essential for both multi-terminal operation and the use of extruded polymer cable.

Because it does not depend on the network to commutate, a voltage-source converter can operate into a passive network and can therefore black start a dead system and energize an offshore collector grid. It also tolerates very low short-circuit ratios, which is precisely the condition created when converter-interfaced generation displaces synchronous machines. Modern voltage-source HVDC terminals are increasingly specified with grid-forming control, so that the converter establishes voltage and frequency in its own right rather than following a phase-locked loop, and provides an inertial and fault-current response comparable in effect to a synchronous machine.

Choosing Between the Families

The comparison is now well settled in practice. Line-commutated converters offer the lowest losses, roughly seven-tenths of a percent per station, the highest voltage and current ratings, and the lowest cost per megawatt, and they are the default for long overhead bulk transfer between strong networks. Voltage-source converters cost more and lose somewhat more, but they need no reactive support, occupy a fraction of the footprint because the filter and capacitor yards largely disappear, connect to weak or passive networks, black start, reverse power without polarity reversal, and work with extruded cable. Anywhere a converter must sit on an offshore platform, feed a weak system, or participate in a multi-terminal scheme, the voltage-source converter is the only practical answer.

Back-to-Back and Monopole

A back-to-back station places rectifier and inverter in the same building with only a short direct-current busbar between them. Because there is no line, the direct voltage can be chosen low and the current high, which minimizes insulation cost, and the whole scheme exists to decouple two networks rather than to move power a distance.

An asymmetric monopole uses one high-voltage conductor and a return path at or near ground potential. The return may be the earth or the sea, using purpose-built electrodes several kilometers from the station, or a dedicated metallic conductor. Earth return is cheap and low loss, and it was the standard arrangement for early schemes, but the continuous direct current in the ground corrodes buried metalwork, deflects magnetic compasses near a submarine route, and raises regulatory and environmental objections. Most modern schemes therefore use metallic return, reserving ground return for temporary operation during a pole outage where permitted at all.

Symmetric Monopole

The symmetric monopole is the standard arrangement for voltage-source cable schemes. Two conductors operate at plus and minus half the total direct-current voltage, and the converter is arranged so that no continuous current flows to ground; the ground connection exists only to define the potential, usually through a high impedance at one end. Both cables are insulated for half the pole-to-pole voltage, which suits extruded cable well, and no electrodes are required. The disadvantage is that a fault on either cable takes the entire link out of service, because there is no independent second pole.

Bipole

A bipole is two complete monopoles of opposite polarity sharing a neutral. In balanced operation the neutral carries only the small difference between the pole currents, so the return path handles almost nothing. The decisive advantage is redundancy: if one pole is lost, the other continues at half capacity, using the metallic return or, where permitted, ground return to close the circuit. For that reason every large land-based bulk transfer scheme is a bipole, and reliability studies treat the loss of a single pole rather than the loss of the whole link as the design contingency. The 2 GW offshore connection systems now being built in the North Sea are bipoles as well, with a bundle comprising a positive pole, a negative pole, a metallic return, and a fiber-optic cable.

Voltage Levels in Service

Typical ratings cluster around a few standard levels. Voltage-source cable schemes were built at plus and minus 150 kV, then plus and minus 320 kV, which remains the most common level for offshore connections in the several-hundred-megawatt class, and are now moving to plus and minus 525 kV as extruded cable qualification has advanced. Line-commutated schemes use plus and minus 500 kV widely, plus and minus 600 kV for some long submarine and overhead links, and plus and minus 800 kV for Chinese and Indian ultra-high-voltage projects, with the single plus and minus 1,100 kV scheme at the top of the range.

Converter Transformers

The converter transformer is the most specialized transformer in the power system. Its valve-side winding is stressed by a direct voltage as well as an alternating one, so the insulation must be designed for a combined stress that ordinary power transformers never see and that redistributes slowly according to resistivity rather than permittivity. It carries heavy harmonic current, which increases eddy and stray losses and demands careful thermal design. In line-commutated stations it also provides the thirty-degree phase shift between the two bridges of a twelve-pulse group and an on-load tap changer with a wide range, used to keep the firing and extinction angles inside their control windows as the alternating voltage varies. Units are often built as single-phase transformers simply because a three-phase unit at these ratings cannot be transported.

Reactors, Filters, and Switchyards

A line-commutated station places a smoothing reactor, typically several hundred millihenries, in series with each pole to keep the direct current continuous, limit the rate of rise of fault current, and detune the direct-current circuit from resonances. Direct-current filters tuned to the twelfth and twenty-fourth harmonics limit the interference current on the line, and the alternating-current filter and capacitor yard supplies reactive power and absorbs alternating-side harmonics. A modular multilevel station replaces most of this with arm reactors and, frequently, nothing else, which is why its footprint is so much smaller.

Both types include a direct-current switchyard with disconnectors, neutral bus arrangements, and metal-oxide surge arresters at every significant node. Arrester coordination is a large part of HVDC insulation design, because the direct-current side has no natural current zero to help clear anything and because the energy absorbed during a fault or a valve misoperation must be shared among arresters in a predictable way.

Cooling and Auxiliaries

Valves are cooled by deionized water circulated through the heat sinks at high potential, with the coolant conductivity held very low by ion exchange resin so that the water column does not become a leakage path. The heat is rejected to outdoor coolers, and the cooling plant is fully redundant because losing it trips the link. Valve halls are shielded enclosures with controlled humidity, since condensation on a valve stack at hundreds of kilovolts is intolerable, and the shielding also confines the switching transients that would otherwise interfere with control and communication equipment.

Control and Protection Hierarchy

HVDC control is organized in layers. At the top, a bipole or master controller sets the power order and coordinates the stations over a telecommunication link. Below it, pole control implements the current or voltage regulation, and valve control converts the resulting firing orders into gate pulses through fiber-optic links that cross the insulation boundary. In a line-commutated link the rectifier normally regulates direct current and the inverter regulates extinction angle, with a current margin between the two orders so that only one station controls current at a time and control transfers automatically if the direct voltage collapses. A voltage-dependent current order limiter reduces the current order when the direct voltage falls, which helps the link recover from an alternating-current fault without repeated commutation failures.

In a voltage-source link the same layering applies with different variables: one terminal regulates direct-current voltage, acting as the slack node for the direct-current circuit, while the others regulate active power, and each terminal independently regulates reactive power or alternating voltage. Protection functions run alongside, watching for direct-current line faults, valve short circuits, submodule failures, converter transformer differential current, and electrode line problems, and they must act in milliseconds because the direct-current circuit has very little natural damping.

Overhead Lines

A direct-current overhead line carries two conductors for a bipole instead of the three of an alternating-current circuit, and it is stressed by a constant voltage rather than by a peak that is the square root of two times the nominal value. Both effects shrink the tower and the right-of-way for a given transmitted power. The compensating difficulties are direct-current corona, which behaves differently from alternating-current corona because space charge accumulates in the air gap and because the ion current reaches ground, and pollution performance of insulators, which is worse under direct voltage because a steady field attracts contaminants electrostatically. Insulator creepage distances are therefore longer than alternating-current practice would suggest, and composite insulators are widely used.

Mass-Impregnated and Extruded Cables

Two insulation systems dominate. Mass-impregnated cable, paper insulation saturated with a high-viscosity compound, has been used since the earliest links, tolerates polarity reversal without difficulty, and has been built for voltages beyond plus and minus 500 kV. Its variant with polypropylene laminated paper extends the temperature and voltage range further. Extruded cross-linked polyethylene cable is cheaper to make, lighter, easier to joint, and can run hotter, but space charge accumulation in the polymer made it unsuitable for the polarity reversals of line-commutated schemes. It became practical for HVDC only alongside voltage-source converters, which never reverse polarity, and qualification has now advanced from plus and minus 320 kV through plus and minus 525 kV and beyond.

The longest submarine links illustrate the range. NorNed, between Norway and the Netherlands, entered service in 2008 with 580 km of mass-impregnated cable carrying 700 MW at plus and minus 450 kV. NordLink, between Norway and Germany, began commercial operation in 2021 over 623 km, carrying 1,400 MW at plus and minus 525 kV on mass-impregnated cable. North Sea Link, between Norway and Great Britain, followed in the same year at 720 km and 1,400 MW. Viking Link, between Great Britain and Denmark, was energized at the end of 2023 over a combined subsea and underground route of roughly 765 km; it is rated 1,400 MW but operated under an 800 MW cap at first, because of constraints in the Danish network rather than in the link itself. Each of these is a point-to-point trade route between two markets, and each would be physically impossible in alternating current.

Why Three Terminals Are Harder Than Two

A two-terminal link has a simple control division: one end sets the current, the other sets the voltage. Add a third terminal and the direct-current circuit becomes a network whose node voltages and branch currents must be coordinated, with the additional problem that the sum of injections must balance at every instant because a direct-current circuit stores very little energy. Line-commutated multi-terminal schemes are further constrained because reversing the power at one terminal requires reversing the polarity of the whole system, so a tapped terminal must be equipped with mechanical switches to swap its connections.

Voltage-source converters remove that constraint, since each terminal reverses power by reversing its own current. Coordination is usually achieved with direct-current voltage droop, in which several terminals adjust their power in proportion to the deviation of the direct-current voltage from its reference, so that the loss of any single terminal is absorbed by the others without waiting for a communication channel. The scheme is the direct-current analogue of frequency droop in an alternating-current system, and it serves the same purpose: a fast, local, communication-free primary response, backed by a slower secondary layer that restores the voltage to its nominal value and redistributes the power economically.

Schemes in Service

Multi-terminal HVDC remained rare for decades. The Sardinia to Corsica to Italy link, which added a small tap at Corsica, and the Hydro-Québec to New England scheme commissioned in 1990 at plus and minus 450 kV and 2,000 MW, were the principal line-commutated examples, and the latter has usually operated with fewer terminals in service than it was built with.

The voltage-source era changed this. China built a three-terminal scheme at Nan'ao in 2013 and a five-terminal scheme at Zhoushan in 2014, both modest in rating and explicitly demonstrative. The Zhangbei project, in service since June 2020, is a four-terminal plus and minus 500 kV network: converter stations at Zhangbei and Kangbao collect wind and solar generation at 3,000 MW and 1,500 MW, Fengning connects a pumped-storage plant at 1,500 MW, and Beijing takes 3,000 MW at the load center. It is the first scheme designed from the outset as a direct-current grid with selective protection rather than as a point-to-point link with a tap. The Kunliulong project, named for its three converter stations and often called Wudongde after the hydroelectric plant whose output it evacuates, was commissioned in December 2020. It is a hybrid three-terminal ultra-high-voltage scheme, about 1,452 km long at plus and minus 800 kV, in which a line-commutated rectifier of 8,000 MW at Kunbei, north of Kunming, feeds two voltage-source inverter terminals, Longmen at 5,000 MW and Liuzhou at 3,000 MW. The arrangement combines the low losses of a line-commutated rectifier with the commutation-failure immunity of voltage-source inverters at the receiving end, which matters because both inverters land in the dense networks of Guangdong and Guangxi.

Interoperability

The obstacle to a meshed offshore direct-current grid in Europe is less technical than commercial. Converter control algorithms are proprietary, and terminals from different suppliers have never been required to share a direct-current circuit and cooperate through it. European transmission operators and manufacturers have accordingly pursued interoperability work aimed at defining the functional interfaces, control interactions, and protection coordination that would let multi-vendor terminals form one network, including the Horizon Europe InterOPERA project, launched in January 2023 to specify those interfaces and demonstrate them in simulation and in hardware before the first multi-vendor scheme is committed. Standardized converter ratings, such as the 2 GW offshore systems described below, are part of the same effort: a common building block is a precondition for a grid rather than a collection of point-to-point links.

Why a DC Fault Is Different

An alternating-current circuit breaker interrupts at a natural current zero, which arrives at worst every ten milliseconds. A direct current has no zero, so a direct-current breaker must create one, and it must do so quickly, because the fault current in a direct-current circuit is limited only by resistance and by the modest inductance of the reactors and the line. In a half-bridge modular multilevel converter the problem is worse than it appears: even with every device gated off, the antiparallel diodes form an uncontrolled three-phase rectifier that feeds the fault from the alternating-current side, so blocking the converter does not stop the current.

Two-terminal links live with this. A direct-current line fault is cleared by tripping the alternating-current breakers at both ends, waiting for the arc to deionize, and restarting the link; overhead-line schemes perform this sequence automatically as a direct-current line reclosure, and it takes several hundred milliseconds. That is acceptable when the whole link is lost anyway. It is not acceptable in a multi-terminal network, where a fault on one branch must be cleared without shutting down every terminal.

Fault-Blocking Converters

One answer avoids breakers. A converter built from full-bridge submodules, or a hybrid mixture of full-bridge and half-bridge cells, can generate a counter-voltage that opposes the alternating-current source and drives the direct-current fault current to zero within a few milliseconds. Fast mechanical disconnectors, which need only to open at zero current, then isolate the faulted branch, and the converters restore the voltage. The cost is more semiconductors and higher conduction losses in normal operation, paid continuously in exchange for a capability used rarely. Full-bridge capability also allows a converter to keep operating at reduced direct voltage, which is valuable on overhead lines where temporary faults are common.

Mechanical, Solid-State, and Hybrid Breakers

The other answer is a genuine direct-current breaker, and three architectures compete. A passive or active resonance breaker uses a conventional interrupter in parallel with a resonant branch that superimposes an oscillation on the arc current to force a zero; it is inexpensive and has low conduction losses but is slow, taking tens of milliseconds. A solid-state breaker uses semiconductor switches in the main current path and interrupts in microseconds, but it dissipates conduction losses continuously, which at gigawatt scale is unacceptable for a device that is normally closed.

The hybrid breaker resolves the trade-off. A fast mechanical disconnector with a small commutating semiconductor branch carries the load current at low loss. On a fault, the auxiliary branch commutates the current into a parallel main semiconductor stack, the mechanical contacts part at nearly zero voltage, and the main stack then turns off, driving the current into surge arresters that absorb the inductive energy and force the current to zero. Devices of this type were built for the Zhangbei four-terminal grid at a rated direct voltage of plus and minus 535 kV, specified to interrupt 25 kA within three milliseconds; reported type testing cleared a 25 kA fault in a little under 2.9 ms. Those figures indicate the performance envelope a direct-current grid demands.

Protection Selectivity

The breaker is only half the problem. Protection must decide which branch is faulted within one or two milliseconds, far too fast for the current-differential comparison used in alternating-current line protection, which needs a communication round trip. Direct-current grid protection therefore relies on single-ended methods that examine the traveling wave arriving at the relay: the rate of change of voltage and current, the polarity and shape of the wavefront, and the effect of the series reactor that deliberately terminates each line and creates a measurable discontinuity between an internal and an external fault. Setting these schemes so that they are both fast enough and secure against maloperation is one of the harder open problems in direct-current grid engineering.

The Case for HVDC Offshore

An offshore wind farm collects power at 66 kV within its array, steps up at an offshore substation, and exports to shore. Alternating-current export at 220 kV is the cheaper choice for distances up to roughly eighty kilometers and moderate capacity. Beyond that, cable charging current consumes so much of the rating that alternating-current export becomes uneconomic, and since offshore development has moved steadily into deeper water further from land, HVDC export has become the norm for large projects. A voltage-source converter is mandatory in this role, because the offshore terminal must energize and control a network that contains no synchronous generation whatever.

The Offshore Platform

Placing a converter station on a platform in the North Sea imposes constraints that no onshore station faces. Weight and volume drive the cost of the jacket and the installation vessel, so the compact footprint of a modular multilevel converter is not a convenience but a requirement. Equipment must tolerate salt, motion, and limited access; the station is normally unmanned and remotely operated; and everything must be commissioned to a standard that makes intervention rare, since a repair may wait weeks for weather. The German North Sea program that began with BorWin1, a 400 MW link at plus and minus 150 kV connecting a wind cluster roughly 125 km offshore, established the pattern around 2010, and successive projects converged on ratings between roughly 800 and 900 MW at plus and minus 320 kV. British offshore development followed. The Dogger Bank project introduced HVDC export to United Kingdom offshore wind in three phases of 1.2 GW each, all at plus and minus 320 kV; the first phase delivered power in October 2023, although commissioning to full output extended well beyond that date, a reminder that a first-of-a-kind rating carries schedule risk of its own.

The 2 GW Standard

The current step change is the standardized 2 GW connection system adopted by TenneT for the Dutch and German North Sea, and taken up in similar form elsewhere. Each system is a bipole at plus and minus 525 kV using extruded cable, with a four-cable bundle comprising the two poles, a metallic return, and a fiber-optic cable, and a single platform design repeated across a series of projects. The first of them, the IJmuiden Ver connections in Dutch waters, are targeted for the end of this decade, with the rest of the programme following into the early 2030s. Standardization is the point: repeating one design across many projects compresses engineering, manufacturing, and installation schedules, and it establishes the common building block that a future meshed offshore grid would require. The 66 kV array voltage, the 2 GW converter, and the 525 kV bipole together define the present state of the art in offshore export.

Hybrid Assets and Alternative Architectures

Two directions extend the pattern. A hybrid interconnector, or multi-purpose interconnector, connects an offshore wind farm to two countries at once, so that the same converter and cable assets serve both export and trade, improving utilization at the cost of a considerably more complicated regulatory and market treatment. Separately, the diode rectifier unit concept proposes replacing the offshore voltage-source converter with an uncontrolled rectifier, which would be far lighter, cheaper, and more robust, shifting all control to the turbines and the onshore station. The concept has been studied extensively but has not been built at transmission scale, and the obstacle is a real one: it requires grid-forming turbine converters able to establish and regulate the offshore alternating-current network with no help from the export station, together with a separate arrangement for reactive support, for start-up, and for any power flow back to the platform.

Loss Accounting

Total link losses are the sum of two converter stations and the line or cable. Station losses at full load are approximately seven-tenths of a percent for a line-commutated station and approximately one percent for a modular multilevel station, with the older two-level voltage-source designs closer to one and a half percent. Line and cable losses depend on conductor cross section and length and are lower than the alternating-current equivalent for the same power, since there is no charging current, no skin effect, and no reactive component in the conductor current. In a loss study the fixed converter penalty must be weighed against the per-kilometer saving, which is another way of stating the break-even distance.

Availability and Redundancy

Mature HVDC schemes achieve high energy availability, and CIGRE has published a biennial survey of HVDC reliability performance for decades that allows operators to benchmark against the fleet. Redundancy is designed in at several levels: redundant thyristors and submodules in every valve, redundant cooling and control systems, and, in a bipole, an entire second pole. Scheduled maintenance is typically annual and concentrated on valve cooling, filters, transformers, and control-system upgrades. Cable failures dominate the outage duration statistics of submarine schemes, because locating and repairing a fault on the seabed takes weeks rather than hours, and spare cable lengths and repair contracts are part of the project rather than an afterthought.

Standards Framework

The IEC 60700 series governs thyristor valve testing and IEC 62501 the testing of voltage-source converter valves. IEC 60919 addresses the performance of line-commutated systems and IEC 62747 the terminology and performance specification of voltage-source systems, while IEC 61975 covers system tests. IEEE and CIGRE supply the complementary engineering literature, with CIGRE Study Committee B4 producing the technical brochures that in practice define much of the state of the art before it reaches a standard. In Europe, Commission Regulation (EU) 2016/1447 establishes a network code on requirements for grid connection of HVDC systems and direct-current-connected power park modules, giving system operators a common basis for specifying what a converter terminal must do.