Power Systems

HVDC Converter

Thyristor "valves" convert AC to ±800 kV DC for losses-free 2,000-km transmission

A high-voltage direct-current (HVDC) converter station converts three-phase AC into ±800 kV DC for long-distance transmission, then back to AC at the destination. Two architectures dominate: line-commutated converter (LCC) using ~1 kHz-switched thyristor "valves" stacked in 12-pulse bridges (China's Xinjiang–Anhui ±1100 kV, 12 GW, 3,293 km — record-holder); and voltage-source converter (VSC) using IGBTs at ~2 kHz with full active control. HVDC eliminates skin effect, dielectric loss, and reactive power demand — making 2,000 km cables and undersea links practical. Conversion efficiency reaches 99.0% per station; line losses fall ~30% below equivalent AC.

  • Voltage range±100 kV to ±1100 kV
  • Power range100 MW to 12 GW
  • TopologiesLCC (thyristor) vs VSC (IGBT)
  • Per-station efficiency~99.0% (LCC), ~98.5% (VSC)
  • Subsea recordNorthConnect (660 km), NeuConnect (725 km)
  • Record voltage±1100 kV (China Xinjiang-Anhui)

Interactive visualization

Press play, or step through manually. The visualization is yours to drive — try it before reading on.

Open visualization fullscreen ↗

Watch the 60-second explainer

A condensed visual walkthrough — narrated, captioned, under a minute.

Why HVDC matters

  • Renewable corridors. Wind in Inner Mongolia, hydro in Yunnan, solar in the Atacama and Sahara are all hundreds to thousands of kilometers from load centers. HVDC is the only way to wheel that power without burning ~10 percent in line losses.
  • Offshore wind. Every offshore wind farm beyond ~80 km from shore connects via VSC HVDC. Germany's North Sea cluster (BorWin, DolWin, HelWin) injects ~10 GW into the European grid through 320 to 525 kV cable bundles.
  • Solar mega-farms. The proposed SunZia line in the U.S. Southwest will carry 3 GW of New Mexico wind/solar to Arizona load centers via ±525 kV HVDC over 880 km. The Australian Sun Cable plans ±525 kV for 4,300 km Darwin-to-Singapore.
  • Grid interconnection. Asynchronous AC grids (Quebec/Ontario, Texas/Eastern Interconnect, Japan east 50 Hz / west 60 Hz) connect only through HVDC back-to-back ties. The North Sea Wind Power Hub is designed as a multi-terminal HVDC mesh linking UK, Norway, Germany, and Denmark.
  • Load balancing across timezones. Morning peak in Beijing happens 3 hours after morning peak in Mumbai. A pan-Asian HVDC backbone could shift renewables across solar-noon offsets, smoothing demand without storage.
  • City-center cable feed. ±320 kV VSC links can feed dense urban load via underground cable without right-of-way for new AC corridors. London's London Power Tunnels and the Manhattan-area Champlain Hudson are examples.
  • Black-start and grid stability. VSC HVDC can re-energize a collapsed AC system from a remote source, faster than restarting thermal generation. It can also damp inter-area oscillations by modulating power flow at sub-cycle rates.

Common misconceptions

  • "AC always loses to DC over distance." Only above the break-even distance. For overhead lines that crossover is ~600 km; below that, AC wins because two converter stations cost more than the AC line losses save. For subsea cables the crossover drops to ~50 km because cable capacitance murders AC at any length.
  • "Thyristors are obsolete." They are not. LCC is still cheaper per MW than VSC, handles ~12 GW per station versus ~3 GW for VSC, and has 30+ years of field-proven reliability. Every modern bulk-transfer ±800 kV and ±1100 kV project (Xinjiang-Anhui, Changji-Guquan, Pacific DC Intertie upgrade) uses thyristors. VSC is reserved for cases requiring active control or weak-grid connection.
  • "VSC is just better." VSC has black-start capability, four-quadrant reactive control, no commutation failures, and no large filter banks. But it costs 30 to 50 percent more per MW, has ~50 percent higher conversion losses (1.5 percent per station versus 0.8 percent for LCC), and is power-limited by IGBT current ratings (~3 kA per device). For 12 GW into a stiff grid, LCC wins.
  • "HVDC means no losses." Each conversion station consumes ~1.0 percent (LCC) or ~1.5 percent (VSC) of the transmitted power. A point-to-point link has two stations, so 2 to 3 percent is lost in conversion alone. Line losses (~3 percent per 1000 km at full load) add to that. The savings come from the comparison: equivalent AC over the same path would lose 5 to 10 percent in skin effect, dielectric, and corona losses combined with reactive power demand.
  • "DC is safer than AC at the same voltage." The opposite for arc clearing. AC current naturally crosses zero 100 or 120 times per second, allowing standard circuit breakers to extinguish arcs. DC has no zero crossings, so DC circuit breakers (hybrid mechanical-semiconductor designs by ABB and Hitachi Energy) must force a zero crossing within 5 to 10 ms. Multi-terminal HVDC mesh networks have been gated for years on the maturity of HVDC breakers.
  • "HVDC corridors are always overhead lines." Many are cable. Overhead conductors are ~10 times cheaper per km, but rights-of-way through populated regions are politically blocked. The 2,500 MW SuedLink (Germany north-south) is entirely buried cable at ±525 kV — the largest land cable HVDC project in the world.
  • "HVDC is mature; nothing new is happening." Modular multilevel converters (MMC), introduced commercially around 2010, replaced two-level VSCs and made VSC HVDC viable above 320 kV by stacking hundreds of half-bridge submodules. Hybrid LCC+VSC stations, multi-terminal DC grids, and ultrahigh-voltage ±1100 kV bipolar systems are all post-2020 commercial deployments.

Frequently asked questions

Why convert AC to DC and back instead of just using AC?

Long AC lines carry three losses absent in DC. First, capacitive charging current: the line's distributed capacitance to ground demands reactive current proportional to length, and at ~600 km overhead or ~50 km undersea this charging current saturates the conductor before any real power flows. Second, skin effect: 50/60 Hz AC concentrates current in the outer ~1 cm of the conductor, raising effective resistance ~10 percent. Third, dielectric loss in cable insulation, dominant in subsea XLPE. DC has none of these — current uses the full cross-section, no reactive power, no dielectric heating. The two ~99 percent-efficient conversion stations cost about 2 percent total, paid back by line savings beyond a break-even distance.

What is a thyristor valve?

A thyristor is a four-layer P-N-P-N silicon switch (~6-inch wafer for high-power) that latches on when triggered and stops when current crosses zero. A "valve" is a stack of 50 to 200 series-connected thyristors in a single physical assembly, rated for ±800 kV by adding their breakdown voltages. Modern valves are suspended from porcelain insulators in the converter hall — a building the size of a hockey arena. Each thyristor switches at line frequency (50 or 60 Hz), so the valve effectively performs commutation at 1 kHz when 12 thyristors share a 12-pulse bridge cycle. Cooling is deionized water; a single ABB or Siemens valve handles 500 kV and 4000 A.

What's the difference between LCC and VSC?

LCC (line-commutated converter) uses thyristors that turn on by gate pulse but only turn off when AC current naturally crosses zero. It needs a stiff AC grid to provide that commutation voltage and consumes 50 to 60 percent reactive power that must be supplied by capacitor banks. Power flow reverses by reversing DC voltage polarity, requiring physical switching. VSC (voltage-source converter) uses IGBTs that switch on and off independently, modulating a sine wave from a DC bus by pulse-width modulation at 1 to 2 kHz. VSC can supply or absorb reactive power, black-start a dead grid, reverse power by reversing current direction (no polarity flip), and connect to weak or islanded systems. LCC is cheaper and handles more power per station — most ±800 kV bulk-transfer projects are LCC.

Why does undersea transmission need HVDC?

Subsea cables have ~5 to 10 times the per-km capacitance of overhead AC lines because the conductors are surrounded by polymer insulation in seawater (a near-conductor at 50 Ω·m). At 50 Hz the charging current of a 50 km AC subsea cable approaches the cable's rated current, leaving no margin for real power. Beyond 50 to 100 km AC subsea is impossible. DC has zero capacitive charging in steady state, so length is limited only by I²R conductor losses. The 660 km NorLink (Norway-Germany, 1.4 GW) and 725 km NeuConnect (UK-Germany, 1.4 GW) are both VSC HVDC. The Sun Cable proposal (Australia–Singapore, 4,300 km, ±525 kV) would carry desert solar across the Timor and Java Seas.

What does "12-pulse bridge" mean?

A 6-pulse bridge is a Graetz arrangement of 6 thyristors that rectifies three-phase AC into DC with a ripple frequency of 6 × 50 = 300 Hz (or 360 Hz at 60 Hz). A 12-pulse bridge stacks two 6-pulse bridges in series, fed by two transformers — one Y-Y, one Y-delta — that produce a 30-degree phase shift between their secondary voltages. The two bridges' outputs sum into 12 pulses per AC cycle (600 Hz at 50 Hz, 720 Hz at 60 Hz), cancelling the 5th and 7th harmonics on the AC side and the 6th harmonic on the DC side. Real ±800 kV stations stack four 12-pulse "poles" for redundancy and to split insulation stress; that's 96 thyristors per quadrant in series, ~400 in total per pole.

Can HVDC connect to weak grids?

LCC requires a short-circuit ratio (grid stiffness) of at least 2.5; below that, commutation failures cascade. This excludes islanded systems, small offshore wind farms feeding a remote shore, or remote oil platforms. VSC has no such requirement — it generates its own AC waveform from the DC bus and can even black-start a dead grid (energizing transformers and synchronous loads from zero). VSC is the architecture used for offshore wind connection in the North Sea (BorWin, DolWin, HelWin in German waters) and for embedded HVDC links in stressed AC corridors. The cost per MW is 30 to 50 percent higher than LCC, and the upper power limit per station is ~3 GW versus ~12 GW for LCC.