WHAT IT IS
RBMK is the Soviet graphite-moderated, light-water-cooled, pressure-tube power reactor. In Western lists it is an LWGR (light-water graphite reactor). Water boils inside the fuel channels and the steam goes straight to the turbines. There is no separate secondary circuit and no giant reactor pressure vessel around the whole core.
It is not a VVER. A VVER is a pressurized-water vessel reactor, like a Western PWR. An RBMK is closer in some ways to a boiling-water reactor and in other ways to a pressure-tube machine such as CANDU — except the moderator is graphite, not heavy water, and the coolant is ordinary water.
That pairing — graphite as the main neutron moderator, boiling light water as the coolant — exists in no other commercial power-reactor family. It is the design feature that made the plant buildable with ordinary industry, and the same feature that made Chernobyl possible.
HOW IT WORKS
Think of a huge stack of graphite bricks, about 7 m tall and 12 m across, sitting in a concrete vault. Vertical holes run through the stack. Down most of those holes go thin metal pressure tubes (technological channels). Inside each tube hang two fuel assemblies, one above the other.
Cold water is pumped up from the bottom of each fuel channel. In the core it heats and partly boils. A steam–water mix leaves the top of the channel and runs to large horizontal drums called steam separators. Steam is dried and sent to the turbines. Water falls back to the main circulating pumps and is sent into the channels again.
That loop is the multiple forced circulation circuit (KMPTs). An RBMK-1000 has two such loops, each serving half the core: separators, downcomers, four main pumps per side, a pressure header, distributing headers, and hundreds of individual feeders, one per channel.
Because each channel is its own little pressure boundary, you can isolate one tube and pull the fuel while the rest of the reactor keeps running. Online refuelling is a defining RBMK trait.
Graphite does the real neutron slowing. Water in the tubes also slows neutrons a bit and, more important, absorbs them. When water turns to steam it absorbs fewer neutrons. In the original lattice that meant more neutrons available for fission — a positive void coefficient. More steam, more power, more steam. At low power and with the control rods pulled far out, that feedback could run away.
THE MACHINE IN PIECES
Graphite stack. Thousands of square graphite blocks (about 250 × 250 mm, various heights) stacked into columns. Active core of an RBMK-1000: height about 7 m, diameter about 11.8 m. Total graphite on the order of 1,700 tonnes. About 5–6% of fission heat is deposited in the graphite. Operating graphite temperature is hundreds of degrees C (design talk around 650 °C). A thin steel jacket around the stack holds a helium–nitrogen mix.
Pressure tubes. Fuel-channel tube: about 80 mm inside diameter, 4 mm wall. The part in the core is Zr-2.5%Nb. Top and bottom sections are stainless steel. An RBMK-1000 has on the order of 1,660–1,700 fuel channels plus control and instrument channels, about 1,880 holes in the lattice at a 250 mm pitch. Each fuel channel is plugged at the top through the upper biological shield. That floor of plugs is the pyatachok.
Fuel. UO2 pellets in zirconium cladding. Eighteen rods around a central carrier make one assembly about 3.65 m long. Two assemblies stacked fill one channel. A full RBMK-1000 load is on the order of 190 tonnes of fuel. Original enrichment was about 1.8–2.0% U-235. After Chernobyl it went to 2.4% and later higher, sometimes with erbium as a burnable absorber.
Control. Boron-carbide rods in their own channels, plus short scram rods and, originally, a design where the rod’s graphite displacer entered the core before the absorber. That last trick could add a burst of reactivity at the start of a scram — the positive scram effect. After 1986 the rods and the rules for how many must stay in the core (operating reactivity margin) were changed.
Containment — or the lack of it. The core sits in a concrete vault with an upper biological shield. There is no large dry containment building of Western PWR style. Steam from a ruptured channel was meant to go to suppression pools or the reactor cavity relief system. The cavity can only swallow so many tube ruptures at once. A whole-core power excursion can lift the upper shield, open the core to the hall, and start a graphite fire. That is what happened at Chernobyl-4.
WHY THE SOVIETS BUILT IT
The 1954 Obninsk AM-1 reactor already used graphite plus water in channels. Beloyarsk AMB-100 and AMB-200 scaled the idea. In the mid-1960s the USSR wanted a 1,000 MWe-class plant that did not need a giant forged pressure vessel, could be assembled from ordinary piping and graphite at inland sites, could be refuelled on line, and could, in principle, make plutonium as well as electricity.
Nikolai Dollezhal’s NIKIET design bureau led the reactor. Leningrad (Sosnovy Bor) Unit 1 was the first RBMK-1000, construction from 1970, grid 1973. Kursk, Chernobyl, Smolensk followed. Ignalina in Lithuania got the uprated RBMK-1500. Three design generations appeared as Soviet safety rules tightened. First-generation plants (Leningrad 1–2, Kursk 1–2, Chernobyl 1–2) were the weakest. Smolensk-3, started after Chernobyl, is the last RBMK that ever went into service (1990).
CHERNOBYL, IN DESIGN TERMS
26 April 1986, Chernobyl-4, during a safety-related turbine coast-down test. Operators had the reactor at very low power, with almost all manual rods withdrawn, xenon poisoning in the fuel, and coolant so close to saturation that a small power rise made a lot of steam. In that state the void coefficient was strongly positive. Power began to climb. The scram (AZ-5) dropped rods whose graphite tips briefly added reactivity as they entered. Power spiked to many times design. Fuel shattered, coolant flashed, channels burst. A second explosion wrecked the core and the roof. Graphite burned for days.
The accident was not “an RBMK doing what every RBMK would do on a normal Tuesday.” It was an RBMK in a forbidden corner of its operating map, with a test that disabled protections, plus design defects that turned that corner into a disaster: positive void worth, slow and initially positive scram, no robust containment, and procedures that allowed the operating reactivity margin to collapse.
WHAT THEY CHANGED AFTERWARDS
On every running RBMK: more fixed absorbers in the core, higher minimum rod worth in the core, richer fuel (2.4% and up; later U-Er fuel), void coefficient cut sharply, redesigned control rods without the dangerous graphite-tip behaviour, better emergency core cooling, better cavity over-pressure relief, new process computers and trip logic, and bans on the low-power operating region that killed Unit 4. Pressure tubes were replaced as graphite swell and tube sag became life-limiting.
None of that turns an RBMK into a VVER. It made the remaining plants operable under modern Russian rules. It does not make a Chernobyl-type excursion the expected end state of a well-run unit in 2026.
RBMK VERSUS VVER AND WESTERN PLANTS
RBMK: graphite moderator, boiling water coolant, pressure tubes, online refuel, direct-cycle steam, huge core, low power density (~4 kW/litre), no full containment on the original design, historically positive void coefficient.
VVER: water moderator and coolant, one steel vessel, offline refuel, two-circuit PWR, compact core, high power density, containment on later models, negative or manageable void behaviour.
Western BWR: water is both moderator and coolant, so more steam usually means less moderation and a negative void coefficient. Graphite is absent. CANDU: pressure tubes and online refuel, but heavy-water moderator in a calandria, not graphite.
HISTORY IN BRIEF
1954 — Obninsk AM-1, the ancestor.
1964–66 — RBMK concept set.
1973 — Leningrad-1 on the grid.
1977 — Chernobyl-1 on the grid.
1983–86 — Chernobyl-4 starts; accident April 1986.
1986–90 — remaining builds finished or cancelled; Ignalina-2 (1987) and Smolensk-3 (1990) are the last.
2000s — Lithuania shuts Ignalina-1 (2004) and -2 (2009). Chernobyl-1 and -3 already shut; Unit 2 shut after a 1991 fire.
2018–24 — Russia begins retiring first-generation units: Leningrad-1 (2018), Leningrad-2 (2020), Kursk-1 (2021), Kursk-2 (2024).
Kursk-5 cancelled. Replacement at Leningrad and Kursk is VVER-1200 / VVER-TOI.
STATUS IN 2026
Only Russia still runs RBMKs. Seven RBMK-1000s are in service: Leningrad 3 and 4 (Unit 4 received a further licence extension in January 2026; both are due to give way to new VVERs), Kursk 3 and 4 (aimed toward late 2020s / about 2030), and Smolensk 1, 2, and 3 (Smolensk-3 has the longest advertised remaining life).
They still make a noticeable share of Russian nuclear electricity and, at Leningrad, some medical isotopes. They are not an export product and will not be built again. Bilibino’s small EGP-6s and the old Beloyarsk AMB units were related graphite-water cousins, not serial RBMK-1000s.
LIMITS THAT NEVER WENT AWAY
IN ONE SENTENCE
An RBMK is a huge graphite pile full of individually cooled boiling-water pressure tubes, built so the Soviet Union could make a thousand megawatts without a giant vessel — and the same layout that gave it online refuelling and cheap construction also gave it a positive void coefficient, a flawed scram, and the accident at Chernobyl.