R4 was supposed to be the bridge.
Ågesta had shown that a natural uranium reactor could be built and run. Marviken was meant to prove that it could be scaled into a real power system — and, quietly, it still carried the earlier logic of the Swedish line: a fuel cycle that could, if ever decided, support weapons.
That logic ended in 1968, as the weapons program was abandoned. That single decision cut one of the hidden pillars under Marviken.
Before that, heavy water, natural uranium, on-load refueling — all of it had a dual justification:
- energy independence
- potential plutonium production
After 1968, only the first remained. And economically, it was already weakening.
But Marviken did not reset. It continued as designed. The structure was still the same:
- AB Atomenergi (state) designing and holding the concept
- ASEA building it
- Vattenfall expected to operate it
But now the motivations were no longer aligned with the design.
Technically, the reactor had already drifted.
What began as a pressure-tube heavy-water system — modular, refuelable, coherent in its own way — had been pushed into something more complicated to keep up with the rapidly advancing light-water technology. Boiling was added. Then superheating. And in the process, even the architecture shifted. The first stage ended up in a single pressure vessel, made possible by lowering the pressure to allow boiling in the core.
But downstream, the complexity remains:
boiling, dense, strongly cooled regions
superheating, sparse, weakly cooled regions
Different physics, coupled together.
At the same time, the reactor is directly tied to the turbine cycle. Steam goes out, condenses, and returns to a system that now contains heavy water. That makes the condenser leak-tightness crucial.
A leak is no longer just inefficiency. It is a loss of expensive heavy water.
So the plant is coupled in two directions:
inside — mismatched thermal regimes
outside — shared boundary with the turbine system
And now, after 1968, the one argument that justified carrying this complexity — strategic independence with latent weapons capability — is gone.
What remains is a technically intricate reactor trying to compete with a world that has simplified.
Light water reactors arrive as complete systems:
- one vessel
- one regime
- one supply chain
Marviken moves the other way
- more interfaces
- more coupled behavior
- more things that must all work at once
The superheating system becomes the point of no return. Too much invested to discard easily. Too little benefit to defend. Fixing it means redesign. Keeping it means accepting a system no longer aligned with its purpose.
By 1970, the decision is finally made.
The project is stopped. Marviken never runs.
Natural uranium. Direct cycle. Superheated steam. A combination so seductive that it’s hard for any nuclear engineer to keep calm and consider the neutronics.
But we must. There is no way around physics.
***
The Soviet RBMK and Sweden’s Marviken R4 were both brilliant reactor concepts. Both also developed serious safety problems. Yet R4 was canceled before fuel loading, while the RBMK became a large operating fleet.
Why?
Both promised to remove expensive limitations of conventional nuclear power. Excellent neutron economy reduced dependence on enriched uranium. Refueling during operation avoided regular shutdowns. Direct steam cycles eliminated steam generators. Each offered a route to an independent national reactor industry.
Their engineering solutions were quite different. The RBMK combined graphite moderation with boiling light water in individual pressure channels, allowing huge power output without manufacturing an enormous reactor pressure vessel.
R4 combined heavy water, natural circulation and a direct steam cycle, with provision for nuclear superheating. Remove the main circulation pumps, remove the steam generators, improve turbine efficiency, and change fuel while operating. From a process engineer’s perspective, there was plenty to admire.
The difficulty was making the complete plant as good as the concept. R4 accumulated complicated requirements for fuel handling, startup, shutdown, and protection. By 1969, testing and revised calculations had revealed problems with steam quality and core stability. Nuclear superheating was abandoned, and the fuel, safety systems, and shutdown arrangements needed further changes.
Fixing the plant would mean more delay, more expense and less of the economic advantage that had justified building it. Meanwhile, light-water reactors were becoming commercially established, including designs from Swedish industry itself.
In 1970, Sweden stopped.
The RBMK’s economic advantages remained attractive, but no equally effective challenge addressed its safety problems. A large positive void effect, difficult spatial reactor behavior, and a shutdown system that could initially add reactivity formed a very dangerous combination.
A positive void coefficient alone does not condemn a reactor. The complete design must reliably accommodate it. The original RBMK failed that test. Nor did the warning arrive only in April 1986. Earlier events, including Leningrad in 1975, had already exposed serious weaknesses. They did not produce adequate corrections or ensure that operators understood the hazards.
The different outcomes cannot be reduced to individual engineers' competence or conscience. The institutions surrounding them mattered.
Sweden could abandon Marviken while continuing its nuclear program. The Soviet system failed to force an adequate reconsideration of the RBMK despite accumulating evidence.
A brilliant concept deserves serious development. It also deserves a serious possibility of rejection.
Marviken is often remembered as a reactor project that failed. From a nuclear safety perspective, a completed reactor that never starts can also represent a decision that succeeded.