Chernobyl, Fukushima, Three Mile Island: Could Any of That Happen With an SMR?
The three disasters that define nuclear fear were each caused by something specific — and SMR designers have spent decades engineering around exactly those things.
The word “nuclear” carries a lot of luggage. Say it in a crowded room and watch people’s faces. You’ll see Chernobyl’s burning graphite. You’ll see the HBO miniseries. You’ll see a vague but powerful dread that no amount of carbon-free electricity seems to fully displace. And that dread has a home address: three accidents spread across four decades that together wrote the story of nuclear risk in the public imagination.
Here’s the thing though. Chernobyl, Fukushima, and Three Mile Island were not random. They were not the nuclear equivalent of lightning strikes. Each one failed in specific, diagnosable ways, for specific, identifiable reasons. And the engineers now building small modular reactors have had all three case studies pinned to their whiteboards for years. The question worth asking is not “is nuclear dangerous” in the abstract, but something sharper: could the specific chain of events that caused each of those disasters actually happen inside an SMR? In most cases, the answer is a pretty firm no — with some important asterisks worth keeping honest about.
What actually broke at each accident site
Start with Three Mile Island in 1979. The reactor core melted partially because the cooling system failed — and then operators, misreading their instruments, actually turned off the emergency cooling water. It was a cascade of equipment malfunction and human error. No explosion, no mass casualties, no meaningful radiation release beyond the plant boundary. But it terrified the public and strangled the U.S. nuclear industry for a generation, which is a kind of damage in its own right.
Chernobyl in 1986 was a different beast entirely. The RBMK reactor design — a Soviet-era, graphite-moderated, water-cooled type that no Western country ever operated — had a positive void coefficient, meaning the reactor actually got more reactive as coolant boiled away. That’s a physics property so dangerous it would never pass modern regulatory review anywhere in the world. Add a reckless safety test run by undertrained staff with the emergency systems disabled, and the result was a runaway reaction, a steam explosion, and an open graphite fire that burned for days and sent radioactive particles across Europe. 🔥
Fukushima Daiichi in 2011 was different again. The reactors themselves shut down automatically and correctly when the earthquake hit. The problem arrived 40 minutes later, in the form of a tsunami that knocked out the diesel backup generators. Without power, the active cooling pumps stopped. Decay heat — the residual heat that keeps cooking even after a reactor shuts down — had nowhere to go. Three cores melted. The lesson, bluntly stated: when your only line of defense is electric pumps, and the electricity disappears, you have no line of defense.
Three accidents. Three very different root causes:
A design with dangerously unstable physics, run recklessly by people who didn’t fully understand it (Chernobyl)
A coolant system that failed, then got shut off by confused operators (Three Mile Island)
Passive decay heat with no backup cooling when power vanished (Fukushima)
The physics of SMR safety: boring by design
Here is where the engineering gets genuinely interesting. Modern SMR designers did not simply promise to run their plants more carefully than the Soviets did in 1986. They went back to first principles and tried to make accidents physically impossible, or at least to ensure the reactor’s own physics work against disaster rather than accelerating it.
The first change is negative temperature coefficient. Most modern reactor designs, including the light-water SMRs from NuScale and the gas-cooled designs from Kairos and X-energy, are engineered so that as temperature rises, the reaction automatically slows down. The physics self-regulates. This is the direct opposite of the RBMK’s positive void coefficient that turned Chernobyl’s test into a runaway. 🔬 You don’t need an operator to recognize the problem and throw a switch. The reactor does it on its own.
The second change is passive cooling, which is the direct lesson from Fukushima. X-energy’s Xe-100 reactor description from the company is unambiguous: it is designed to be “walk-away safe.” Here’s what that means in practice:
Gravity feeds cooling water without pumps or electricity
Natural convection moves heat away through the laws of physics, not mechanical systems
The reactor self-stabilizes over hours and days without a single human intervention or external power source
Emergency planning zones shrink from a 10-mile radius to the plant site boundary — a few hundred meters
NuScale demonstrated this to the U.S. Nuclear Regulatory Commission successfully enough that the NRC approved the reduced emergency planning zone. That’s not marketing language. That’s a regulatory body with nuclear accidents in its institutional memory signing off on the physics.
The third change is the fuel itself. TRISO fuel — used by X-energy’s Xe-100 and Kairos Power’s designs — embeds the uranium in ceramic and graphite coatings that maintain structural integrity above 1,600°C. As researchers at the Nuclear Technology journal documented in late 2024, TRISO particles remain stable under conditions far worse than those seen at Three Mile Island, Fukushima, or Chernobyl. The fuel physically cannot melt in the way conventional fuel rods do. Each tiny particle is its own miniature containment structure. 🧬
Could the Chernobyl scenario specifically happen?
Probably not — and the reasons are structural, not procedural. The Chernobyl disaster needed three specific ingredients to come together: a reactor design with inherently unstable physics, the deliberate disabling of safety systems, and operators who didn’t understand what they were doing.
No Western regulator would license a reactor with a positive void coefficient today. The IAEA’s safety standards for new reactor designs explicitly require inherent safety in loss-of-power and loss-of-cooling scenarios. As the World Nuclear Association notes, Department of Defense specifications for mobile microreactors demand meltdown be “physically impossible” in various complete failure scenarios. That is the design floor, not the aspirational ceiling.
The graphite fire at Chernobyl is also not replicable in most modern designs. Light-water SMRs don’t use graphite as a moderator. And while some gas-cooled SMRs do use graphite, their geometry and fuel type prevent the kind of rapid oxidation that turned Reactor 4 into a burning, open-air radioactive source for 10 days.
Does that mean a bad operator could never make a serious mistake at an SMR? No. Humans will always be humans. But the specific nightmare — a design flaw that amplifies operator error into a runaway chain reaction — is not a feature of any currently licensed or near-licensed SMR design. The physics works against it, not with it.
Could Fukushima happen?
This is the more honest question, and the answer is more nuanced. 💡 Fukushima happened because active cooling pumps lost power. If an SMR relies on the same active cooling architecture, it has the same vulnerability.
Most next-generation SMR designs have specifically addressed this. The key distinctions:
NuScale’s design uses natural circulation — no primary coolant pumps at all in normal operation, making pump failure a non-issue
X-energy’s Xe-100 uses TRISO fuel that generates so little decay heat, relative to its heat capacity, that temperature rise in a complete loss-of-cooling scenario stays well within safe limits
TerraPower’s Natrium reactor uses liquid sodium, which has a very high heat capacity and provides passive decay heat removal without pumping
Harry Keeling, head of Business Development at Rolls-Royce SMR, told Data Center Dynamics in March 2026 that their reactor is deliberately designed to be “boring.” In nuclear engineering, boring is a five-star review. A boring reactor is one where nothing interesting happens when things go wrong.
That said, Britannica’s SMR entry fairly notes that SMRs are still “vulnerable to natural disasters and weather-related events that can cut power used to control reactor operations.” The degree of vulnerability is massively reduced in passive designs, but it is not zero. Honesty matters here. No engineered system is risk-free. What passive SMR safety systems do is change the failure timeline from minutes (Fukushima’s pumps stopped, the clock immediately started) to days or weeks — enough time for an orderly response rather than a desperate scramble.
What do you think is the bigger obstacle to SMR adoption: public fear of a Fukushima-style accident, or the actual engineering tradeoffs that still exist in current designs? The difference matters for how the conversation should go.
Could Three Mile Island happen?
Three Mile Island is the most instructive case for light-water SMRs, because most of the near-term deployments — NuScale, GE Hitachi’s BWRX-300, Holtec’s SMR-300 — are still light-water designs. They use the same basic physics as the reactor at TMI.
The direct TMI scenario — operators turning off emergency cooling water while misreading instruments — is much harder to replicate in modern designs for two reasons. First, digital instrumentation has replaced the confusing, contradictory analog gauges that misled TMI’s operators. Second, and more important, modern designs incorporate passive emergency core cooling systems that don’t require operators to activate them. If coolant drops below a threshold, gravity-fed water flows in automatically. There is no switch to misread or turn off.
As Wikipedia’s passive nuclear safety article explains, the TMI accident directly exposed the design deficiency of reactors that couldn’t remove decay heat without active cooling after shutdown. Third-generation and Generation IV designs specifically exist to close that gap, using:
Gravity-driven coolant injection systems
Natural convection heat removal that requires no power
Large water reservoirs above the reactor that drain by gravity in emergencies
The ADVANCE Act of 2024 also introduced regulatory reforms that streamline SMR licensing while embedding modern safety requirements, including 50% fee reductions for applications and faster parallel review of early site permits and design certifications. ⚡ More speed in the system, but no loosening of the physics-based safety floor.
The honest remaining concern for TMI-style scenarios is not the reactor core itself — it’s the spent fuel pools. Fukushima’s spent fuel at Unit 4 was nearly as alarming as the reactor cores, and spent fuel management remains a legitimate open question for any nuclear technology. The IAEA and OECD/NEA have both flagged that the backend nuclear fuel cycle needs to be addressed from the very beginning of SMR design, not retrofitted later.
The honest accounting
The three famous accidents each had a specific failure mode. Chernobyl needed a uniquely bad reactor design and deliberate disabling of safety systems. Fukushima needed active cooling to fail completely, permanently, with no backup. Three Mile Island needed confused operators working with bad instrumentation on a reactor that couldn’t passively remove decay heat.
Modern SMR designs address all three failure modes — some through physics, some through fuel chemistry, some through passive system design that removes humans and electricity from the critical path. That’s not nothing. That’s actually quite a lot.
But here’s what’s also true. Allison Macfarlane, former chair of the U.S. Nuclear Regulatory Commission, has been blunt: most SMR designs are still on paper. As of early 2026, only Russia and China have operational SMRs. The U.S. Department of Energy selected Tennessee Valley Authority and Holtec for grants of $400 million each in December 2025 to support early deployment. First commercial SMRs are targeted for 2029-2030, and those timelines have slipped before.
The safety architecture of SMRs is genuinely different from what failed at Chernobyl, Fukushima, and Three Mile Island. The physics is different. The fuel can be different. The cooling strategy is different. Whether that difference translates into a generation of reactors that never produce a headline-grabbing incident — that’s something the next decade will actually prove or disprove. 🌱
So here’s the question to sit with: if the safety case for SMRs is as strong as the engineering suggests, what is actually driving the delay between the design and the deployment?