Case Study 1: Three Accidents — TMI, Chernobyl, Fukushima: What the Physics Tells Us

Introduction: Three Reactors, Three Failures, Three Different Lessons

The three major accidents in the history of commercial nuclear power occurred in three different countries, with three different reactor types, across three different decades. At first glance, they appear to have little in common. But viewed through the lens of reactor physics, a unifying pattern emerges: in each case, the accident was the result of a specific physical mechanism interacting with specific human and organizational failures. Understanding the physics is not optional — it is the only way to separate what actually happened from the mythology that surrounds these events.

This case study examines each accident in detail, focusing on the nuclear physics mechanisms that drove the event sequence.

Part I: Three Mile Island (March 28, 1979)

The Physics: Loss of Coolant and Decay Heat

Three Mile Island Unit 2 was a 2,772 MW(th) Babcock & Wilcox PWR in its first year of commercial operation. The physics of the accident is straightforward — it is a textbook case of decay heat overwhelming a compromised cooling system.

Event sequence (physics-relevant steps):

The critical physics:

1. Decay heat. After SCRAM, the fission chain reaction stopped immediately. But the ~200 radioactive fission product species in the core continued to decay, generating heat at ~6% of operating power (initially ~170 MW). This decay heat is not optional — it is dictated by the laws of radioactive decay and cannot be turned off. It decreases following an approximate $t^{-0.2}$ power law but remains at tens of megawatts for hours.

2. Zirconium-water reaction. When fuel cladding temperatures exceeded ~1,200°C, the exothermic reaction $\text{Zr} + 2\text{H}_2\text{O} \to \text{ZrO}_2 + 2\text{H}_2$ accelerated. The reaction releases ~6.5 MJ/kg Zr — adding significant heat to an already overheating core and producing hydrogen gas. The hydrogen later burned in the containment building (a brief pressure spike to ~2 bar gauge) but the containment held.

3. The PORV: a small-break LOCA. The open valve created a ~2-inch hole in the primary pressure boundary, leaking roughly 20 kg/s of coolant. At this rate, the core would uncover in roughly 2 hours — and it did. The operators' decision to throttle the ECCS converted a manageable event (with ECCS providing adequate makeup flow) into a severe accident.

What the physics teaches: The accident was not caused by an exotic nuclear failure. The chain reaction stopped correctly. The containment held. The decay heat was manageable with the available safety systems. The accident occurred because operators, trained to fear "going solid" in the pressurizer (a real but far less dangerous concern), overrode the safety systems that would have prevented core damage.

Health impact quantification: The total radioactive release was approximately 370 GBq of ${}^{131}\text{I}$ and 555 GBq of noble gases. The maximum calculated dose to any individual within the 16-km zone was estimated at 0.8 mSv — less than a single chest CT scan and far below the threshold for any deterministic health effect. Multiple long-term epidemiological studies (Hatch et al., 1990; Talbott et al., 2000) found no statistically significant increase in cancer incidence.

Part II: Chernobyl (April 26, 1986)

The Physics: Positive Void Coefficient and Prompt Criticality

Chernobyl Unit 4 was an RBMK-1000 — a graphite-moderated, light-water-cooled reactor of uniquely Soviet design, producing 3,200 MW(th). The accident at Chernobyl was fundamentally different from TMI: it was a reactivity-driven event, not a loss-of-coolant event.

The setup: a test gone wrong

On April 25–26, operators were conducting a turbine rundown test — measuring whether the spinning-down turbine could generate enough electricity to power the emergency cooling pumps during the ~60-second gap between a station blackout and diesel generator startup. This test had been attempted before and postponed. The operators were under pressure to complete it.

Event sequence with physics annotations:

The critical physics — a quantitative analysis:

1. The positive void coefficient. In the RBMK at low power:

$$\alpha_v = \frac{\partial \rho}{\partial \alpha_{\text{void}}} \approx +4 \text{ pcm per percent void}$$

As coolant flow decreased during the test, void fraction increased from ~3% to ~30% or more. The positive reactivity insertion:

$$\Delta\rho \approx 4 \times 27 = +108 \text{ pcm} \approx +0.23\$$$

This alone would not be catastrophic. But with almost no control rods in the core, there was no negative reactivity margin to compensate.

2. The AZ-5 catastrophe. The RBMK control rods had a fatal design feature: 4.5-meter graphite "displacer" sections at the bottom of each rod, with the neutron-absorbing section above. When the rods were fully withdrawn and the SCRAM signal was given, the graphite tips entered the lower core first — displacing water (a neutron absorber) with graphite (a moderator). This briefly increased reactivity by an estimated 0.5$ in the lower core, exactly where the power was highest.

The total reactivity insertion — from voiding plus the graphite-tip effect — pushed the reactor past prompt critical:

$$\rho > \beta = 0.0048 \quad (\text{i.e., } > 1\$)$$

3. Prompt supercriticality. At $\rho \approx 100\$$ (estimated), the reactor period was: $$T \approx \frac{\ell_p}{\rho - \beta} \approx \frac{10^{-4}}{100 \times 0.0048 - 0.0048} \approx \frac{10^{-4}}{0.475} \approx 2 \times 10^{-4} \text{ s}$$ Power doubled every $T \ln 2 \approx 0.14$ ms. In roughly 3 seconds, the power rose to an estimated 30,000 MW(th) — *100 times* the nominal rating — before the fuel itself disintegrated, ending the excursion by dispersing the fissile material. **4. The graphite fire.** 1,700 tonnes of graphite at ~700°C, now exposed to air through the blown-off reactor lid, caught fire. Graphite burns slowly but at high temperature, and the fire lofted radioactive material to altitudes of 1–2 km over the following 10 days, spreading contamination across northern Europe. **Why Chernobyl cannot happen in a Western reactor:** | Feature | RBMK | PWR/BWR | |---------|------|---------| | Void coefficient | Positive (at low power) | Negative | | Moderator | Graphite (separate from coolant) | Water (same as coolant) | | SCRAM mechanism | Graphite-tipped rods (initial positive effect) | Absorber-only rods (always negative) | | Containment | Partial (reactor hall, not pressure-rated) | Full pressure-rated containment | | Can operate with rods fully withdrawn? | Yes (operators violated rules) | Physically impossible (interlocks prevent it) | **Health impact:** 28 of the 134 emergency responders diagnosed with acute radiation syndrome died within 3 months (individual doses of 1–16 Gy). The most significant public health effect was thyroid cancer in children exposed to ${}^{131}\text{I}$ in the first days before evacuation and dietary restrictions: ~5,000 cases as of 2005, with a 99% survival rate. The overall excess cancer mortality is estimated at ~4,000 by the WHO (2006) but remains uncertain because of the difficulty of detecting a small excess signal against the large natural cancer background. --- ## Part III: Fukushima Daiichi (March 11, 2011) ### The Physics: Decay Heat Without Electricity Fukushima Daiichi Units 1–3 were GE BWRs (Mark I containment) with thermal ratings of 1,380–2,381 MW. The accident at Fukushima was a *decay-heat removal* failure — the same fundamental physics as TMI, but at a far larger scale and driven by a natural disaster that exceeded all design assumptions. **The physics of station blackout:** | Time after earthquake | Event | Decay heat (MW, Unit 1) | |----------------------|-------|------------------------| | 0 s | Earthquake; auto-SCRAM | 1,380 → 0 (fission); decay heat ~83 | | +50 min | 14-m tsunami arrives | ~65 MW | | +50 min | All AC power lost (SBO) | Batteries provide DC for ~8 hours | | +5 hours | Battery power exhausted (Unit 1) | ~40 MW; no cooling | | +15 hours | Core fully uncovered (Unit 1 est.) | ~25 MW | | +24 hours | Hydrogen explosion (Unit 1) | ~20 MW | | +62 hours | Hydrogen explosion (Unit 3) | ~12 MW | | +87 hours | Hydrogen explosion (Unit 4 building) | — | **The decay heat problem in quantitative terms:** At shutdown, each reactor produced decay heat at approximately 6% of operating power. For Unit 1 (1,380 MW(th)): $$P_{\text{decay}}(t = 0) \approx 0.06 \times 1380 = 83 \text{ MW}$$

At 1 hour: $P_{\text{decay}} \approx 55$ MW. At 1 day: $P_{\text{decay}} \approx 15$ MW.

A typical BWR fuel assembly can tolerate loss of cooling for roughly 2–4 hours before cladding temperatures reach the ~1,200°C threshold for the Zr-H$_2$O reaction. The station blackout at Fukushima persisted for days. Core damage was inevitable once AC power was lost and battery power was exhausted.

The hydrogen problem: The same Zr-H$_2$O reaction that occurred at TMI produced large quantities of hydrogen at Fukushima. Unlike TMI, where the hydrogen burned inside the containment (which held), the Fukushima hydrogen leaked from the containment through hardened vent lines and accumulated in the reactor buildings above, where it detonated with devastating force — destroying the buildings but not the primary containment.

What the physics teaches:

1. The reactors worked correctly. The control rods inserted, the chain reaction stopped, and the pressure vessels initially maintained integrity. The failure was not nuclear but electrical.

2. Decay heat is non-negotiable. You cannot argue with radioactive decay. Any reactor design must have a means of removing decay heat that does not depend on external power, or the consequences of prolonged station blackout are core damage.

3. The 14-meter tsunami was roughly 2.5 times the design-basis tsunami (5.7 meters). The seawall was inadequate. The diesel generators were in the basement. The backup batteries lasted only 8 hours. Each of these was a known vulnerability; together, they were fatal.

Health impact: No deaths from radiation. The UNSCEAR 2021 report found "no adverse health effects among Fukushima residents are directly attributable to radiation exposure." The evacuation itself caused approximately 2,300 deaths (the Reconstruction Agency's official figure) from stress, medical care disruption, and suicide — a stark reminder that the response to a nuclear accident can be more harmful than the radiation.

Synthesis: What the Physics Demands

The three accidents, taken together, reveal the non-negotiable requirements that reactor physics imposes on reactor design:

The most important lesson is perhaps the simplest: the physics does not care about your intentions, your procedures, or your schedule. Neutrons multiply according to $k_{\text{eff}}$, fission products decay according to their half-lives, and decay heat must be removed regardless of whether the pumps have electricity. Any reactor design that does not accommodate these physical realities is fundamentally unsafe. Any design that does — with passive, physics-based safety margins — can achieve a safety record unmatched by any other energy technology.

Discussion Questions

1. Defense in depth. The concept of "defense in depth" requires multiple independent barriers between radioactive material and the environment. Identify the barriers for a PWR and explain which barrier failed in each of the three accidents.

2. The role of human factors. In TMI, operators worsened the situation; in Chernobyl, operators caused it; in Fukushima, operators were heroic but powerless. What does this suggest about the role of automation vs. human judgment in reactor safety?

3. Positive vs. negative void coefficient. A student argues: "Chernobyl proves that nuclear power is inherently dangerous." Using your understanding of void coefficients, explain why this statement is physically incorrect as a generalization. What would be a more accurate statement?

4. Decay heat as a design constraint. Gen III+ reactor designs (AP1000, BWRX-300) use passive decay heat removal systems. Describe qualitatively how natural convection and gravity-driven water injection can remove decay heat without pumps. Why was this not part of the original GE BWR Mark I design used at Fukushima?

5. Risk communication. The Fukushima evacuation killed more people than the radiation. How should nuclear safety policy balance the risk of radiation exposure against the risk of evacuation? What quantitative tools (dose projections, LNT model, hormesis data) should inform such decisions?