Assessing and managing tsunami risks
Fukushima Daiichi crisis | Tsunami27 October 2011
Since the Fukushima accident there has been greater awareness of the risk from tsunami to reactors in seismically-active areas. It is important, however, to realise that tsunami include a wider range of phenomena that could impact a variety of nuclear facilities in different geographical settings. By Ian G. McKinley, W. Russell Alexander and Hideki Kawamura
The tragedy of the March 11th Tohoku earthquake and tsunami has focused attention on the risks of such events for reactors located on the coast in seismically active regions.
The potential for large tsunami impacting the north east coast of Japan was known (see photo) and the particular concerns for reactors and other nuclear facilities in such settings has been identified [1, 2]. Although post-mortem assessment is currently evaluating the extent to which such warnings were ignored by the operator, TEPCO, it is fair to say that, in the absence of previous nuclear experience of an event of this type, the risk of such massive failure of defence-in-depth had also fallen under the radar in most countries. This reality forms the basis of many of the recommendations for action for regulatory bodies (as identified in the IAEA fact finding mission and Weightman reports [3, 4]).
To date, the heroic efforts of Japanese reactor staff have limited the off-site consequences of the disaster, to the extent that nobody has received life-threatening radiation doses despite major damage to three reactor units and four fuel storage ponds. Even ‘statistical deaths’ in the future as a result of contamination (both on- and off-site) are likely to be very few – certainly in comparison to Chernobyl  and even the less well-known accidents at Windscale  and Kyshtym . Nevertheless, Fukushima has dominated both Japanese national and international media. Indeed, when comparing international media coverage, the Fukushima incident has even eclipsed the 2004 Indian Ocean tsunami, which caused loss of life and damage to infrastructure on a scale of an order of magnitude larger than that in Japan. It is clear, therefore, that—regardless of absolute hazard —nuclear facilities require special assessment of risks and, even if of low likelihood, tsunami need special consideration.
The historical record
A tsunami may be defined as a series of waves that are generated when a large volume of water...is displaced by an impulse disturbance such as an explosion, earthquake, volcanic eruption, landslide or meteorite impact .
The two great tsunami of the 21st century were caused by huge ‘megathrust’ earthquakes, which are a characteristic of the subduction zones found at the boundaries of tectonic plates. The locations where such events may occur are well-known  and, although individual events cannot be predicted, the global frequency of magnitude 9 quakes like these with the potential to generate large tsunami is probably in the order of five per century. Such tsunami may cause significant damage and loss of life at distances thousands of kilometres from their epicentres.
Looking at the major destructive tsunami of the last few centuries, however, it is clear that many were generated by earthquakes of considerably smaller magnitude. Although global effects are more limited, major devastation may result locally as the tsunami follows the destruction caused by the initial earthquake, as in the 1775 Lisbon earthquake and tsunami, one of the deadliest in recorded history. There are several more recent examples, such as the tsunami resulting from the magnitude 7.9 quake in Mindanao, Philippines in 1976; the 7.9 magnitude quake in Tumaco, Columbia in 1976; the 7.8 magnitude quake in Hokkaido, Japan in 1993, and the 7.1 magnitude quake in Papua New Guinea in 1998.
It is important to realise that earthquakes are not the only tsunami sources and, indeed, are not directly responsible for the very largest waves in the geological record. In particular, volcanoes and landslides are common tsunami sources and may, indeed, produce waves so large that they are classed as megatsunami: many tens or hundreds of metres height. Nevertheless, it is often difficult to decouple the tectonic and volcanic events that may contribute to the large earth slides that usually cause megatsunami.
A volcanic example is the 1792 Mt Unzen eruption in Kyushu, Japan where volcanic activity (possibly along with associated earthquakes) cause the collapse of the eastern flank of the volcanic dome, resulting in a tsunami with waves reaching 100m high. More recently, the 1883 Krakatoa eruption in Indonesia generated tsunami reaching heights of 40m. As an example of tsunami generation distant from the coast, the 1980 eruption of Mt St Helens in Washington USA caused a major landslide into Spirit Lake that produced a 260m-high megatsunami.
Even major landslides without clear tectonic or volcanic links can also give rise to major tsunami. Of particular importance in terms of nuclear facilities may be localised rock- or ice-falls into constricted bodies of water that can generate huge waves. Examples are the 500m high megatsunami in Lituya Bay, Alaska in 1958 or the 200m high wave from the Vajont Dam in Italy in 1963.
Landslides big enough to generate regional or global scale tsunami are relatively rare, but include several Storegga Slides in the Norwegian Sea (the most recent about 8000 years ago), which are assumed to be generated as a result of glacial unloading. Nevertheless, such submarine collapse of coastal shelf slopes has been postulated as a potential consequence of methane clathrate destabilisation as a result of global warming—so geological records should not be blindly extrapolated to calculate probabilities of future recurrence of such events.
Even bigger tsunami can result from the collapse of seamounts such as the Canary and Hawaii Islands. Even though the most recent event of this type occurred about 10,000 years ago, there is good geological evidence of multiple debris flows at these sites and no convincing reason why they could not also occur at similar, less-studied locations.
Finally, for completeness, it can be noted that meteorite impact is also a source of large tsunami. These are, however, very rare events, the last substantial one being the Eltanin Impact 2.15 million years ago .
In general, the probability of a tsunami as a function of peak wave height follows a power-law distribution, but the actual risk is very site-specific. In order to carry out quantitative risk analysis, local palaeotsunami data can be combined with formal methods that combine information on potential sources with regional topographic data that determines wave propagation and impact .
It is notable, however, that the development of a tsunami assessment method for nuclear power plants in Japan  considered only earthquake-induced tsunami and a much lower observational database derived from relatively recent historical records. This led to very obvious underestimates of total inundation risk (see  for example).
The impact of a tsunami of particular size is influenced by the source of the water displacement, particularly if it is a major earthquake and/or volcanic eruption relatively nearby . In such a case there may be little warning of the tsunami and additional complications due to damage and regional disruption caused by ground movement, ash falls, and so on.
Major nuclear facilities may generally be sufficiently strong to withstand the initial impact of even large waves, but it has to be ensured that they are strong enough to resist also the high drag forces, which are rarely considered during design, but can greatly exceed the surge force for larger waves (Figure 1). Also, as the dramatic video material from Japan has shown, it is not only the wave itself that can cause damage, but also large objects carried with it, including ships torn from moorings, buildings ripped from their foundations, oil tanks, etc. It is also important that access hatches, ventilation intakes and so on are protected from flooding.
It is also essential to ensure that all critical infrastructure providing control, communication, monitoring, power, cooling, ventilation, drainage, transportation, etc. is protected from both mechanical damage and inundation. Especially in areas where earthquake and inundation risks are considered to be low, these may be more vulnerable than the major facilities that they support. But even in an area with known risks like Fukushima, scenarios producing common-mode failure of redundant systems may be insufficiently analysed.
Rapid shutdown (‘scram’) of reactors can be achieved either automatically or manually as a result of either warning from tsunami-monitoring networks or seismic networks picking up the initiating event. While these are well-proven in areas of high risk, they are not universally adopted and might be considered even in cases where probabilities are low.
As illustrated at Fukushima Daiichi, however, core cooling after the scram is more challenging and requires that both the cooling system is mechanically intact and power is available for pumps, valves, critical instrumentation and safety equipment (for example, hydrogen recombiners). For modern reactor buildings, the mechanical forces from even large waves should not exceed design specifications, especially when they are designed to contain pressure, and hence are less vulnerable to surge forces.
It is certain that robustness of emergency power supplies will be a focus for future studies, especially when these are currently situated in un-hardened structures outside the reactor building. Nevertheless, it should not be forgotten that there are a number of secondary perturbations that could give rise to cooling problems. As shown already in the 2004 Indian Ocean case, the initial drawback of water before the arrival of the tsunami can approach the level of sea cooling water intake pipes, which could potentially lead to early pump failure [1, 2]. This is directly addressed in the current Japanese hazard assessment, which quantifies maximum draw-down in addition to wave height [10, 11]. A further concern could be blocking of cooling water intake by tsunami debris (or, more seriously, infill of the cooling water intake by ash in the case that local vulcanism initiates the tsunami).
Although the discussion tacitly applies to power reactors, the basic principles are applicable also to research reactors. Although their power densities and contained radionuclide inventories may be much lower, this may be balanced by less robust engineering and institutional defence systems.
A wide range of large-scale nuclear facilities may be vulnerable to tsunami perturbations, including uranium conversion, enrichment, fuel fabrication, research laboratories and, in some countries, military installations. The most potentially-significant type of nuclear facility, however, is reprocessing plants, as these tend to have, by far, the greatest inventories of radiotoxic materials on the premises. Relative to a reactor, such plants contain a more complex array of active facilities, which include those producing, transporting and storing highly active liquid wastes. As is known from past accidents (particularly at Kyshtym), highly-active liquid waste management requires active control and, in case its fails, is liable to temperature excursions, build-up of explosive gases and in some cases, even criticality. Ensuring safety thus requires that all key buildings are resistant to both the initiating event (if nearby), the tsunami itself, and subsequent loss of services that are not in hardened structures.
The Fukushima accident has illustrated the vulnerability of fuel storage pools in the event of complete loss of power. This is especially the case for the BWR designs involved, as their raised location (to aid fuel handling) makes them particularly susceptible additionally to earthquake damage. Nevertheless, fuel storage ponds are a general concern: delays in implementing central storage, reprocessing and geological disposal have all resulted in fuel reracking to increase density of storage, increasing the potential impact of major perturbations. Although inventories of long-lived radionuclides may be larger than those in reactor cores, absence of primary and secondary containment may make pools inherently more vulnerable to both direct impacts of tsunami (plus the initiating event) and also to loss of services, particularly cooling (as seen at Fukushima Daiichi, and particularly for the highly-loaded unit 4 pool).
Dry storage systems for both spent fuel (SF) and other highly active material (such as vitrified high-level waste (HLW)) are generally more physically robust, especially if stores are strong (for example, designed to withstand aircraft impacts) and rely on natural convection for cooling. Flooding could certainly give rise to management problems, but barrier degradation to the point that there is a risk of release of radioactivity is probably unlikely.
Although the hazard associated with stored low active wastes is relatively low, container damage and subsequent contamination of the local tsunami debris field would certainly be of great public concern. Concerns include physical damage to storage buildings and waste containers (which may be coupled in the case of building collapse), flooding and destruction of infrastructure (power, drainage, monitoring and control systems).
For open surface or near-surface repositories, the situation is similar to waste stores, where physical damage of containers and flooding can combine to produce a potential source of off-site contamination. Even in cases where a repository is partially closed, wave impacts and scouring could damage containment and flood monitoring systems, requiring costly remediation.
Deeper disposal facilities are inherently less directly vulnerable to tsunami, but concerns focus on the risk of flooding via access/ventilation ramps and shafts and destruction of surface infrastructure–in particular the supply of power, drainage and ventilation. Risks of releases off-site may be small, but the hazard to operators and the potential need for expensive remediation needs to be considered.
To date, despite the existence of many coastal disposal sites, there has been little assessment of tsunami risks and only recently more detailed analysis for deeper repositories .
For new facilities, a clear question is whether the risk of tsunami impact justifies consideration of siting measures to limit it. Alternatively—and this is the only option for existing facilities—risk can be managed by consideration of tsunami barriers or modified facility design.
There is clearly a trade-off between the advantages of locations close to the sea or large water bodies (availability of cooling water; ease of ship transport) and the desire to avoid risk of tsunami. Nevertheless, measures can be taken to assess local probability of very large tsunami (for example, potential for landslides or volcanic eruptions, Quaternary palaeotsunami record) and, where possible, high-risk locations can be avoided. It is important to note here that the local impact of any tsunami may be greatly influenced by regional topography and/or bathymetry which can either reduce or increase the maximum wave size by a considerable margin (factor of 2-3). This can be readily assessed on a site-specific basis by current tsunami propagation models.
More generally, siting on a raised position will always provide benefits; hills tend to divert waves around it. Distance from the coast (or water body) will also reduce risks, although large tsunami may have a run-up kilometres long on flat plains.
For protection of an entire site, tsunami walls are a common approach. These can be combined with offshore breakwaters to disperse or refract the wave. Although such systems work well for smaller-wavelength storm surges and small tsunami, they may be overwhelmed by larger waves. Long walls, in particular, if perpendicular to the tsunami, may simply cause large waves to build up until the wall is overtopped. Nevertheless, when combined with local tsunami propagation models, the effectiveness of different wall designs can be readily assessed.
Due to the difficulty of predicting low-probability, high-consequence events, it is prudent, where possible, to ensure that key safety functions are not lost even in the event of barrier overtopping. This is particularly the case for very large industrial facilities (such as reprocessing plants) where the cost of a barrier against low-probability events surrounding the entire site may be prohibitive.
Where the option exists, construction of nuclear facilities (or key infrastructure) underground may reduce risks—although here supply of cooling water and protection of access points needs to be assured. A possible advantage of this option is that it is not specific to tsunami and may, in parallel, reduce risks from earthquakes, extreme weather, terrorist attacks, and so on.
It is important not to overstate the risks to nuclear facilities after Fukushima: seen objectively, both the direct and long-term hazards from radiation released are less than those from the subsequent bio-beansprout epidemic in Europe . Nevertheless, great public concern and the resultant huge costs associated with remediation mean that tsunami risks should be carefully evaluated for all nuclear facilities.
This article was originally published in the October 2011 issue of Nuclear Engineering International (p14-17)
We thank many colleagues in Japan and abroad who have provided insight into the Fukushima incident and background on the wider topic of tsunami risk assessment in Japan.Related ArticlesTEPCO rebuts media reports about 2008 tsunami predictionsExternal weblinksNuclear Engineering International is not responsible for the content of external internet sites.References