Lessons Learned from Fukushima: part I – the Technical

I’ve spent some time over the last few weeks talking about “lessons learned from Fukushima” to several different audiences. There are many different ways to look at this event and many different “lessons” that are to be taught. However, I think it is useful to consider those lessons in about three different general classes.

1) Technical Lessons

These are internal lessons about what equipment or systems worked in what ways at the plant. There are some very specific lessons that all nuclear facilities should be looking at and learning from

2) Corporate Lessons

Corporate lessons are the business responses that TEPCO had to the event. These are also internal to companies that are managing nuclear facilities, but could be learned by any corporation operating large industrial facilities.

3) Political Lessons

It is always hard to take lessons at the political level, but understanding how the interactions between corporations, regulators, and government, both nationally and globally worked is critical in improving emergency response.

Each should be examined for what worked, what didn’t, and why. In the next three weeks, we will examine each. Whether or not people have died in this event, the terrible toll being paid by the evacuated people and the workers at the plant site and the immense cost of response require that we look at what happened and try to determine ways to prevent it from happening ever again.


Let’s look at things in the order of occurrence, as a sort of step through the defense-in-depth features of these plants.

1) Natural disasters

The specific events (earthquake and tsunami) shouldn’t be the question. I have had a number of people asking me if they should be worried about the plants in their neighborhood (like Iowa or North Carolina). My stock answer has been “If you have a 9.0 subsidence earthquake and a 15 meter tsunami in Iowa, we have WAY more to worry about than the nuclear facility.” In other words, we need to look at each facility site and understand the risks and potentials for natural and man-made disasters at that facility.

In the U.S. this is an ongoing effort. Every time something happens through the INPO reporting systems and the NRC assessments, the entire industry looks at each facility and assesses any lessons learned or changes that need to be made.However, it is clear that we need to remain vigilant against complacency while balancing cost vs. risk assessments of these potentials.

2) Long term Station Blackout (SBO)

Generally two issues stand out. Either the possibility of common cause failure needs to be eliminated, or the facility needs to be able to manage for a longer period before regaining power. The inherent issue here is how long is long enough and the fact that batteries aren’t really a practical option for driving pump power.

3) Ultimate Heat Sink

The underlying issue of SBO is one of maintaining the ultimate heat sink during those early critical hours when the decay heat in the reactor is significant and can cause major fuel failure. Loss of the heat sink is the ultimate reason for the catastrophic failure of the fuel in core. Whether a solution separate from the SBO issues is required isn’t clear, but the issue is one to consider.

4) Spent Fuel Pools

The issues with the spent fuel pools are still evolving. Claims made internationally regarding the status of the pools in the early days of the event have been clearly proven false. However, at a minimum these pools represented a significant diversion of resources for TEPCO that could have been better spent elsewhere.

5) Hydrogen

aside: I’ve been told that many believe that hydrogen explosion to be like a hydrogen bomb. That is not the case. The explosion we’re talking about here is that of hydrogen and oxygen recombining rather violently to make water. It is the same mechanism that caused the explosion of the Challenger Shuttle in the 1980’s :end aside.

Much speculation has been made as to the sources of the hydrogen that caused the explosions in units 1, 3, and 4. The only thing we KNOW at this point is that the unit 4 pool was NOT the source of the hydrogen in that explosion. I explained this in more detail in a prior entry (Nuclear Power and the Witch Hunt). Given everything else, it is reasonable to assume that the hydrogen came from the zirconium cladding inside the reactor cores of units 1, 2, and 3.

How that hydrogen migrated to places where it could freely combine with oxygen is not understood at this point. Having that knowledge is absolutely necessary to determine appropriate mitigation. Until that is understood, comments about hydrogen recombiners or hardened vents or other issues and or solutions are mere speculation and do not serve to improve plant safety.

We can go on forever in considering possible technical issues and actions to prevent or mitigate them. As this event stabilizes and more time can be given to expert review of the equipment and events, knowledge will be gained. That knowledge will be used to make plants in the US safer and more secure.

Next week, corporate lessons learned.

12 thoughts on “Lessons Learned from Fukushima: part I – the Technical”

  1. Margaret,

    Good write-up. Thanks.

    With regard to the hydrogen explosion at Fukushima Unit 4, one possibility seems to be cross-flow from Unit 3 through the combined vent stack for Units 3 and 4. While the stack arrangement should have prevented such flow, it seems possible that it may not have.

    Just a thought . . . .


    Lars Hanson

  2. Good start, Margaret. If a may add some other thoughts?

    1 – Back up power. In the UK and Europe, consideration of use of equipment not physically connected to the plant, and certainly not anything stored off-site is not considered as part of the safety cases, and therefore doesn’t form part of requirements for operation (with the exception of fire-fighting). I assume this is the same in the US.

    I now believe this to be a profound error. The I understand the great majority of the damage at Fukushima had spare DG sets been available, stored offsite in secure, near-proximity storage. This could even be split across sites, to introduce further redundancy.

    Some plant modifications would be required, to facilitate connections of external gennies, however overall costs would be small (sub $10m/plant).

    It wouldn’t prevent problems due to damage to switchgear and dstribution boards, of course, but these are relatively easily protected.

    2 – Ultimate heat sink. Again, true – but once more cheap to solve. On site or near-site storage for sufficient water for a “once through” cycle, and scavenge/storage for contaminated water for several days worth of post shutdown cooling ought not to be a major cost or challenge. Assuming a 10MW decay heat removal requirement for 5 days for a 1GW unit, and keeping the upper limit of temperature to about 60C looks like needing a supply reservoir 5m deep and 200m square. Large, but not infeasible, and certainly not a cost that’s prohibitive in the context of cost or space requirements.

    3 – Hydrogen. I no longer have access to the sorts of computer codes I’d need to validate this – but, are there options other than zircalloy viable in LWRs? When we in the UK went from (natural uranium) fuelled MAGNOX to AGRs, we switched to stainless steel cladding. Partly because magnesium was seen as too reactive in accident circumstances!

    Stainless cladding has advantages – it’s mechanically stronger than ziracalloy, easier to fabricate, and much cheaper. Thermally it’s similar (slightly better in that it keeps strength better at very high temperatures. But there is some neutron capture.

    That necessitated enrichment – but not to scary levels; we use 2.5% (i.e 1.8% over natural)

    Making the very flaky assumption that the additional enrichment is addative, that would suggest LWRs running on 4-5% enrichment. Assuming reasonably low cost enrichment, that’s not infeasible. If it is unattractive, are there passivating treatments that would make prevent surface production of hydrogen?
    As an aside, I’m not adverse to using seaawter for short periods – recognising that this is a “last ditch” option.

    • Andy,

      1 – they brought in DGs in a fairly short period of time, but could not hook them up. There was significant damage to the boards as well apparently. However, your point is a good one. There are some additional requirements in the US as a result of 9/11 assessments (sometimes called B5b assessments), but because of security concerns not all of that info is publicly available.

      2 – Yes, but not something that had been considered necessary in the past. Pumping and decon and all the other considerations make it all more complicated. No two events run the same way, so making sure the system is robust will be more complex.

      3 – The original fuel designs were SS cladding, but the neutron scavenging was considered unacceptable. LWRs are, indeed, operating at 4-5% enrichment, near the maximum of the manufacturing facilities, so additional enrichment is not feasible. However, some interesting work in advanced materials might get accelerated with this event. The neutron flux makes the environment a challenge.

      • I was aware of the hook-up problems, and indeed, that was largely the inspiration for the problem – had they had compatible voltages/connectors, I’ve a hideous feeling this accident might now be looking like a triumph of system resilience and operator improvisation!

        Agreed re on-site storage, howver, a highly visible (hence reassuring to the public) and low cost measure.

        As to fuel – true, 4-5% probably was near the commercial operating limits for early generation GD enrichment – but not for more recent kit, most notably centrifuges) or, if GE-Hitachi can make the laser process work commercially). I understand the Us is not only now allowing URENCO to build a centrifuge plant, but also developing an indigenous design?

        I don’t know if you are aware, but AREVA have designed in an option for EPR fuel to be enriched to 4%+ – it’s done alongside doping with burnable poisons to extend life between refuellings, and running to notably higher burn-ups. Again, I’m not privy to the numbers, but that suggests strongly that those enrichment levels are commercially viable.

        I’m not sure neutron flux is too big an issue – I’d need to check, but I’m not sure that neutron flux at the fuel surface is notably lower in an AGR than an LWR.

        We needed very high levels of fuel integrity in the AGRs – they’re slightly leaky (the sheer size of the damned things, and the number of penetrations into the PV make total sealing near impossible). They lose something like 1-2 tonnes/week of coolant to atmosphere, so significant levels of FPs in the coolant gas weren’t acceptable. Agreed that there’d be some work to do re water chemistry, etc, but these are material routinely used in boilers etc. so I suspect a good knowledge base exists as a start.

        As an aside, on the AGRs and their leakiness…..I was working on Heysham II in the mid eighties. At that time, there was a huge industrial dispute running – our Mineworkers Union had been out on strike for months, protesting against pit closures and as you’ll appreciate the nuclear stations were seen as one of the major obstacles to them exerting enough pressure to win.

        I finally became convinced they were going to lose when I stood at the main gate, watching the pickets. They were trying to stop fuel shipments – and of course, we had enough loaded and on site for years of running. They were waving the CO2 tankers through. We had 1-2 months storage on site.

        • The enrichment limit at manufacturers is not a technical problem per se. The Enriching facilities can go significantly above 5%. It is a problem of handling on the site and licensing limits. As the enrichment goes up, geometry and material accumulations controls become more difficult. Going above 5% will require significant investment on the part of the manufactures to handle the powder.

          The use of burnable poisons in BWRs primarily gadolinia has been standard practice for over 30 years. This primarily helps control during the first cycle of operation and isn’t really a player in higher burn-ups, but rather longer cycle length. Currently fuel is loaded with Gadolinia up to about 8%, much above that, the ceramic properties become sufficiently degraded that the pellets don’t sinter well.

  3. Just thought from a non-engineer (just a physicist):

    When the explosion happened in Apollo 13, the ground controller was able to turn around and immediately get a checklist for what to do in an explosion. Why wasn’t there an equivalent for “OMG Earthquake! Tsunami alert!” at Fukushima? Had they never thought to plan? Had they never had a drill? Had they never said “how do we get this thing cold quick?”

    Looking at recent coverage in the NY Times, it seems like they lulled themselves to sleep (here). I guess if you convince yourself there will never be an accident, then you’ll never bother preparing for one.

    NASA learned that lesson in the Apollo 1 fire — “a failure of human imagination”. Let’s not have any more.

    • This would be a discussion for next week. And the answer is that, at least in the U.S., there are such checklists called Severe Accident Management Guidelines or SAMGs.

      I’m not convinced that the NY Times article is an accurate depiction of the Japanese industry. Advertising safety to the public is not the same thing as believing that nothing needs to be done.

      • Agreed, advertising to the public is not the same as internal preparations. But, the engineers refused to take steps to deal with a catastrophe. No robots? No plan? No plan for backup power? Estimating the worst tsunami based on less than the whole geological record? That is a failure of imagination all around, not just for the public face.

        • As I said, there will be many more lessons as we have more information.

          You are again going after something other than TECHNICAL issues. This post was meant to discuss equipment failures and technical problems on the site.

          Having robots or adequately preparing are CORPORATE and POLITICAL considerations, by my definitions.

          I’m not sure it is reasonable to expect that TEPCO would have radiation hardened all terrain robots sitting in a warehouse waiting to be deployed. Frankly, I’m not sure such equipment was even available. TEPCO did request help too slowly, as I mention in the next post regarding corporate lessons learned. Drones and other equipment and expertise could have been available much earlier had they reached out.

          Earthquake and tsunami preparation was a political failure. In fact, NISA and other oversight bodies KNEW of the potential for greater events and failed to enforce upgrades. Should TEPCO have done them anyway? Perhaps, but if the regulator doesn’t think it is important, then the only reason to do it is to protect assets. Potentially expensive modifications for extremely low frequency events are hard to justify to shareholders. I discuss this in the next post as well.

    • That was mentioned at the American Nuclear Society plenary this morning. Could be very, very interesting. Problem is always length of development cycle. Stuff will have to run in test facilities to demonstrate robustness in reactor conditions up to the expected life of the fuel. Huge time and $$$ investments required. My guess is that you don’t see a new material re cladding for at least 5 and probably more like 10 years.

    • Now, that’s an interesting one – I’ve been developing a sneaking interest in some of the non-sodium LMFR designs, mostly lead or lead-bismuth cooled. That had prompted thoughts as to the use of ceramics and ceramic coatings for various components, obviating corrosion issues.

      After all, we routinely use ceramics for major components in the metal smelting industry.

Comments are closed.