Japan Earthquake: Nuclear Plants at Fukushima Daiichi

[Azby]

Having 'partially damage' (light green) above significant melting (brown) is rather odd/strange/seemingly unlikely. Molten fuel would not support any solid structure. Anything above the melting would likely be severely damaged.

I'm imagining that the newsaper's graphic designer had access to a more technical representation, and got some details wrong. I wonder if anyone can find the TEPCO files this diagram was based on?

 

[zapperzero]

I'm imagining that the newsaper's graphic designer had access to a more technical representation, and got some details wrong. I wonder if anyone can find the TEPCO files this diagram was based on?

Well, it could be that the colors actually represent temperature ranges and the chart just gives those temp ranges in terms of how damaged a notional fuel rod would get upon being heated to somewhere in that temp range.

 

[SteveElbows]

I'm imagining that the newsaper's graphic designer had access to a more technical representation, and got some details wrong. I wonder if anyone can find the TEPCO files this diagram was based on?

The TEPCO originals for reactor 2 & 3 are shown within this document:

http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110524e14.pdf

 

[tsutsuji]

The following is dated April 21st, so it might have already been posted here : Kanazawa University and an Akita prefecture based company called Kumaken Kougyou claim they have invented a process to treat 1000 tons of water per hour, which is 20 times more than Areva's 50 tons per hour : http://blogs.wsj.com/japanrealtime/2011/04/21/chemist-i-can-clean-fukushima-water-faster/

A May 22 issue of a Colorado local newspaper reports about the involvement of a company called "Industrial Constructors/Managers" :

The Pueblo company is coordinating the production of 97 steel containment units for the radioactive water at Tokyo Electric Power Co.'s stricken nuclear plant in Fukushima, Japan.
http://www.chieftain.com/business/local/answering-japan-s-call-for-help/article_a549f206-841b-11e0-abba-001cc4c002e0.html

 

[dh87]

I don't know if that fully answers your question, but page 14 (15/41) https://www.sfen.fr/content/download/30419/1606950/file/2-ICAPP Fukushima Accident_Masui.pdf speaks about a "desalination system" producing "concentrated seawater" as ultimate waste, being installed at the end of the water decontamination unit.

The diagram shows a "Cs adsorption tower" before the reverse osmosis treatment. My understanding was that Areva was talking about a selective precipitation procedure. For Cs+ removal to be effective, the selectivity has to be enormous. The Na+ and K+ concentrations exceed the Cs+ concentration by many orders of magnitude. This couldn't be a simple ion-exchange resin. In addition, the recent published data show significant amounts of Sr-89 and Sr-90. In the diagram you reference, these radioisotopes will end up in the concentrated seawater.

I don't know much about reverse osmosis, but doesn't it produce a small amount of purified water and a large amount of slightly concentrated starting material? Presumably, that's why the diagram shows evaporation of the reverse osmosis concentrate. But if you're going to evaporate anyway, why not just dispense with everything else?

 

[tsutsuji]

The diagram shows a "Cs adsorption tower" before the reverse osmosis treatment. My understanding was that Areva was talking about a selective precipitation procedure. For Cs+ removal to be effective, the selectivity has to be enormous. The Na+ and K+ concentrations exceed the Cs+ concentration by many orders of magnitude. This couldn't be a simple ion-exchange resin. In addition, the recent published data show significant amounts of Sr-89 and Sr-90. In the diagram you reference, these radioisotopes will end up in the concentrated seawater.

I don't know much about reverse osmosis, but doesn't it produce a small amount of purified water and a large amount of slightly concentrated starting material? Presumably, that's why the diagram shows evaporation of the reverse osmosis concentrate. But if you're going to evaporate anyway, why not just dispense with everything else?

See also the somewhat more detailed diagram page 10 of http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110517e5.pdf

 

[NUCENG]

Ahh. But this is the very crux of the matter, isn't it? TEPCO proposed and the gov't approved a modification of the design basis numbers. Were they right, or wrong to do so?

There are now two design basis numbers for Fukushima. One for inland earthquakes and one for the type of earthquake at sea like that on March 11. The 600 Gal number is for the latter. TEPCO had completed reanalysis and mods (unknown what they did) for the inland earthquakes, but had not finished evaluations for the 600 Gal limit.

See:
https://www.physicsforums.com/showthread.php?p=3289365&highlight=seismic#post3289365
https://www.physicsforums.com/showthread.php?p=3287872&highlight=seismic#post3287872

 

[tsutsuji]

extract from http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110524e13.pdf" showing that the isolation condensers where shut down 11 minutes after automatic start up with a note justifying the reason for this intervention ( we presume that this is a manual operation)
[PLAIN]http://k.min.us/i2Dqk.JPG[/QUOTE]

I guess that the closing of 3A and 3B are what is shown with red dots on pages 32 and 33 of http://www.tepco.co.jp/cc/press/betu11_j/images/110524a.pdf .

Page 19 (attachment 1-11) shows the "PLR pump entrance temperature" rising again at the point marked with arrow ③, which is when the isolation condenser stops.

See also pressure rising at 15:03 on the graph in the middle of page 24.

The TEPCO originals for reactor 2 & 3 are shown within this document:

http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110524e14.pdf

Some similar-looking diagrams are shown on http://www.tepco.co.jp/cc/press/betu11_j/images/110524a.pdf

page 160 for unit 1
page 173 and 183 for unit 2
page 195 and 206 for unit 3

 

[NUCENG]

As Reuters puts it :


But http://www.chunichi.co.jp/s/article/2011051790014812.html says that while Tepco is emphasizing that the shutdown of the isolation condenser by a worker, in accordance with the manual which instructs to do so when reactor temperature falls by more than 55°C per hour, is one possibility, the other possibility is that the isolation condenser was damaged by the earthquake before the tsunami struck. Tepco says that the second possibility can't be ruled out.

Looking at the graphs for F1-1 at:
http://www.tepco.co.jp/nu/fukushima-np/index10-j.html

The Isolation Condenser was initiated from 1452 hr to 1503 hr as stated in the TEPCO report.
In that time the RPV pressure dropped from 6.8 MPa to 5 MPa. Using steam tables for saturated steam temperatures at those pressures the cooldown rate if allowed to continue was about 212 degC/hr. In fact, in the condition they were in, they should have continued cooling down and ignored the cooldown rate. That rate is for normal shutdown conditions to reduce stress cycling of the vessel and nozzles. If they deliberately isolated the IC, it indicates their training was deficient and they were blindly following normal cooldown limits when they were in a non-normal condition.

Once they secured flow through the IC, there is a potential that noncondensible gases can (eg., hydrogen, amonia, fission product gases) collect in high points in the IC piping. This can air-bind the system defeating natural circulation which is the driving force for this system. That is why they are uncertain if it worked when they later tried to restart the IC.

I see no indication of leaks or damage to the IC system or piping during the time prior to the tsunami covered by the TEPCO data at the link above. It was holding pressure, so I am inclined to believe the isolation was deliberate by operators.

TEPCO is apparently being careful in stating it may have been damaged in the earthquake, but during the few minutes it operated there is normal system response. If they suspected damage to the system they should have isolated the 4 normally open supply valves from the reactor vessel steam area to the IC system shown in the diagram in the TEPCO report, but they haven't indicated that in any of the releases I've seen.

 

[Rive]

Looking at the graphs for F1-1 at:
http://www.tepco.co.jp/nu/fukushima-np/index10-j.html

The Isolation Condenser was initiated from 1452 hr to 1503 hr as stated in the TEPCO report.
In that time the RPV pressure dropped from 6.8 MPa to 5 MPa. Using steam tables for saturated steam temperatures at those pressures the cooldown rate if allowed to continue was about 212 degC/hr. In fact, in the condition they were in, they should have continued cooling down and ignored the cooldown rate. That rate is for normal shutdown conditions to reduce stress cycling of the vessel and nozzles. If they deliberately isolated the IC, it indicates their training was deficient and they were blindly following normal cooldown limits when they were in a non-normal condition.

Even without knowing about the tsunami on way?

 

[AntonL]

Breaches in containment vessels 1 & 2 before explosion
http://search.japantimes.co.jp/cgi-bin/nn20110526a1.html

Another analysis by Tepco has shown that breaches may have occurred at containment vessels encasing reactors 1 and 2 at the power plant, possibly causing leaks of highly radioactive water there.

The possible breaches to the containment vessels there are certain to compound efforts to deal with accumulating contaminated water at the sites, raising questions about the viability of a Tepco plan to re-establish a stable cooling system by around mid-July.

Tepco said if it hypothesizes that a breach of about 3 cm wide occurred at the reactor 1 containment vessel 18 hours after the quake and widened to about 7 cm 50 hours later, that corresponds well to changes in pressure readings inside the containment vessel.

The utility also hypothesized that a breach roughly 10 cm wide occurred at the No. 2 reactor's containment vessel 21 hours after the quake due to elevated temperatures, among other factors.

This finding also corresponds with data obtained.

Tepco also said it believes that parts used to ensure air tightness may have broken from overheating

explosion unit 1: 3:36PM March 12 = 25 hours after earthquake but containment breached 18 hours after the quake

explosion unit 2: 6:14AM March 15 = 88 hours after earthquake but containment breached 21 hours after the quake

Elevated temperature should not be a ground for the breach in containment - surely it is designed to withstand such temperatures. Has Tepco made this statement as as defence against possible damage by earthquake, which would imply that all containment of NPP of similar designs are at extreme danger of failing by a stronger quake or they already have failed but not yet tested or noticed as no elevated pressure in containment under normal operating or shutdown condition.

Are there any other references to this report?

 

[NUCENG]

Even without knowing about the tsunami on way?

Oops, that is right! Great point. They had cooling and AC power and expected to be able to control cooldown. The operators may have done what they thought was right based on what they knew at 1503 hr. It can be tricky to keep hindsight out of our evaluations. After they lost All AC after the tsunami, they would have been dealing with all sorts of lost instrumentation and trying to figure out what was going on. They may not have thought about restoring the IC to prevent air-binding the IC system.

 

[bytepirate]

...I am inclined to believe the isolation was deliberate by operators.

how can one explain, that TEPCO does not *know* what the operators did?

this draws a picture in my head: panicing people running around pushing random buttons...

i am pretty sure, that it has not been like this, there should not be more than one or two people, that could have pushed that button (or issued the order to do so).

how about asking them?

 

[Marita]

Even without knowing about the tsunami on way?

Yep it's real easy for us to sit here without all the alarms screaming and lights flashing after the whole building was shaking for 5 minutes and say "what were they thinking" why didn't they just do x y or z. Been in those situations and sometimes it is really hard to sort out the the highest priority emergency you need to handle "Right Now" and then your incoming information is hidden by a power loss , I can only imagine the "pucker factor" those operator experienced.

 

[NUCENG]

Breaches in containment vessels 1 & 2 before explosion
http://search.japantimes.co.jp/cgi-bin/nn20110526a1.html



explosion unit 1: 3:36PM March 12 = 25 hours after earthquake but containment breached 18 hours after the quake

explosion unit 2: 6:14AM March 15 = 88 hours after earthquake but containment breached 21 hours after the quake

Elevated temperature should not be a ground for the breach in containment - surely it is designed to withstand such temperatures.

Are there any other references to this report?

It wasn't just temperature. Unit 1 pressure was more than double the pressure limit and containment would have been expected to start leaking at pressures as low as 0.18 MPa. At 0.81 MPa the predicted leakage area is 5.42 cm^2 per: (units converted to metric)

http://www.osti.gov/bridge/servlets/purl/5630475-EX87x5/5630475.pdf

If containment temperatures exceeded 260 degC to 371 degC other seals and containment penetrations are also expected to begin failing. The typical design basis limits for containment are about 140 degC and 0.39 MPa. Once the plants at Fukushima lost all decay heat removal, containment cooling, and fuel pool cooling in an extended station blackout the inevitable results were what we see now.

 

[NUCENG]

how can one explain, that TEPCO does not *know* what the operators did?

this draws a picture in my head: panicing people running around pushing random buttons...

i am pretty sure, that it has not been like this, there should not be more than one or two people, that could have pushed that button (or issued the order to do so).

how about asking them?

The graphs show that the IC was isolated at 1503 hr. The May 24 TEPCO report cited above says it was done to slow cooldown rate. Rive pointed out that based on what they knew at that time, their actions were probably justified. Everything I see in the data released to date is consistent with an operable IC system that was secured after a few minutes of normal operation. Other releases and news stories may indicate something different, but based on the discussion here it looks doubtful that the earthquake damaged the system. This will probably be a point covered in any investigations.

As to paniced operators pushing random buttons, I wasn't there, but as Marita pointed out there was likely a strong "pucker factor." Operator training and drills in simulators is pretty intense and they have a lot of practice at upset conditions. The fact that this was real was a bigger stress factor, I am certain. Have any news reports actually included interviews of the operators?

 

[WhoWee]

Oops, that is right! Great point. They had cooling and AC power and expected to be able to control cooldown. The operators may have done what they thought was right based on what they knew at 1503 hr. It can be tricky to keep hindsight out of our evaluations. After they lost All AC after the tsunami, they would have been dealing with all sorts of lost instrumentation and trying to figure out what was going on. They may not have thought about restoring the IC to prevent air-binding the IC system.

That's what I was thinking - they had to prioritize - deal with certainties first.

 

[NUCENG]

Having 'partially damage' (light green) above significant melting (brown) is rather odd/strange/seemingly unlikely. Molten fuel would not support any solid structure. Anything above the melting would likely be severely damaged.

I am picturing a similar natural phenomena - volcanic lava. In Hawaii you can walk on a crust over liquid magma. The crust acts to insulate the magma allowing it to remain liquid. In the reactor they were using core spray to inject into the vessel over the top of the core. This would promote forming a crust. Would that make sense for why a cooler area might be flloating on a hotter area?

 

[tsutsuji]

TEPCO is apparently being careful in stating it may have been damaged in the earthquake,

Because it is "more political", I am answering in the "more political" thread at : https://www.physicsforums.com/showthread.php?p=3322455#post3322455

 

[NUCENG]

Good you posted this again Anton:
http://search.japantimes.co.jp/cgi-bin/nn20110516a3.html

As here they are explaining (hopely some expert has made some calculations?) that rising pressure alone cannot explain the findings in unit 1:

It is very difficult to overpressurize the RPV itself due to the capacity of the Safety Relief Valves. These valves relieve pressure to tghe torus which thenh relieves back to the Drywell through the torus-DW vacuum breakers. Containment design pressure is about 0.39 MPa and can be reached fairly quickly in an extended SBO like occurred at Fukushima. At the same time the pressure is rising, the temperature is rising.

 

[NUCENG]

I believe the reactor (power generation) was shutdown manually just after the quake. The reactor cooling was working at the time. Normally, even at shutdown, the cooling continues.

That very well could be. In other words, the quake may have damage the core or cooling systems and/or containment.

The press/media are frequently not careful with wording.

We need to be careful with wording too but it can be tough. According to the logs the three operating reactors at Fukushima Daiichi all received automatic scram signals from the seismic sensors. The shutdown was NOT manual. In the case of Unit 1, the Isolation Condenser was initiated manually and probably isolated manually a few minutes later but before the tsunami.

I think these two distict events have been confused.

 

[NUCENG]

That very well could be. In other words, the quake may have damage the core or cooling systems and/or containment.

The press/media are frequently not careful with wording.

The possibility of radiation release prior to the tsunami is something I have been looking to confirm. There are numerous press reports of a single radiation sensor alarm about 1.5 km from Unit 1 that was received at 1529 hr about 15 minutes before the 14 m tsunami wave hit the site.

However if you look at the timeline and press releases you can find that the first tsunami waves hit the site at 1529 hr. Coincidence?

Why was only one sensor triggered? None of the graphs or logs published by TEPCO so far show any radiation levels or alarms prior to the tsunami. Many instruments in nuclear plants alarm on downscale or loss of power as well as at a setpoint to alert operators to instrument failures. It is possible that the sensor that alarmed was inundated by the lower waves that hit the site at 1529 hr.

 

[ThomS]

http://www.tepco.co.jp/en/press/corp-com/release/11052508-e.html

This is in regards to detection of radioactive materials, specifically Pu-238, 239, and 240.

What does this tell us? The report states that TEPCO tested up to 500m from the reactors, but should they be testing further away. I *assume* that these materials could travel as other materials such as cesium and iodine.

Is it significant that they are still finding it? I understand that Pu has a very long half-life, however, the second report (pdf) says that the values have NOT changed greatly.

Thank you for your time and thoughts.

 

[NUCENG]

I am quoting some posts from the political thread








and now http://www3.nhk.or.jp/daily/english/25_30.html


So it seems that the quake did cause some severe damage to the integrity of the plants

Also remember Unit 6 is collecting water in both the basements of the reactor and turbine buildings.

In conclusion - the NPP are not that earthquake safe as previously thought!

AntonL,

To some extent the following is playing devils advocate. Your interpretation is absolutely plausable.

However, another possibility requires consideration of the difference in design of the HPCI and RCIC systems. The design basis for HPCI is a small break LOCA that does not depressurioze the RPV. The RCIC system is sized for a Loss of Feedwater accident. In this event the MSIVs closed at the time of the SCRAM and there did not appear to be significant loss of coolant. This was more likely to use RCIC, which is the smaller of the two systems. However from the graphs for unit 3 neither HPCI nor RCIC was operated for some time after the SCRAM. There were multiple SRVs cycling losing water from the RPV to the suppression chamber. From the latest reports on 5/24 the RCIC system eventually was able to recover level in Unit 3. It is unclear why the HPCI system auto started on 3/12 but it is possible the RCIC system failed or was secured for some reason. (?) Anyway, when HPCI started it would have much higher flow rate and would also draw more steam. Together with a lower decay heat after 24 hours from the SCRAM, both of these effects would have reduced RPV pressure faster than with the RCIC system in operation.

The HPCI system uses 250 VDC batteries. RCIC uses 125 VDC. There are early press releases that indicate Unit 3 tried to restart the RCIC system but it failed to start. This could be due to 125 VDC battery depletion or that the steam pressure was too low to run the system.

This alternative sequence of events is possible without postulating earthquake damage. Again, this is not to say that your sequence is wrong, but it is not a sure thing.

 

[ThomS]

http://search.japantimes.co.jp/cgi-bin/nn20110526a1.html

Has this been discussed? Apologies if it has.

 

[Bodge]

http://search.japantimes.co.jp/cgi-bin/nn20110526a1.html

Has this been discussed? Apologies if it has.

It looks like an 'old school' BWR cannot take a direct hit from a 6-7 magnitude quake, tsunami or not.

 

[ThomS]

http://enenews.com/japan-switches-measurements-reactor-no-1-drywell-not-suppression-chamber-now-at-192-sieverts-per-hour

I can't find the original source for this. Can someone point me to it?

Also, (apologies for all the questions), assuming this is correct, why the big jump? Why measure from this point instead of the prior point?

 

[Bodge]

Astronuc, NUCENG, et. al.

What are we to understand from the 215 Sv/Hr readings in No.1's drywell?
[215 is the latest number from TEPCO's daily update]*

Does this indicate ongoing criticality or has a fragment of corium 'just' migrated nearer the sensor?

What should we expect to happen next if there is corium in the drywell, with or without ongoing fission?

Additionally, why is the pressure in the RPV edging ever upwards?

TIA

* see also: http://www.cbsnews.com/stories/2011/05/25/ap/business/main20065975.shtml

"In addition, some chunks of the fuel appeared to have entered the inner containment chambers, or drywell, causing some damage."

 

[LabratSR]

http://enenews.com/japan-switches-measurements-reactor-no-1-drywell-not-suppression-chamber-now-at-192-sieverts-per-hour

I can't find the original source for this. Can someone point me to it?

Also, (apologies for all the questions), assuming this is correct, why the big jump? Why measure from this point instead of the prior point?

I believe this is the original source for that and note status = Instrument Failure

http://atmc.jp/plant/rad/?n=1

This is TEPCO's Unit 1 parameter page and if you go al the way to the bottom under CAMS data you will see they also state Instrument Failure.

http://www.tepco.co.jp/en/nu/fukushima-np/f1/images/11052506_level_pr_data_1u-e.pdf

 

[Bodge]

From the UN:

"In addition, some chunks of the fuel appeared to have entered the inner containment chambers, or drywell, causing some damage."

Maybe sensor A is malfunctioning and sensor B is working OK?

Sensor B shows a history of 30-40 Sv/Hr readings that aren't shown at the atmc link.

Sensor A always shows a non-worrying level of radiation.

...when I look back at the reactor 1 drywell CAMS readings I noted in the past, I have just one sensors data for the date range march 20th->april 8th, and it was yoyoing around the 30-50 range over this time, tending to be towards the lower end of this range as time went on. Then it suddenly went up to 100 on the 8th, then back down to 68.3 later on the 8th april. Then no more data was published from unit 1 D/W CAMS all the way until 17th may where we get 2 sensors readings ever since, one of which is usually well below 1 and the other which darts around from either 25-35 or sometimes leaps to 200 or more.

Here is a snapshot of No.1 on the 23rd March:

0.5 Sv at Sensor A, 33 Sv at sensor B. Sensor B correlates much better with the sensors in No.2 and 3 drywells.

http://www.meti.go.jp/press/2011/05/20110523006/20110523006-3.pdf

 

[Azby]

The TEPCO originals for reactor 2 & 3 are shown within this document:

http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110524e14.pdf

Thanks. Seeing that the diagrams in that presentation are so similar to what's in the newspaper, I now get the feeling that they're based on a software simulation that uses a similarly simplified graphic image.

(Am I the ony one who has trouble with multi-quote? Using Firefox 3.5.7...)

 

[NUCENG]

Thanks. I had seen that data before, but I forgot to compare it to the original data that I had been recording manually from each status update.

By looking at both sets of data, it seems highly likely that the reason they originally stopped publishing unit 1 CAMS for the drywell after April 8th is not just because it spiked upwards and then wobbled all over the place, but because they got the 2nd sensor working at this time and it completely disagrees with what the other sensor was saying, by quite some orders of magnitude. I am not really sure why they started showing it on the overall status updates again as of may 17th, because the 2 sensors still disagree a lot.

Probably a self-defense strategy - the old "even a broken watch is accurate twice a day" strategy. If they weren't publishing the data someone would certainly accuse them of witholding vital data.

 

[robinson]

Have any news reports actually included interviews of the operators?

I saw two interviews on TV with people who were actually working at the plant, inside the plant at the time of the earthquake. It was pretty much get out any way you could, and run away. During the quake, which lasted quite a while. Eyewitness reports of large cracks in the reactor buildings, massive damage to the facility, one American there stated they pried open the safety doors to get out of the building.

Then later when the tsunami was coming everyone ran for high ground. Nobody stayed at the controls when the tsunami hit. This was on television, CNN I believe.

 

[ThomS]

I believe this is the original source for that and note status = Instrument Failure

http://atmc.jp/plant/rad/?n=1

This is TEPCO's Unit 1 parameter page and if you go al the way to the bottom under CAMS data you will see they also state Instrument Failure.

http://www.tepco.co.jp/en/nu/fukushima-np/f1/images/11052506_level_pr_data_1u-e.pdf

I appreciate it. But this really does not answer my question. Let me explain if I can. I see the note, but it doesn't *mean* anything to me. I see the numbers jumping around some over the last five days. So, does this mean that these high reading is incorrect or regarded as dubious at best? And should it be ignored? Honestly, I don't see how amtc came up with the numbers.

OTOH, did the meter fail on or after 5/25? Before?

Frankly, I don't know what to make of that headline, readings, etc. I am looking the amtc and enenews but it seems that it is either 1) sensationalist and not reliable, or 2) really scary.

Based on the data, really which one it it? This sort of headline scares and worries everyday people without technical backgrounds to dig into the guts of all those numbers. Remember that people here in Japan (including me) do want to understand and make informed decisions on reliably interpreted data.

Apologies if this came across a bit bluntly.

Thank you

 

[NUCENG]

That's pretty unbelievable... I really think that we will need some time to draw all the conclusions about the technical flaws enlightened by this tsunami. And I'm really convinced that whatever the disaster is already, to some extent, japanese and ALL OF US also have been very lucky when you consider all the parameters involved, with all the uncontrolled risks.

We talked a lot about the EDG and the electric blackout, but really i think the problems related to the pumps and more generally the loss of the cold source have been understated or perhaps not even fully revealed, and the design flaws are also to be considered in this department...

If the EDG can in a way be installed at a much higher level from sea level (if we want to secure them), the problem of the pumps is i think even more problematic to solve. By definition, a seawater pump has to be at sea level, at list the turbine part. Then, of course, the electrical part can be installed at a much higher level, with a long shaft linking the motor and the impeller. I'm not sure that in the Daichi plant, and even at Daini, this is sufficient to secure them though in case of tsunami.

We can also imagine that a so big wave rushing into the impellers can probably break them, or damage them to a certain extent. Then, even with electrical power left, the cold source is lost or diminished. We can also imagine that the intakes can be heavily obstructed by big debris. All these risks explain why the big wall with Y concrete blocks has been constructed into the sea, to create the port where the remaining waves are supposed to be small and the debris kept at a minimum. But a tsunami easily go over this wall... then the cold source is at risk.

The articles reveal an other domino effect with the cooling of the EDG being done with the pumps... which is kind of scary thinking about it! The backup for electrical power is dependent of the same cold source than the one used for the reactors, so no cold source leads to big reactors cooling problem AND electrical backup failure after maybe one hour.

Definitely, this problem related to the pumps and the possible loss of cold source is in my mind as inportant, and even more important perhaps, than the one related to the EDG. And it has not been discussed a lot in reality (here or in the medias), even if i think that at the very beginning, i heard from some reports that Daichi lost its cold source before losing its power (I mean the backup power).

Do we know more on this for the various reactors/plants? Which pumps did fail?

These pumps should at least be put in a bunker if they have to stay close to the sea...

At Daini, from the picture i posted, i don't see the main pumps being inside a building. This picture is from the 12th of March, one day after the tsunami.

US BWRs I am familiar with have pump houses to protect the cooling water pumps for General Service Water, Emergency Service Water and Circulating Water systems. These buildings are safety-related, sesmically qualified and water tight where to protect the pump motors. Impeller shafts drive the impellers through a water sealed fitting on the motor floor. Power cables can be elevatedor routed in water tight conduits. So it is possible to protect pumps from flooding damage even if they are located near the water edge.