Japan Earthquake: Nuclear Plants at Fukushima Daiichi

[jpquantin]

About unit 4 explosion breaking gate theory ...

This theory is interesting, but it relies on the assumption that the pool was quite empty at the time of the explosion, in order to explain that in the days after there were no signs of a dry pool, like for instance zirconium fire. I've even seen drawings with uncovered fuel assemblies at the time of the explosion.

Now looking at SFP4 losses rates, there are still missing pieces to me to explain such a point of departure.

Explosion happened Mar. 15th around 6 AM. If fuel assemblies were uncovered at this time, this means that all water above it would have boiled, leaked or splashed before the explosion. According to my calculation, this is about 870 tons of water. Without considering splashing, this means 217 tons losses per day. Losses of water due to splashing in SFP in other units (5 & 6) would help know the impact of it. Splashing may be one missing piece. Any information anyone?

Then: they put 1554 tons of water afterwards in order to reach the "pool is full" signal, which was first reached on Apr. 13 14:55. This is an average 46 tons per day loss. There are a few injections / spray for which we lack volumes (two helicopter drops, and one injection between Mar. 26 06:05 and Mar. 26 10:20). Even considering a big injection for the missing one (180 tons), and 2 drops of 20 tons each, one gets an average loss rate of 52 tons. Even splitting calculations between period with or without skimmer surge tank levels, you got roughly a rate about 50 - 60 tons per day loss.

So the situation would have been the following:
- prior to explosion, they suppose SFP4 had an average 217 tons per day loss
- after it, data only gives 50 to 60 tons per day.

Well, not only explosion saved SFP4 from a zirconium fire, and fuel melt, but it also stopped leaks apparently! How can you explain that? Rubbles filling the pool? But the pool is full of casks, and it may be very difficult for rubbles to go the places where there are leaks.

Moreover last video did not show that much rubble in the pool (attach1). Which seems weird also, as in another picture (attach2) shows SFP4 surrounded by a LOT of rubbles.

Next, looking at a http://www.youtube.com/watch?v=AUx7lIUsogA&NR=1", there is a view that is I think of the reactor wel (see circle in attach3). No steam above it, while lot of steam left of it (red / green circles in attach4), above SFP. We would expect at least some steam above reactor wel is gate would be broken, wouldn't we?

The gate broken theory is interesting, but it seems to contradict with other observations.

Pictures of unit 4 before and after explosion clearly qualify an hydrogen explosion. The key to me would be to estimate the amount of hydrogen with different water level drop rates, and see with this volume if one can explain the damages in unit 4. I've heard here and there that there were models to calculate hydrogen release from zirconium-steam reaction, maybe it would help qualify one or the other assumption. Maybe still covered assemblies, but surrounded by boiling water, can produce enough hydrogen to explain damages to unit 4? I don't know. I've seen a post on this question, quite interested in it.

Additionnally, we can also quantify the volume of water brought through the broken gate. It has to be significant in order to "save SFP4".

Well if the gate broke during the explosion it brought water that was above the reactor wel in the SFP. My own estimate of this volume is 1200 m^3, using this input:

- Top of RPV is at 31660
- Service floor is at 38900
- This is 7.24 meters height
- Assuming width identical to SFP, 12.2 meters
- Third length is 2 times distance between 2 pillars: 2 x 6.76 = 13.52 meters
- Estimated volume: 1200 m^3

Any remark anybody on this volume estimate?

NB: Some part of this providential water may have turned into steam due to the heat of the explosion, reducing its benefits...

 

[jpquantin]

Last attachement for previous post

 

[jlduh]

Maybe it's all just garbage readings.

That's a possibility that cannot be dismissed! That's why i mentionned several weeks ago that doing some "scientific analysis" on datas that are not confirmed reliable (to say the least) or that are coming from sources which are not scientifically neutral (Tepco for example to say the least) is more "educational guess" than science anyway!

But let's live with this without forgetting about it, you are right!

 

[fluutekies]

[...]
(Several days ago I tweeted @tep_co about the CAMS drop and got back a reply that they would look into it "later".

Please be advised that @tep_co is some sarcastic parody...

 

[artax]

There's a new update at all things nuclear.

http://allthingsnuclear.org/

 

[jlduh]

About unit 4 explosion breaking gate theory ...

This theory is interesting, but it relies on the assumption that the pool was quite empty at the time of the explosion, in order to explain that in the days after there were no signs of a dry pool, like for instance zirconium fire. I've even seen drawings with uncovered fuel assemblies at the time of the explosion.

Now looking at SFP4 losses rates, there are still missing pieces to me to explain such a point of departure.

Explosion happened Mar. 15th around 6 AM. If fuel assemblies were uncovered at this time, this means that all water above it would have boiled, leaked or splashed before the explosion. According to my calculation, this is about 870 tons of water. Without considering splashing, this means 217 tons losses per day. Losses of water due to splashing in SFP in other units (5 & 6) would help know the impact of it. Splashing may be one missing piece. Any information anyone?

Then: they put 1554 tons of water afterwards in order to reach the "pool is full" signal, which was first reached on Apr. 13 14:55. This is an average 46 tons per day loss. There are a few injections / spray for which we lack volumes (two helicopter drops, and one injection between Mar. 26 06:05 and Mar. 26 10:20). Even considering a big injection for the missing one (180 tons), and 2 drops of 20 tons each, one gets an average loss rate of 52 tons. Even splitting calculations between period with or without skimmer surge tank levels, you got roughly a rate about 50 - 60 tons per day loss.

So the situation would have been the following:
- prior to explosion, they suppose SFP4 had an average 217 tons per day loss
- after it, data only gives 50 to 60 tons per day.

Well, not only explosion saved SFP4 from a zirconium fire, and fuel melt, but it also stopped leaks apparently! How can you explain that? Rubbles filling the pool? But the pool is full of casks, and it may be very difficult for rubbles to go the places where there are leaks.

Moreover last video did not show that much rubble in the pool (attach1). Which seems weird also, as in another picture (attach2) shows SFP4 surrounded by a LOT of rubbles.

Next, looking at a http://www.youtube.com/watch?v=AUx7lIUsogA&NR=1", there is a view that is I think of the reactor wel (see circle in attach3). No steam above it, while lot of steam left of it (red / green circles in attach4), above SFP. We would expect at least some steam above reactor wel is gate would be broken, wouldn't we?

The gate broken theory is interesting, but it seems to contradict with other observations.

Pictures of unit 4 before and after explosion clearly qualify an hydrogen explosion. The key to me would be to estimate the amount of hydrogen with different water level drop rates, and see with this volume if one can explain the damages in unit 4. I've heard here and there that there were models to calculate hydrogen release from zirconium-steam reaction, maybe it would help qualify one or the other assumption. Maybe still covered assemblies, but surrounded by boiling water, can produce enough hydrogen to explain damages to unit 4? I don't know. I've seen a post on this question, quite interested in it.

Additionnally, we can also quantify the volume of water brought through the broken gate. It has to be significant in order to "save SFP4".

Well if the gate broke during the explosion it brought water that was above the reactor wel in the SFP. My own estimate of this volume is 1200 m^3, using this input:

- Top of RPV is at 31660
- Service floor is at 38900
- This is 7.24 meters height
- Assuming width identical to SFP, 12.2 meters
- Third length is 2 times distance between 2 pillars: 2 x 6.76 = 13.52 meters
- Estimated volume: 1200 m^3

Any remark anybody on this volume estimate?

NB: Some part of this providential water may have turned into steam due to the heat of the explosion, reducing its benefits...

Your explanations are interesting, especially the significant difference in loss rate of water before and after the explosion at N°4.

I've been surprised also to see the underwater images at the bottom of the pool (only a small part of it though) with very few debris indeed. That's not what I would imagine after such destructions around, but Tepco may have selected a clean spot!

 

[GJBRKS]

That's true. And the internal reactor parameters don't all seem to point the same direction, either. (#3 temperature goes up, but drywell radiation readings continue down. #2 radiation reading goes up in one monitor, but not in the others nor in temperatures...)

Maybe it's all just garbage readings.

In my view ,garbage readings do not follow real world dynamics.
I see some correlation , and I do not consider this a closed system , so there will be a lot of interconnected and unknown differentials.

But of course not all is well

 

[jlduh]

There's a new update at all things nuclear.

http://allthingsnuclear.org/

Thanks, but nothing really new i think.

But i saw just below an interesting article on SFPs, which describes the real flaws (and even gaps!) in the safety issues and scenarios retained for assessing them by NRC.

http://allthingsnuclear.org/post/4814761753/susquehanna-spent-fuel-pool-concerns-and-how-i-ended

Lochbaum is refererencing excerpts of his book already posted here:

http://www.ucsusa.org/assets/documents/nuclear_power/nuclear-waste-disposal-crisis-excerpts.pdf

 

[SteveElbows]

Latest updates mention 58t of water added to number 2 pool today, and a hefty 270t added to unit 4 pool on the 5th, think that might be the most they have stuck into that pool in one go.

 

[bytepirate]

130 Sv/h in no 2 torus - that is quite a lot?

How can that be?

it can not...
according to the graphs here: http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110427e19.pdf
130Sv/h would mean a core damage rate of above 100%

but as the readings of A and B differ by magnitudes, i would assume, that the readings are almost meaningless.
as long as we don't know, what exactly these monitors *see*, we simply don't know what these readings mean.

 

[Jorge Stolfi]

Here is an annotated version of the Air Photo Service snapshot of reactor #3, from nearly above:

[PLAIN]http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/foto/edited/out/reactor3-Z-3-c-A-i.png

To the best of my knowledge:

(A) Outline of the service floor, out to the outer surface of the building.
(B) Stairwell (?).
(C) Elevator shaft.
(D) Spent-fuel pool. The North edge is guessed, the East edge may be a bit off.
(E) Steam-dryer storage pool (= equipment pool). The South edge is guessed.
(F) Overhead crane (outline of upper surface).
(G) Winch box of the overhead crane.
(H) E-W and N-S cuts through the reactor axis (as in blueprints).
(I) Estimated broken edge of service floor slab.
(J) Grappling attachment for the containment cap?

Now, where is the Fuel Handling Machine? In orbit perhaps?

The tracks of the FHM allow it to travel only between the spent-fuel pool and the reactor opening. Perhaps it was over the reactor, and was squashed into the refueling pit by the falling crane? Were the concrete plugs in place at the time of the explosion?

Higher resolution images:
http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/good/un3_foto_exploded_Z_1.png
http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/good/un3_foto_exploded_Z_2.png
http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/good/un3_foto_exploded_Z_3.png
http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/foto/edited/out/reactor3-Z-3-c-A-e.png (annotated version of the latter)

 

[jim hardy]

piet

"A chemical explosion could have triggered prompt criticality."

can you explain the mechanism for that?

 

[artax]

piet

"A chemical explosion could have triggered prompt criticality."

can you explain the mechanism for that?

Well I assume they're considering the possibility that the chemical explosion changed the geometry of the nuclear fuel in such a way that the number of neutrons escaping was less or close to that causing fission.

The first atomic bombs used a chemical explosion to compress a 'sub critical' sphere of fissile material into a critical one of same mass but slightly samaller volume.

The following is from wikipedia

TEPCO claimed that there was a small but non-zero probability that the exposed fuel assemblies could reach criticality.[243][244] The BBC commented that criticality would never mean a nuclear explosion, but could cause a sustained release of radioactive materials.[243] Criticality is usually considered highly unlikely due to the low enrichment level used in light water reactors.[245][246][247]

There was, however, speculation on Russia Today by low-dose radiation researcher and anti-nuclear activist Christopher Busby that the explosion that destroyed the reactor 4 building was a "nuclear explosion" of some kind in the spent fuel pool.[248] Similarly, as noted above, Arnie Gundersen surmised a prompt criticality for the 13 Mar 2011 explosion at the spent fuel pool located on top of the Reactor-3 (see Reactor-3 above). [249]

Visual inspection of the spent fuel rod pool on reactor 4 on April 30 has however shown that that there is no significant visible damage to the fuel rods in the pool. This observation is inconsistent with speculation of prompt criticality

 

[jlduh]

piet

"A chemical explosion could have triggered prompt criticality."

can you explain the mechanism for that?

See an hypothesis (without any calculation though) in this video that I posted in this previous message:

https://www.physicsforums.com/showpost.php?p=3284941&postcount=5890

As i understand it the principle would be the same than the one for the first Nuke bombs (despite with far than "optimum" conditions resulting in a low energy nuclear explosion).

Have not enough knowledge to assess if this is a stupid idea or a possible one.

See also these wiki:
http://en.wikipedia.org/wiki/Prompt_critical#Critical_versus_prompt-critical
http://en.wikipedia.org/wiki/Critical_mass

Gunderson says that the answer is probably is the Xe data which could indicate the occurence of this phenomenon (don't ask me how).

Which Xe data do we have on hand for Daichi?

 

[elektrownik]

it can not...
according to the graphs here: http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110427e19.pdf
130Sv/h would mean a core damage rate of above 100%

but as the readings of A and B differ by magnitudes, i would assume, that the readings are almost meaningless.
as long as we don't know, what exactly these monitors *see*, we simply don't know what these readings mean.

I saw table, but I can't find it now with information how long reactors were without water, tahere was: 2,5h for 1&3 and 12,5h for 2 (because they were not able to open some valve) and then additional 50min after one of aftershocks, I think that 12,5h is enought to melt 100% of core ?

 

[jlduh]

I saw table, but I can't find it now with information how long reactors were without water, tahere was: 2,5h for 1&3 and 12,5h for 2 (because they were not able to open some valve) and then additional 50min after one of aftershocks, I think that 12,5h is enought to melt 100% of core ?

Don't know if you read this complete document listed yesterday (169 pages!) but to date this is by far the best documented and complete document on the chronology of the events for each reactor, with precise timing indicated. Have a look at some of the listings pages with times and events that this guy reconstructed, you'll find maybe something useful for your question. BUT, it's worth the reading for anybody on this thread, and also for new members or readers who didn't follow the stuff from scratch.

http://www.galcit.caltech.edu/~jeshep/fukushima/ShepherdFukushima30April2011.pdf

But for the "veterans" of this thead it's also a very good read, very informative on related subjects, and excellent to recap the bits of this complex puzzle (even if the guys is cautionning people that anyway, thuth an reality will only have a chance to be partially or fully understood in several years!)

I cannot put it in attachment to invite you to download it because of its size (11 Mo!)!

 

[jlduh]

A Japanese panel of seismologists knew about the risk of a major tsunami in northeastern Japan before disaster struck on March 11th. The group had put together a report in February and was planning to release it in April".

[...]

Associate Professor Yoshinobu Tsuji at the University of Tokyo's Earthquake Research Institute says the committee needs to further review past disasters to reassess future risks.

http://www3.nhk.or.jp/daily/english/06_27.html

 

[default.user]

I saw table, but I can't find it now with information how long reactors were without water, tahere was: 2,5h for 1&3 and 12,5h for 2 (because they were not able to open some valve) and then additional 50min after one of aftershocks, I think that 12,5h is enought to melt 100% of core ?

On 28 March 1979 at 4:36 clock time closed while work on the condensate polishing plant by a malfunction of the pneumatic control valve in supply line from the condenser to the two main feed pumps in the secondary circuit

german orginal:

Am 28. März 1979 um 4:36 Uhr Ortszeit schloss während Arbeiten an der Kondensatreinigungsanlage durch eine Fehlfunktion der pneumatischen Steuerung ein Ventil in der Speiseleitung vom Kondensator zu den beiden Hauptspeisepumpen I am sekundären Kreislauf

Since the beginning of the incident 165 minutes had passed when radioactively contaminated water reached the sensors. At this time, the radioactivity in the primary cooling circuit 300 times higher than expected: The meltdown was in full swing.

The operators in the control room was not aware for a long time, as little water as the primary cooling circuit still contained. About three and a half hours after the incident began to recognize the scope hurrying experts - new water was pumped into the primary circuit. Later it was found that fused together for roughly half of the inventory and a supercritical state was narrowly avoided.

german orginal:

Seit Beginn des Störfalls waren 165 Minuten vergangen, als radioaktiv kontaminiertes Wasser die Sensoren erreichte. Zu diesem Zeitpunkt war die Radioaktivität I am primären Kühlkreislauf 300-mal höher als erwartet: Die Kernschmelze war in vollem Gang.

Den Bedienern I am Kontrollraum war lange Zeit nicht bewusst, wie wenig Wasser der primäre Kühlkreislauf noch enthielt. Ungefähr dreieinhalb Stunden nach Beginn des Störfalls begannen die herbeigeeilten Fachleute die Tragweite zu erkennen – neues Wasser wurde in den Primärkreis gepumpt. Später wurde festgestellt, dass bereits etwa die Hälfte des Inventars zusammengeschmolzen und ein überkritischer Zustand nur knapp vermieden wurde.

It had been almost 16 hours, when the pumps were turned on again in the primary circuit and the core temperature began to fall.

german orginal:

Es waren fast 16 Stunden vergangen, als die Pumpen I am Primärkreislauf wieder eingeschaltet wurden und die Kerntemperatur zu fallen begann

http://de.wikipedia.org/wiki/Kernkraftwerk_Three_Mile_Island#Unfallhergang

http://en.wikipedia.org/wiki/Three_Mile_Island_accident

http://en.wikipedia.org/wiki/Fukushima_Daiichi_Nuclear_Power_Plant#Reactor_data

Will certainly have to consider the difference in the types of reactor.
But we can see that could melt within 10 to 16 hours the entire inventory.

It may help you.

kind regards

 

[elektrownik]

Thanks !, I was searching for this, also this 169 pages presentation from few post up give 17h without water, I think that it is enought to full melt of core, unit 1 was without cooling only 2,5h and they give 70% core damage for it, so if #2 was 17h without cooling...

 

[Samy24]

Maybe "they" have learned the lession?

http://www3.nhk.or.jp/daily/english/06_27.html"
http://www3.nhk.or.jp/daily/english/06_31.html"

 

[Samy24]

Thanks !, I was searching for this, also this 169 pages presentation from few post up give 17h without water, I think that it is enought to full melt of core, unit 1 was without cooling only 2,5h and they give 70% core damage for it, so if #2 was 17h without cooling...

The first time I read this document I believed the cores could have been melted completely.
Time without water and CAMS radiation readings point straight to this.
(I'm not a nuclear engineer.)
http://www-pub.iaea.org/MTCD/publications/PDF/te_955_prn.pdf"

 

[Rive]

... so if #2 was 17h without cooling...

Be careful, TMI2 was on full power when the water stopped. Fukushima was ~1 hour after SCRAM when the diesel lost. During that one hour the initial ~10% remanent thermal power was down to ~2%. And even after the diesel gone there was some partial cooling for some time.

Ps.: and fuel damage is not meltdown. It's far from that.

 

[jlduh]

Thanks !, I was searching for this, also this 169 pages presentation from few post up give 17h without water, I think that it is enought to full melt of core, unit 1 was without cooling only 2,5h and they give 70% core damage for it, so if #2 was 17h without cooling...

By the way Elektrownik did you manage to extract the time without water at N°3?

If time without water was 2,5h at N°1 and 17h at N°2, how can it be that Tepco is giving a much higher percentage of damage on N°1 (70%) than on the two others? I know from where they are getting these numbers (froms CAMS right?) but i mean, how can it be correlated with these times without water? Of course the height of exposed fuel rods is also a factor (without water means without any new water injected to compensate for steam conversion) but still this is strange, no?

Concerning the "damaged versus melted" difference, i understand it of course, but still, if it stays uncovered for a long time (and we know that, if the readings of water level inside the cores are working, which is NOT sure i admit, it stayed like that for weeks now at level like -2000mm below top fuel level for example at N°3), it is going either to melt, or to break (before melting) and fell into bottom water at some point. But if it breaks and fell but is blocked by other debris or bent stuff inside the core before reaching water level, the pellets are going to melt, don't you think?

So anyway, after a so long time (weeks), everything that is (or was!) higher than the water level has been "damaged" and relocated somewhere at a lower level inside the RPV, right? Damaged means either broken and relocated before melted, or broken, melted and relocated further down.

Am I wrong?

 

[yakiniku]

Don't know if you read this complete document listed yesterday (169 pages!) but to date this is by far the best documented and complete document on the chronology of the events for each reactor, with precise timing indicated.

Great link. I have only just started making my way through it. Thought I would add some anecdotal information that the WSJ wrote regarding pressure limits:

"Containment vessels can withstand higher pressures, some studies have indicated. Among these are studies conducted in the 1990s by Japanese operators and equipment manufacturers, in preparation for Japan's first set of severe-accident protocols, that say such vessels can withstand twice the design pressure. Many Japanese operators have adopted this as their benchmark for releasing contaminated air.

Tepco spokesman Yoshikazu Nagai confirmed that if there is a risk of releasing radiation, the company doesn't vent until pressure hits roughly twice the design limit. "Venting is a last resort," Mr. Nagai said.

General Electric Co., the designer of the vessel at Fukushima Daiichi, said it is unaware of any such Japanese studies or venting protocols."

http://online.wsj.com/article/SB10001424052748703922504576273234110896182.html

Edit: This is regarding the pressure limits mentioned on page 10 in the presentation

 

[Dmytry]

Don't forget to double-check all third party facts or compilations.

I've seen several articles that simply asserted gross untruths. This for example:
http://atomicinsights.com/2011/04/fukushima-nuclear-accident-exceptional-summary-by-murray-e-miles.html
"Something nearly miraculous occurred to prevent units 1, 2, and 3 from early catastrophic failure. About eight hours after the earthquake Tokyo Electric Company made the decision to pump seawater into the reactors and reactor containment buildings of units 1, 2, 3, and 4."
which is patently false; the quake did strike at march 11th, 14:46 ; according to NISA, http://www.nisa.meti.go.jp/english/files/en20110317-1.pdf
the water injection has started at:
unit 1: march 13th, 11:55 (seawater) (or quake+45 hours 10 minutes)
unit 2: march 13th, 14:00 (seawater)
unit 3: march 13th, 11:55 (freshwater, followed by seawater shortly thereafter)
unit 4: no fuel in reactor

On topic of venting: I'm pretty sure that's wrong. You can't vent safely at beyond the design pressure. Venting at such pressure is probably how they managed to explode 3 reactors out of 3 (the #2 less impressively though than the other two). Overpressure beyond limit is known to cause leakage of the vented gasses (steam, hydrogen) into the building, hence the explosions; in #3 there is a lot of reasons to suspect the lid of the containment vessel was lifted a bit briefly, leading to explosive venting and massive mushroom cloud 'explosion' in the hollywood style.

 

[jpquantin]

The reaction is 2 H₂O + Zr --> ZrO₂ + 4 H₂. For your high end estimate, making 1e+06g of H₂ requires (1e+06g/2g/mole)/4*91g/mole = 11e+06g Zr. This seems like a lot of Zr available for oxidation since only the surface Zr is immediately available, if I understand the chemistry correctly. The low end is probably too low because the escaping hydrogen will fall below 4% and won't explode, whereas there were explosions in the escaping material.

Could you tell us how much energy does one reaction would release ? Thanks.

 

[fluutekies]

Could you tell us how much energy does one reaction would release ? Thanks.

Zr + 2H₂O --> ZrO₂ + 2H₂ + 14.6*10⁶ J/kg Zr

Source: http://www.galcit.caltech.edu/~jeshep/fukushima/ShepherdFukushima30April2011.pdf page 62

BTW: Great source!

Edit: another source http://oa.upm.es/718/1/Miguel_Angel_Jimenez_Garcia.pdf dH = 576 kJ/mol Zr

My Spanish is not good, but it seems that this PhD thesis is worth reading ...

 

[kerkko]

Something seems weard here.
Lower floors must have had an implosion instead of the explosion.
From 30 sec on at the video. I mean there is no way that the powder you see in video could have stayed there if there was that strong explosion, it would have blown out that debris.

 

[elektrownik]

From data which I saw unit 3 was 2,5h without water (#1 also 2,5h). In case of #3 explosion I will stay with my opinion: hydrogen explosion in SFP and then explosion in core or drywell, if you look on explosion video you will see that the second big explosion is from core location, not from SFP...

 

[kerkko]

Something seems weard here.
Lower floors must have had an implosion instead of the explosion.
From 30 sec on at the video. I mean there is no way that the powder you see in video could have stayed there if there was that strong explosion, it would have blown out that debris.

Im not an physics expert or anything, my field of work is demolitions, but i have seen enough to tell from that video that the explosion in plant 3 was not normal..
I think that the top of the reactor blew up, sucking air to lower levels causing an implosion at the lower levels, creating a pipelike effect.