Japan Earthquake: Nuclear Plants at Fukushima Daiichi

[SteveElbows]

I also don't understand why unit 4 sfp water temperature is always 84C and 62C for unit 3 sfp, this is not possible that there is no change... also why water level in unit 2 is so low, it looks like they can't fill it with water like in unit 4... it is ~4000 max, but usualy 2000-3000 and unit 4 is 6500 after injection

Just look at the dates - they take temperature of 3 & 4 via sensor attached to concrete pump, and they don't do it very often at all. The 84c is from may 7th and 62c for reactor 3 was measured on may 8th.

As for the water levels, these are the water levels of the skimmer surge pool, not the fuel pool itself. Same with temperature readings for reactor 2, its the skimmer water, which is why temp goes up when they inject water. When pool gets near top, water can move from pool to skimmer surge tank(s), and so we see skimmer level go back up and temperature also rises as hot pool water moves into skimmer tank.

Now its certainly true that when pool 2 is filled, skimmer level does not go as high as it used to. It goes up to about 4000mm these days, but when they used to fill it in april it went as high as 5000 or 6000mm, even higher than that at times. I can't tell you why, but there are likely a few explanations which are nothing to worry about.

 

[AntonL]

Staring soon -
this should be interesting if they zoom in
as a static view it will be boring

We can start the first discussion - is the nearest exhaust stack leaning or not?
I think the camera is not mounted horizontal [PLAIN]http://k.min.us/icHSBk.JPG

 

[tonio]

I have followed up on the potential for radiolysis and believe it is a plausible explanation for the Unit 4 explosion and damage. I am still looking at the possible scenario of Hydrogen gas from unit 3 through the SBGT system ducting.

1. I calculated decay heat for Unit 4 based on a full core offload 102 days before 3/11/2011 using rated thermal power of 2381 MWt and three batches of spent fuel aged 1 year (204 bundles), 2 years (204 bundles) and 3 years (171 bundles). I used the methodology of ANSI/ANS-5.1-1979. This calculation estimated decay heat in the Unit 4 SFP as 3.05 MW.
2. Using the various plant drawings posted on the forum I estimated spent fuel pool water volume of 1170 m^3 or 1.17E6 kg.
3. Assuming adiabatic heating and an initial temperature of 25 degC at the time of Loss of Spent Fuel Pool Cooling (15:45 3/11/2011) the estimated time to boiling was 33.5 hours or 3/13/2011 13:12 hr.
4. Using the ORIGEN2 analysis I have referenced before and the same core/batch assumptions used in #1 above I estimated total ionizing radiation in the spent fuel pool as 6.18E24 eV/sec. I determined that alpha and neutrons interactions are several orders of magnitude less than Beta and Gamma so these were ignored. The total used includes Bremsstrahlung radiation and x-rays.
5. Then assuming a production rate of 44 molecules of H2 per 1E4 eV of radiation from the Hydrogen Manual AntonL has referenced, I come up with a production rate of .0903 g of H2/sec. The method used in the Hydrogen Manual indicated it was probably a significant over-estimate of hydrogen generation and this calculation is about one tenth of that estimate. One note in the calculation used by AntonL is that the rates in the table he used are average rates over the entire production time, and need to be converted to instantaneous production reates which will also lower the production rate. I assumed that once pool boiling starts the H2 gas will be released from the pool at the production rate.
6. TEPCO does not know exactly when the damage to Unit 4 occurred, but damage was observed about 3/15/2011 at 06:00. This gives a maximum duration of boiling and hydrogen production.
7. Again using the calculated decay heat rate and latent heat of vaporization of water I calculated a boil-off rate of 1.35 kg/sec from the pool. It would have taken until 3/18/2011 to boil off half the pool volume. Therefore there was no fuel uncovered.
8. Using the discovery time of the Unit 4 building damage, a maximum of 16.9% of the pool volume could have boiled off.
9. In this same time a maximum hydrogen production of 13.25 kg was possible.
10. If we ignore steaming rate and other contributions that could have caused dilution of hydrogen in the refueling floor, a simple linear first order differential equation model would result in a LEL of Hydrogen (4% by volume) in about 15.8 hr or 3/14/2011 05:00. By the time damage was discovered the concentration could have been as high as 65%. This ignores significant dilution due to approximately 20 m^3 of steam entering the refueling floor every second. That steam could have explained why the blowout panels were open on Unit 4 since they would relieve at a low (inches of water) pressure. There was no ventilation assumed due to lass of power, however there could have been some airflow through the blowout panels due to steaming, condensation and buoyancy driven flow.
11. Assuming that all of the generated hydrogen remained in the building, I used an spreadsheet (Estimating Pressure Increase and Explosive Energy Release Associated with Explosions, Version 1805.0), from the NRC website to estimate the explosive power of the hydrogen. This spreadsheet calculated an equivalent TNT weight of 385 kg. Using the perimeter of damage calculation method from US NRC RG 1.91 the radius for overpressure of 1 psi where structural damage could occur is about 130 m.

The intent of these calculations was to determine if the times and sources of heat and radiation would permit explosive concentrations of hydrogen to exist with only radiolysis and fuel pool boiling. More detailed modeling of the interplay between air flow, steam, and hydrogen or use of thermal-hydraulic analysis codes could refine this picture. , but as of now it looks like this scenario is plausible. Thanks AntonL.

Is this realistic? You calculate a steam production of 20 m^3 per second. You state that this steam production explains why the blowout panels were open on unit 4. It seams to me that in a space that is open to the outer world and that is flushed with 20 cubic meters of steam per second (I assume that this means that the upper floor of the reactor is "refreshed" with steam every few minutes or so) any accumulation of hydrogen is very unlikely. It seems more likely to me that the hydrogen concentration in this upper floor will more or less equal the hydrogen/steam production rate coefficient.

 

[Astronuc]

Thanks for explanations, in case of ignition
Emergency lighting batteries giving out - I think that we can exclude this because batteries die 8 hours after blackout (if I remember correct from reports)


@Astronuc yes I understand there is more fuel (and not so old) in 4 than in others, but for example they were able to fill unit 1 sfp more than 2, and also no change in temperature in sfp 3 and 4 suggest sensor failure

Yes - it is quite possible that various sensors have lost calibration. It may be the case that the core level readings were faulty, and that the cores went dry even though the level indicators showed some level of coolant. Similarly for the SFPs.

 

[NUCENG]

Is this realistic? You calculate a steam production of 20 m^3 per second. You state that this steam production explains why the blowout panels were open on unit 4. It seams to me that in a space that is open to the outer world and that is flushed with 20 cubic meters of steam per second (I assume that this means that the upper floor of the reactor is "refreshed" with steam every few minutes or so) any accumulation of hydrogen is very unlikely. It seems more likely to me that the hydrogen concentration in this upper floor will more or less equal the hydrogen/steam production rate coefficient.

The secondary containment (reactor building) is designed for minimal leakage so it can be maintained at a negative pressure with normal ventilation during plant operation and with the SBGT system during accidents. Any significant heatup or energy release can raise building pressure and cause the blowout panels to open. Once they are open some ventilation flow due to wind or steam or hydrogen may occur but the flow rates will be fairly small due to no mechanical ventilation. If anyone can suggest approaches to refine the modeling of steam and hydrogen and airflow in the refueling floor, I can take another stab at it.

I calculated a boiloff rate based on adiabatic heating and came up with 20 m^3/sec. That would be saturated steam at 100 deg C entering the air on the refueling floor. Some of it will heat up the air space and condense. More will condense on surfaces that are below 100 decg C. Some will indeed slip out through the open blowout panels, but without a lot more time and work all I can say is that my numbers are conservative. That 20 m^3/sec will not result in a flow out the openings of 20 m^3/sec. The approximate volume of the refuel floor is about 2E5 m^3. It will be hot and humid and there will be some steam dilution and mixing of hydrogen. However over time the hydrogen concentration has 24 hours from the minimum time I calculated to accumulate to the Lower Explosive Limit in the top of the bulding. If even a few kg of hydrogen reach an explosive concentration there is enough energy to perform significant damage to the building as we see.

My purpose was not to say what happened. I don't know that for certain, but to use the tools I have to see if I could find anything that would rule out the theory of radiolysis as the source for building 4 damage. What I found is that the theory is plausible, not proven, but it may be what happened. It further supports previous evidence that the fuel in unit #4 was not significantly damaged. (Low concentrations of radioctivity in SFP #4, and visual evidence).

 

[tonio]

The secondary containment (reactor building) is designed for minimal leakage so it can be maintained at a negative pressure with normal ventilation during plant operation and with the SBGT system during accidents. Any significant heatup or energy release can raise building pressure and cause the blowout panels to open. Once they are open some ventilation flow due to wind or steam or hydrogen may occur but the flow rates will be fairly small due to no mechanical ventilation. If anyone can suggest approaches to refine the modeling of steam and hydrogen and airflow in the refueling floor, I can take another stab at it.

I calculated a boiloff rate based on adiabatic heating and came up with 20 m^3/sec. That would be saturated steam at 100 deg C entering the air on the refueling floor. Some of it will heat up the air space and condense. More will condense on surfaces that are below 100 decg C. Some will indeed slip out through the open blowout panels, but without a lot more time and work all I can say is that my numbers are conservative. That 20 m^3/sec will not result in a flow out the openings of 20 m^3/sec. The approximate volume of the refuel floor is about 2E5 m^3. It will be hot and humid and there will be some steam dilution and mixing of hydrogen. However over time the hydrogen concentration has 24 hours from the minimum time I calculated to accumulate to the Lower Explosive Limit in the top of the bulding. If even a few kg of hydrogen reach an explosive concentration there is enough energy to perform significant damage to the building as we see.

My purpose was not to say what happened. I don't know that for certain, but to use the tools I have to see if I could find anything that would rule out the theory of radiolysis as the source for building 4 damage. What I found is that the theory is plausible, not proven, but it may be what happened. It further supports previous evidence that the fuel in unit #4 was not significantly damaged. (Low concentrations of radioctivity in SFP #4, and visual evidence).

OK. Some remarks. If I am correct, the size of R4 building is 35 x 45 m. Assuming that the height of the refuelling floor is close to 15 m, it;'s volume is about 20.000 m^3, about a tenth of what you specify. This space, which I assume is well isolated (apart from the open blowout panels) and is not mechanically ventilated anymore, is heated by a large SPF, which generates about 3 MW of heat. I am not a physicist, but I assume that such a powerfull heat source, dumping an amount of steam in this space which is enough to fill it completely in just 1,5 minutes, will heat it to a temperature close to 100 degrees and will effectively refresh it's (steamy) atmosphere every few minutes.

 

[thehammer2]

I just want to bring up a point regarding NUCENG's calculations. When you consider the steam rate and the hydrogen production rate coming off of the pool, the ratio of steam to hydrogen in the space above the pool based on those production rates is only valid right at the surface of the pool. As the gas mixture leaves the pool, the hydrogen is MUCH less dense than the gases around it, so due to buoyancy alone, it will rise to the top and accumulate. The result is stratification, like oil floating on top of water.

Now, to picture what would tend to happen consider this analogy. Let's imagine we have an upside down glass sitting on a table. This glass is the containment building, and the surface of the table is the surface of the SFP. We begin filling the glass somehow from the surface of the table (How is unimportant in this analogy. It's just coming from the surface of the table) with a mixture of water (representing steam) and oil (representing hydrogen). If the glass is filled slowly relative to its volume, the water (steam) and oil (hydrogen) separate with the oil rising to the top. Once the glass (containment building) is full, let's further imagine the water and oil keeps trying to enter the glass from the table surface (more steam and hydrogen leaving the SFP) and it springs a leak (analagous to the blowout panels popping) below the oil-water interface.

Now, what's newly entering the glass from the surface of the table is still a mixture, but what's leaving the glass because the exit points are below the water/oil interface is largely water. Sure, some oil will leave through the holes, but the oil that was above the holes before they opened up will remain, as there's no real way for them to get to the holes. Further, as the water and oil are still entering at the same rate they were before, but the liquid leaving the glass is largely water, the concentration of the oil in the glass begins to rise since the water is preferentially leaving the glass due to the location of the holes. The oil further collecting in the glass is still rising due to buoyancy and increasing the size of the pocket of oil trapped at the top.

I think this analogy can pretty well illustrate the behavior of the gases. Sure, there would be a gradient of concentrations versus a hard interface, and some of the behaviors of the materials would be different due to the differences between gases and liquids, but the point is to illustrate the mechanism by which buoyancy of the gases and the location of the entrance and exit points of the gases can set up a method by which a small hydrogen rate compared to the steam rate can result in stratification and buildup of explosive gases.

 

[Borek]

I am not so sure about stratification. Bubbles of pure hydrogen will go up, no doubt about it. But if the gases are well mixed, from what I remember stratification due to masses of molecules is negligible, as mixing due to thermal motion is way too strong. For reasonably good results you need very high towers and very low temperatures (cryogenic distillation) - neither were present.

 

[elektrownik]

http://jen.jiji.com/jc/eng?g=eco&k=2011053000813"

 

[Bandit127]

http://jen.jiji.com/jc/eng?g=eco&k=2011053000813"

To provide a sense of scale, the 2.9 million becquerels (MBq/ccm) of Cs 137 quoted in the article is very close to the 3.0 MBq/ccm that was reported in water from Unit 2's basement on 19th April. Tepco said that was "extremely high".

http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110419e2.pdf"

 

[Bandit127]

There is a bit more detail on the two workers who exceeded 250mSv here:
http://jen.jiji.com/jc/eng?g=eco&k=2011053000517"

External exposure stood at 73.71 millisieverts for the worker in his 30s and at 88.7 millisieverts for the one in his 40s.

Does this mean that the internal exposure has been normalised to an annual or lifetime dose?

 

[MiceAndMen]

OK. Some remarks. If I am correct, the size of R4 building is 35 x 45 m. Assuming that the height of the refuelling floor is close to 15 m, it;'s volume is about 20.000 m^3, about a tenth of what you specify. This space, which I assume is well isolated (apart from the open blowout panels) and is not mechanically ventilated anymore, is heated by a large SPF, which generates about 3 MW of heat. I am not a physicist, but I assume that such a powerfull heat source, dumping an amount of steam in this space which is enough to fill it completely in just 1,5 minutes, will heat it to a temperature close to 100 degrees and will effectively refresh it's (steamy) atmosphere every few minutes.

Not a huge difference from your estimate, but the height of the refuelling floor is 15.8 m according to the drawings we've seen (for Unit 3), so volume is closer to 25,000 m3.

I would not assume the space is well isolated, however. Various diagrams and descriptions give me the impression that there are a multitude of open hatchways that connect the airspace of every floor to every other floor. If that is true then the atmosphere inside the building as a whole is communicable subject to drafts, convection, and buoyancy factors. There are probably rooms and spaces sealed off due to containment considerations.

But even then the entire volume of the building would be somewhat less than 71,000 m3 (35 x 45 x 45) because the primary containment vessel - which shuold be completely isolated - occupies some fraction of the total volume.

 

[elektrownik]

http://ajw.asahi.com/article/0311disaster/fukushima/AJ201105300291"
http://ajw.asahi.com/article//0311disaster/fukushima/AJ201105300298" interesting !

 

[MiceAndMen]

http://ajw.asahi.com/article/0311disaster/fukushima/AJ201105300291"
http://ajw.asahi.com/article//0311disaster/fukushima/AJ201105300298" interesting !

Ah yes, Megafloat to the rescue. Again. It holds 10,000 tons of water and it has been obvious for quite a while that it's not going to be very useful. A drop in the proverbial bucket.

And hoses. TEPCO has secured hoses for Plan B. Or is it Plan C, D, E or F?

 

[~kujala~]

I don't know where the news reporter got this info but he seems to confirm that groundwater level is quite high:

It has not been confirmed that contaminated water has leaked into the groundwater from the basements in large quantities, but the levels of contaminated water in the basements are currently only a few meters lower than that of the groundwater.

http://ajw.asahi.com/article/0311disaster/fukushima/AJ201105300291

By the way, did you notice that he copied what jlduh said a couple of days ago (May 27th):

So what I foresee is the possibility that when the basements where almost empty (with regular pumps ejecting out the inflows from the watertable, which was probably routine operation to keep these basements dry), then of course the direction of flow was from outside to inside (because of hydrostatic differential). When the basements are filling in with water, the differential is reducing and eventually, this differential can be inverted if water level inside basement becomes higher than water table level outside. Then the flow will invert also, and so leakage from basement towards watertable can happen (with contamination).

https://www.physicsforums.com/showpost.php?p=3325027&postcount=8356

Asahi Shimbun May 30th:

As long as water is flowing from the surrounding groundwater into the contaminated water, because the level of the groundwater is higher than that in the basements, the threat of substantial leaks is not considered acute. But if the level of the contaminated water rises above that of the groundwater, water would begin flowing in the other direction and is likely to spread contamination.

 

[htf]

I calculated a boiloff rate based on adiabatic heating and came up with 20 m^3/sec.

How did you get these figures? I get 2 m^3 / sec.

I calculate 1.35 kg/sec boiloff rate: 3.05e6 W/(2.256e6 J/kg) These are ~75 mol, which give 22.4 * 373/273 * 75 l = 2.3e3 l = 2.3 m^3 steam - make it 2 m^3 because it is saturated and not an ideal gas.

 

[elektrownik]

I don't see any solution for Fukushima (in case of water leaks), they need to inject water to RPV to cool down melted cores, they can't stop, but it look like all RPVs are leaking and drywells also so water is going to reactor and turbine buildings, and to other locations. To install cooling system they need closed loop, but to do this they would need fix at last drywell leak, but we still don't know how big is damage. In theory they could fix leak, but to do this they would need to stop water injection, but they cant, and if they could then there is also extrem radiation problem, I think that it will be not possible to work in leak location for many years, radiation would be too hight. Closed loop with reactors as water tanks is not a solution also because we know that they are not sealed and water is leaking outside...

 

[thehammer2]

I am not so sure about stratification. Bubbles of pure hydrogen will go up, no doubt about it. But if the gases are well mixed, from what I remember stratification due to masses of molecules is negligible, as mixing due to thermal motion is way too strong. For reasonably good results you need very high towers and very low temperatures (cryogenic distillation) - neither were present.

Agreed. I was definitely thinking more of discrete pockets of gas coming out of the water that were large enough to have enough buoyancy to get significantly toward the top of the building before spreading out and mixing into the surroundings, not a nice mixture leaving the SFP. My mechanism requires that the gases aren't very well mixed when they leave the pool.

 

[etudiant]

I don't see any solution for Fukushima (in case of water leaks), they need to inject water to RPV to cool down melted cores, they can't stop, but it look like all RPVs are leaking and drywells also so water is going to reactor and turbine buildings, and to other locations. To install cooling system they need closed loop, but to do this they would need fix at last drywell leak, but we still don't know how big is damage. In theory they could fix leak, but to do this they would need to stop water injection, but they cant, and if they could then there is also extrem radiation problem, I think that it will be not possible to work in leak location for many years, radiation would be too hight. Closed loop with reactors as water tanks is not a solution also because we know that they are not sealed and water is leaking outside...

Is not the plan to have the water recycled after going through AREVAs decontamination?
If the planned 1200 ton/day processing is achieved, they can reuse the cleaned water and still gradually drain the site, because the cooling only uses about half that.
AREVA expects to process 250,000 tons of water, so they plan to be at this well into next year.
Where they put the processed water other than into the Pacific is still an open question. Presumably one could put it into an old supertanker and moor it somewhere out of typhoon prone areas, but for the cesium 137 it would only be a halflife or so before the ship rusted out completely.

 

[frangin]

@swl ,
From the same samples, Cs134 from 1600 up to 4100 and Cs137 from 1700 to 4300 becquerel/liter

i found it in this realease :

http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110529e3.pdf

 

[Jorge Stolfi]

... [ignition of hydrogen in unit #4] ...

Hydrogen/oxygen mixtures can be ignited by contact with a suitable catalyst, such as certain bare metals. The catalyzer will initially combine H2+O2 flameless, but will get hot as a result. (The explosion that convinced Fleischmann and Pons that they had achieved cold fusion was later conjectured to be a chemical D2 + O2 explosion catalyzed by palladium.)

 

[Jorge Stolfi]

As the gas mixture leaves the pool, the hydrogen is MUCH less dense than the gases around it, so due to buoyancy alone, it will rise to the top and accumulate. The result is stratification, like oil floating on top of water.

This does not happen. No matter what the difference in density, gases that are mixed will stay mixed, and gases that are initially stratified will gradually diffuse into each other, even if they are kept perfectly still (without macroscopic motion). Indeed there must have been significant convection currents in the service floor of #4, because of the powerful heat+steam source in one corner.

That said, IMHO radiolysis seems a better explanation than H2 leakage from #3. I cannot see how a warm steam+H2 mixture, much lighter than air, would have chosen to travel backwards into the venting pipe of #4, and force its way through a tortuous path with several valves and tubing into a closed building with the AC turned off --- instead of flowing up the venting tower. I suppose that if some of the H2+steam from #3 managed to get into the chimney, it would have created negative pressure at its base, thus sucking back any gas that happened to flow towards #4

 

[jlduh]

I don't know where the news reporter got this info but he seems to confirm that groundwater level is quite high:


http://ajw.asahi.com/article/0311disaster/fukushima/AJ201105300291

By the way, did you notice that he copied what jlduh said a couple of days ago (May 27th):

https://www.physicsforums.com/showpost.php?p=3325027&postcount=8356

Asahi Shimbun May 30th:

Maybe some journalists are reading this forum
Next time I will write some big BS (humm I'm sure i did already!) and we will see if they copy it

 

[jlduh]

I don't see any solution for Fukushima (in case of water leaks), they need to inject water to RPV to cool down melted cores, they can't stop, but it look like all RPVs are leaking and drywells also so water is going to reactor and turbine buildings, and to other locations. To install cooling system they need closed loop, but to do this they would need fix at last drywell leak, but we still don't know how big is damage. In theory they could fix leak, but to do this they would need to stop water injection, but they cant, and if they could then there is also extrem radiation problem, I think that it will be not possible to work in leak location for many years, radiation would be too hight. Closed loop with reactors as water tanks is not a solution also because we know that they are not sealed and water is leaking outside...

This is called a technical NIGHTMARE. Maybe less spectacular than Tchernobyl, but much more perverse on the long run, IMHO.

 

[hbjon]

How did the area handle the storm that just went through? Or is it still going through?

 

[hbjon]

Maybe some journalists are reading this forum
Next time I will write some big BS (humm I'm sure i did already!) and we will see if they copy it

Someone stole (borrowed) my analogy to the "Titanic" and called the disaster a "Nuclear Titanic", early on. Not even so much as a thank you e mail. Sigh.

 

[swl]

@swl ,
From the same samples, Cs134 from 1600 up to 4100 and Cs137 from 1700 to 4300 becquerel/liter

i found it in this realease :

http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110529e3.pdf

On page two of the above quoted document, TEPCO's data indicates that the Iodine-131 level measured in water from "Screen of 1Fs unit 2 (inside the silt fence)" is much higher than the levels indicated in Units 1, 3 and (of course) 4.

Given similar starting levels it would take well over a month of decay to account for this difference. I understand that the starting points would likely have been different, so I considered the ratio between the Iodine-131 and the Cesium levels (both 134 and 137). Unit two appears to have far more Iodine relative to Cesium than any of the other units.

Unit 1: I131 650Bq/L; Cs134 1,500Bq/L; Cs137 1,600Bq/L
Unit 2: I131 24,000Bq/L; Cs134 4,100Bq/L; Cs137 4,300Bq/L
Unit 3: I131 720Bq/L; Cs134 5,100Bq/L; Cs137 5,400Bq/L
Unit 4: I131 160Bq/L; Cs134 4,500Bq/L; Cs137 4,800Bq/L (shut down mode before accident and much lower Iodine levels than the other units)

So, I am now wondering if this could be an indication that unit 2 had a criticality accident over a month after the tsunami, or is there some other likely explanation?

 

[Atomfritz]

Here some quotes about the hydrogen collecting up in the buildings:

I am not so sure about stratification. Bubbles of pure hydrogen will go up, no doubt about it. But if the gases are well mixed, from what I remember stratification due to masses of molecules is negligible, as mixing due to thermal motion is way too strong. For reasonably good results you need very high towers and very low temperatures (cryogenic distillation) - neither were present.

Hydrogen/oxygen mixtures can be ignited by contact with a suitable catalyst, such as certain bare metals. The catalyzer will initially combine H2+O2 flameless, but will get hot as a result. (The explosion that convinced Fleischmann and Pons that they had achieved cold fusion was later conjectured to be a chemical D2 + O2 explosion catalyzed by palladium.)

No matter what the difference in density, gases that are mixed will stay mixed, and gases that are initially stratified will gradually diffuse into each other, even if they are kept perfectly still (without macroscopic motion). Indeed there must have been significant convection currents in the service floor of #4, because of the powerful heat+steam source in one corner.

That said, IMHO radiolysis seems a better explanation than H2 leakage from #3. I cannot see how a warm steam+H2 mixture, much lighter than air, would have chosen to travel backwards into the venting pipe of #4, and force its way through a tortuous path with several valves and tubing into a closed building with the AC turned off --- instead of flowing up the venting tower. I suppose that if some of the H2+steam from #3 managed to get into the chimney, it would have created negative pressure at its base, thus sucking back any gas that happened to flow towards #4

Just my unqualified 2 cents:
First of all, thanks to Nuceng and all the others for their valuable insights!

Please do also consider the fact that much, if not most of the steam will condensate at the walls etc due to the very low dew point.
The increased pressure even speeds up condensation.
This means that the mixture steam-hydrogen inevitably gets richer in hydrogen with time, because hydrogen cannot condensate under earthly temperatures/pressures.
Until explodable conditions have been achieved, Then just a little spark or hot surface could initiate the "kaboom procedure"...

The observation that only RB#2, the only that allowed hydrogen to escape, remained in shape, appears to confirm my hypothesis as far I see.

Compare this to a fridge where you put a pot of boiling water in. The steam will condense at the walls, some leaking out at the door, but not popping it open.
(Finally, due to lack of (pressed out) air the fridge will develop underpressure, making it difficult to open the door. Unlike a reactor...)

Consider this photo just before explosion.
Do you also see steam leaving through building weak points at wall/ceiling corner of reactor building #1 (left)?
Doesn't this indicate high pressure in the reactor building?
(Or maybe I misinterpret this image ? It could be Daini or some other plant?!? But what is that optical distortion looking like steam? German "Spiegel" posted this photo short before Daiichi explosion #1, with picture description suggesting it was Daiichi 1+2...)

So, could the inevitable outcome have been that eventually an explosive hydrogen-oxygen ratio developed, ready to be ignited by a slight spark or some hot surface?

 

[jlduh]

How did the area handle the storm that just went through? Or is it still going through?

The new live webcam is ON, you can check it by yourself...


http://www.tepco.co.jp/nu/f1-np/camera/index-j.html

 

[StrangeBeauty]

...
Consider this photo just before explosion.
Do you also see steam leaving through building weak points at wall/ceiling corner of reactor building #1 (left)?
Doesn't this indicate high pressure in the reactor building?
(Or maybe I misinterpret this image ? It could be Daini or some other plant?!? But what is that optical distortion looking like steam? German "Spiegel" posted this photo short before Daiichi explosion #1, with picture description suggesting it was Daiichi 1+2...)

So, could the inevitable outcome have been that eventually an explosive hydrogen-oxygen ratio developed, ready to be ignited by a slight spark or some hot surface?

My first thought was prop wash from the helicopter causing that image distortion...? Do you have another image without the chopper?

Thanks to a number of people for all the detailed analysis in the attempt to come up with plausible scenarios for the #4 explosion/fire.