Japan Earthquake: Nuclear Plants at Fukushima Daiichi

[jlduh]

The recent data from the analysis of water from SFP4 would seem to me to be strong evidence of leaking. Taking account of the decay of iodine-131, the data for all three measured isotopes indicates that the pool has lost half of the content of soluble matter it had 14 days ago.
http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110429e13.pdf

When i look at these numbers and compare them to the ones from the analysis of the sub drain water (they are all in Bq/cm3),
http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110428e14.pdf

i have a hard time to understand why the ones from the SFP are not much higher...

What can explain these so low numbers in comparison with the subdrain levels? Are the subdrain so heavyly contaminated that the level ae actually higher than in the SFP?

 

[ascot317]

I wonder why CNN no longer reports about the nuclear plants.. guess it's no longer serious and problems almost solved. (?)

News Media "Interest" in any topic has a half time of two weeks, unless something "new" happens.
Same with Libya. The plant is still a serious problem.

 

[NUCENG]

why put fresh fuel into SPF anyway? Fresh fuel is much more reactive, and requires more boron...

Fresh fuel is valuable and fragile. It is as much to protect the new fuel as anything else. After receit inspection and channeling, it is put into the pool ready to be picked up and loaded into the core. Fuel pool racks have boron neutron absorber as you recall (BORAL) and thet is sufficient to keep them subscritical. Reactor engineers also evaluate fuel placement to ensure Keff stays in the safet range. Why would it be unsafe to store new fuel in the pool?

 

[bytepirate]

The recent data from the analysis of water from SFP4 would seem to me to be strong evidence of leaking. Taking account of the decay of iodine-131, the data for all three measured isotopes indicates that the pool has lost half of the content of soluble matter it had 14 days ago.
http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110429e13.pdf

a rough calculation gives a loss of 50 tons per day. assuming, that cesium can not leave but through a leak, and the water level is held constant by refills.
on the other hand: if the pool was filled only half at the time of the first measure and now is full, the result would be the same. do we know?

 

[zapperzero]

News Media "Interest" in any topic has a half time of two weeks, unless something "new" happens.
Same with Libya. The plant is still a serious problem.

Here's your fresh news: http://online.wsj.com/article/BT-CO-20110429-708521.html" .

 

[ascot317]

Here's your fresh news: http://online.wsj.com/article/BT-CO-20110429-708521.html" .

Nah, that's just background noise. "New" would be something breaking up, exploding, someone getting killed etc. Which could still happen.

That adviser seems to be exactly the right guy to do advising.

 

[RealWing]

Nah, that's just background noise. "New" would be something breaking up, exploding, someone getting killed etc. Which could still happen.

That adviser seems to be exactly the right guy to do advising.

I have great respect for someone with strong public safety principles. Another 'hero" emerges from this event.

 

[MadderDoc]

a rough calculation gives a loss of 50 tons per day. assuming, that cesium can not leave but through a leak, and the water level is held constant by refills.
on the other hand: if the pool was filled only half at the time of the first measure and now is full, the result would be the same. do we know?

You are right, that's a possibility we cannot exclude, although the levels being only half of what's in it now stretches my imagination, I don't think it is full now.

What we know from the data that has been released (http://www.gyldengrisgaard.dk/fuku_docs/fuku_sfp_sprayings.html) is there seems to have been a 3 day lull in the sprayings to SFP4 up to the sampling on March 12th. We know also that shortly after midnight the night after the sampling (00:30 on March 13th), spraying to the pool commenced and continued throughout the night, for six and a half hour, one of the longest spraying periods recorded.

 

[MiceAndMen]

SFP #4 underwater video ! http://www.tepco.co.jp/en/news/110311/images/110428_1.zip
We can see that upper parts of fuel sets are melted

To this layman's eye, there seems to be very little obvious damage. Some zircaloy somewhere in that pool must have oxidized to generate the hydrogen that exploded, and maybe the fuel seen here is damaged down below where we can't see, but I see very little physical damage.

This is but one section of the SFP where little has fallen in on top of the racks. The four 6 x 10 racks visible in the video have space for 280 assemblies total, about 20% of the total number of assemblies in the pool. They don't all seem to be full, meaning we're seeing somewhat less than 20% of the assemblies. I'd love to see what the rest of the racks look like.

People claiming to see damage lurking behind every little shadow of video have overactive imaginations IMO. This video, at least to my untrained eye, is reason to hope that the SFP in Unit 4 is not a huge jumbled mass of chaos and twisted metal.

 

[AntonL]

The recent data from the analysis of water from SFP4 would seem to me to be strong evidence of leaking. Taking account of the decay of iodine-131, the data for all three measured isotopes indicates that the pool has lost half of the content of soluble matter it had 14 days ago.
http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110429e13.pdf

Leaking can be very quickly confirmed by the salt content of the pool, the pool was cooled by adding sea water from March the 20, which has been concentrated by the continuous boiling. March 30 was the first day fresh water was used.
I suppose a salinity test would be too innovative and too simple for Tepco

SFP4 radioactivity is a fraction to what was measured in SFP2
extract from http://www.nisa.meti.go.jp/english/files/en20110421-1-5.pdf page 9

The work of sampling water that flowed out in the Skimmer Surge Tank
from the Spent Fuel Pool was carried out in order to grasp the condition
of water in the pool. (April 16th) As a result of nuclide analysis of
radioactive materials regarding the sampled water of the pool,
4.1×10³Bq/cm³ of 131 I (Iodine),
1.6×10⁵:Bq/cm³ of 134 Cs (Cesium),
1.5×10⁵:Bq/ cm³ of 137 Cs (Cesium)
were detected. (April 17th

 

[MiceAndMen]

I suppose a salinity test would be too innovative and too simple for Tepco

My thoughts exactly.

 

[MiceAndMen]

For a rough estimate of the length and width of buildings, the plant layout map should be able to yield data of sufficient precision:
http://gyldengrisgaard.dk/fuku_docs/plant/

Good estimates of the W and L of the reactor buildings might be of great utility to others, so do not hesitate to post them.

The measurements on the diagram are not consistent. When scaled appropriately along the Y axis (I used Autocad), the distances between the X coordinates are way off. And vice versa. I would trust the XY numerical values given, but simply tracing the outlines of things will not yield accurate shapes or positions.

 

[etudiant]

The measurements on the diagram are not consistent. When scaled appropriately along the Y axis (I used Autocad), the distances between the X coordinates are way off. And vice versa. I would trust the XY numerical values given, but simply tracing the outlines of things will not yield accurate shapes or positions.

Would it be sensible to use the Google Earth images to set the proportions and then derive the floor plan dimensions from that?
Admittedly, that does not help for interior partitions or depth. so this may be a dead end.
In any event, TEPCO is projecting up to 200,000 tons of water in the plant, so that gives a minimum volume.

 

[Caniche]

a rough calculation gives a loss of 50 tons per day. assuming, that cesium can not leave but through a leak, and the water level is held constant by refills.
on the other hand: if the pool was filled only half at the time of the first measure and now is full, the result would be the same. do we know?

JAIF status report details the spraying of 650 tons of freshwater into SPF4 over the period 23-26/4 . Steam and mirrors?

 

[Caniche]

Bloomberg reports robots reading 1,120 mSV/h in reactor building 1 on 26/4 , that's not good is it?
Tepco reckons the readings might be a tad too high to allow their plan to flood the "drywell "containment to proceed.
Hmm, surely flooding the outer containment structure will only guarantee submersion of the remaining core if the RPV containment is breached at a high level /or high level venting of extreme radioactive material is still physically possible?

 

[fluutekies]

The recent data from the analysis of water from SFP4 would seem to me to be strong evidence of leaking. Taking account of the decay of iodine-131, the data for all three measured isotopes indicates that the pool has lost half of the content of soluble matter it had 14 days ago.
http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110429e13.pdf

Agree. Here my try for quantification of the leak of SFP4 based on Cs-isotopes:
L 12.2 m
W 9.9 m
D 11.8 m
V 1425.2 m3
m fuel 264.0 kg U
V net 1200 m3

Cs-137
93 Bq Cs-137/cm3 on 2011-04-28
55 Bq Cs-137/cm3 on 2011-04-13

Cs+ ion is non-volatile & can only leave by leaks or decay, decay is negligible (15 days/30 years).

59% Cs-137 remains in SFP after 15 days. Assuming homogeneous distribution by convection and diffusion.

710 m3 remaining volume with original 93 Bq Cs+
490 m3 lost volume with 93 Bq Cs+
15 days
23 L/min
1.4 m3/h
32.7 m3/day ~ -30cm level/day

This simple model assumes that the total volume is lost once, refilled & homogenized again.
In reality the loss will be continuous & refilled periodically.
The calculated loss by leaking of about 23 L/min therefore is the lower limit. Under the real conditions (periodic refill & continuous homogenization), therefore the leak is probably a factor of 3 -5 higher (my guess). But this quantity (75 - 150 L/min) is still small enough to get undetected somewhere in or outside the building.

4.4E+11 Bq Cs-137 leaked in this period from SFP4. Not a small amount. At least as it is expressed in Bq/L instead of /cm3 and multplied with the volume for the absolute inventory.
Same calculation with Cs-134 gives comparable result.
Edit: Read some remarks after posting: both volumes are assumed equal which seems reasonable due to the fact that TEPCO is refilling daily.

 

[razzz]

I thought I read TEPCO was doing a balancing act for SFP #4, trying to keep the weigh factor down with less water and their figures of re-watering were inline with boil off losses. So they allow the pond to get so hot but not to hot and still keep coverage over the fuel assemblies. In the meantime, they try to figure out if there is a leak or not. First there is then there isn't.

 

[MadderDoc]

The measurements on the diagram are not consistent. When scaled appropriately along the Y axis (I used Autocad), the distances between the X coordinates are way off. And vice versa. I would trust the XY numerical values given, but simply tracing the outlines of things will not yield accurate shapes or positions.

Allright. I took a wrestle with that map, and I do see what you mean, it's not that easy using this map which surely could be better. Now, one should not let 'better' be the worst enemy of 'good enough' , and how much precision do I actually need. I quickly found out that a pair of extra assumptions, the reasonability of which you may not be aware, would make things a lot easier.

a) it is a fair assumption that the basefloors of the reactors are quadratic.
b) it is a fair assumption that units 2-4 have closely the same dimensions.

Using those assumptions, and a ruler measurement of a marked up 125.000 m distance in the NS direction of the map, I get estimates that the basefloors of Unit 2-4 are 46.4 x 46.4 meter (topfloors 46.3 x 33.4 m). Unit 1 is smaller, the estimate for that using the map gave me 40.6 x 40.6 m, impressively close to the data we have in diagrams of sections of the generic sort of reactor, which unit 1 is one of,

I am sure some luck was involved there but I took it, that the method was kinda working. I should add that there is no markup of the topfloor in unit 1 in the map, so the only recourse for that is the generic section diagram, the estimate for the topfloors of unit 1 becomes 40.6 x 30.4 m.

 

[bytepirate]

therefore the leak is probably a factor of 3 -5 higher (my guess).

no need to guess ;-) the factor is ~1.5 (http://en.wikipedia.org/wiki/Compound_interest: e^0.41)

 

[Jorge Stolfi]

[Could you confirm] the depth of the floor basement in the reactor building, below the line of the platflorm ground? (probably n°1 is different because it is a different reactor, but 2 to 4 should be similar). It should be something like 11 meters i think but maybe you can confirm this. The depth of the basement of T/B seems aligned with the one of the R/B so this info could help me to check and scale the sketch i captured on the NHK.

These are the only blueprints that I have which seem to match the photographs of exploded buildings:

http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/good/
http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/good/un3_cut_N_1.png (E-W cut)
http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/good/un3_cut_W_1.png (N-S cut)
http://www.ic.unicamp.br/~stolfi/EX...a/povray/blueprint/good/un3_service_floor.png (service floor plan)

The first one (E-W cut, looking towards north) seems to be a hurried sketch by someone who had at least some technical drafting experience.

The second one (N-S cut) is rather amateurish: the circles seem to be drawn with templates rather than compass, the sides of the building are not parallel, the midline is not exactly in the middle, etc. More importantly, the spacing of floors in the drawing does not match the numbers on the side. Also the N-S size of the SFP does not match the photos.

The third one (service flor plan) is also rather amateurish, just a a sketch done with some common illustrator program (rather than Autocad or such). The original was visibly stretched N-S. I fixed the image to match the true(?) aspect ratio of the building, but then the reactor opening ended up slightly squashed the other way. Still, the positions and sizes of pillars and other details seem to match the photos quite well.

As you can see, the only numbers explicitly given in those drawings are the floor heights (in meters) above the local standard reference level ("O.P."). The external ground surface is at OP+10.000 meters, and the floor of storey 1 ("ground floor") is at OP+10.020 meters (i.e only 2 cm above the external ground). CORRECTION: OP+10.200 , i.e 20 cm above external ground. The basement floor is at OP-2.060 meters (not counting the trench where the torus sits), and the bottom of the concrete base is at OP-6.060 meters. The buried wing on the West side is an extension of the basement, and that on the East side is an extension of basement plus storeys 1 and 2. The terrace roof seems to be at the same level as floor 3 inside the building.

The following file contains the measurements of units 2-4 that I am using in my POV-Ray models. Note that many of the numbers are my estimates ("E") obtained by measuring distances on the drawings and doing the appropriate scaling. Many others are just guesses, still to be confirmed ("TBC").

http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/un4_dimensions.inc

2) seen from the top, and also for scaling purposes, the exact outside dimensions of the R/B 1 to 4 (1 is a little bit smaller maybe)? .

If the N-S and E-W cut drawings can be trusted, the middle part of buildings 2--4 (storeys 1 and 2) seems to be approximately square, 47.5 meters on each side (give or take 1 meter, perhaps). The top part (storeys 3 to 5) is about 8 m narrower: 35.5 meters E-W by 47.5 N-S. The basement extension on the West side is about 10.0 meters wide.

Unit 1 is indeed smaller. For that one we have actual engineering blueprints (with the original caption "Fukushima Daiichi Unit 1" at the corner) with hundreds of accurate measurements. Unfortunately, only for the N-S and W-E cuts thrugh the reactor axis:

http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/good/un1_cut_N_1.png (E-W cut)
http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/good/un1_cut_W_1.png (N-S cut)
http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/misc/un1_blueprint_big.jpg (both, hi-res, big file)
http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/un1_dimensions.inc (dimensions I am using)

All these blueprints were uncovered by other contributors to this thread, and the original documents are available through previous posts.

I hope it helps...

 

[MJRacer]

Ongoing criticalities at Unit 2?

http://www.glgroup.com/News/TEPCO-D...e-Leaking-Fukushima-Daiichi-Unit-2-53751.html

 

[Jorge Stolfi]

Hey Jorge Stolfi, you seem to know your building elevations so...if the RPV cap at the join to the vessel leaks and vents, what level or plane would it be projecting or slicing through in relation to Unit 3? Or if you cut the Unit 3 down to the elevation level with the RPV flange (like cap removed), what would it look like on your modeling?

Here is the current version of my model:

http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/out/fig_un4_building_cut_NW.png

If I understood the blueprints correctly, there is a washer-like steel plate connecting the drywell neck (yellow) to the reactor pressure vessel (gray). That plate separates the bulb-like drywell proper from the refueling pool --- the cavity on the service floor where the shroud is located. That partition is supposed to be water-tight; AFAIK, in normal operation and most core maintenance work the refueling pool is filled with water, while the drywell is, well, dry. So if the drywell is overpressured and the shroud joint gives way, the gases should blast sideways into the refueling pool.

However, the flanges connecting shroud and drywell seem to be rather massive things held together with a zillion heavy bolts. So I would expect that the walls of the drywell will rupture at some weld somewhere else, well before the bolts do. Is that correct?

 

[jlduh]

Ongoing criticalities at Unit 2?

http://www.glgroup.com/News/TEPCO-D...e-Leaking-Fukushima-Daiichi-Unit-2-53751.html

I repost the graphs for clarity:
http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110428e14.pdf

As i got no answer on my question i ask it again: don't you find these numbers globally high comparing them to the water samples of SFP 4?

 

[Jorge Stolfi]

Re Jorge's estimate of the SFP cavity dimensions, yielding a volume ~1690 m³

The NISA number is given as the water volume of the pool. It may be the equivalent of the SFP concrete cavity minus volume of steel liner minus volume of equipment installed inside the pool (e.g. heat exchangers).

Moreover, some sketches and phtographs seem to suggest that part of the raw SFP space is taken up by the cask loading(?) pool (not shown in my models). That would be a block of concrete about 3.8 m x 3.8 m, with a cylindrical well down the middle, flush against the NW corner of the SFP and spanning its whole depth. The well in this block seems to be about 3m in diameter, and is connected to the SFP by a narrow gate of unknown depth. If that thing really exists, presumably it serves to hold the "dry" casks under water while the fuel is loaded and unloaded. In that case, from the ~1690 m³ you should subtract ~95 m³ to get the free volume of the SFP.

Also note that the water level in the pool is normally at some distance below the service floor.

 

[default.user]

I repost the graphs for clarity:
http://www.tepco.co.jp/en/press/corp-com/release/betu11_e/images/110428e14.pdf

As i got no answer on my question i ask it again: don't you find these numbers globally high comparing them to the water samples of SFP 4?

Excuse my bad English.

We talk about criticality in Germany for 4 weeks.
German physicist from very unpleasant conditions in the reactors.
I can give you, unfortunately only German interviews, if you like.

There are no more news to Fukushima Daiichi.

 

[Jorge Stolfi]

Some of those handles (and the caps of the bundles themselves) seem melted, in particular those in the rack that is partly visible in the bottom right corner, while others are simply not visible above the tops of the racks. I believe they call that "total meltdown"?.

Beware that a lot of roof debris rained down on the SFP after the explosion. The top edges of the rack walls sem to be higher than the top of the assemblies. If so, it is expected that the debris have collected over the latter. Only the larger pieces remained atop the walls. So what seem like damaged assemblies may be just assemblies covered by debris.

By the way, those tech reports on convective cooling of fuel in the SFP seem to ignore the possibility that the water flow up through the assemblies could get blocked by a layer of concrete rubble...

 

[default.user]

http://de.wikipedia.org/wiki/Edmund_Lengfelder

He said four weeks ago:

There is no way to cool a meltdown.
He says that a meltdown is to stop only by its own momentum.

All work at the reactor are unnecessary.
Are only means to avoid the radiation exposure of people and staff.

Humanity before ecology.

 

[Jorge Stolfi]

The video taken on March 11th shortly after the tsunami shows a large portion of the staircase to be still on the building, except the uppermost part of the staircase which is missing. The video does not give a view to the lowermost part of the staircase close to the ground level to see whether is has sustained any damage, but it would be reasonable to expect at least some damage to it since this part of the staircase was inundated by tsunami water with floating debris.

My comments were based on the right half of this image from a previous post, allegedly a photo taken after the earthquake (and tsunami?) but before the #4 explosion:

[PLAIN]http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/foto/edited/src/reactor4-S-7.png

In this photo the staircase seem to be still there, whereas the Mysterious Green Box has been replaced by the Mysterious Hole With Mickey Mouse Ears.

Note that the "Dark Goo Flowing Down From The Terrace" is already there.

If [by the Big Greenish Closet] you mean the apparent green box standing at the _foot_ of the wall, I think it is about the right size (I estimate the dimensions of the green box to be about 4 x 4 x 2 m).

The Mysterious Green Box should be at least 6--7 m tall to match the height of the Mickey Mouse Ears. On the other hand, judging by the ground-level photo below, the Big Green Closet may be quite a bit taller than that. (The terrace is 17 meters above ground level.)

http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/foto/edited/src/reactor4-S-6.png

If the scenario of earthquake damage making the box come off is assumed, this is one of the places it could have ended up. It might initially have fallen down more or less vertically during the earthquake, taking with it the upper part of the staircase.

The photo above seems to indicate that the staircase was still there after the Mysterious Green Box disappeared.

The tsunami waters could then have made the box end up in this corner, behind other debris.

No need to invoke the tsunami. See:

http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/blueprint/foto/edited/src/reactor4-S-5.png

The Big Green Closet either was attached to the wall all along (covering part of the window?) , or is hanging from the terrace by cables/pipes/whatever (the true nature of the "Dark Goo Flowing Down from the Terrace").

 

[default.user]

To put it together.

Both scientists believe that a particular [edit] nuclear fusion can not be controlled.

Both scientists have exquisite experience through Chernobyl.
Both scientists are not particularly popular in Germany.
Neither in East nor in West Germany.

So, how can you cool a core melt with water, without the danger of a steam explosion?

 

[Jorge Stolfi]

Someone posted this map of Fukushima I plant:

http://gyldengrisgaard.dk/fuku_docs/plant/

The overall layout may be correct, but the outlines of reactor buildings 3 and 4 are somewhat misleading. Apparently the drawing shows the outline of the ground floor of each reactor (which is indeed square). The conspicuous top part of the building has actually a rectangular floorplan. The 3rd floor terrace on the east side, between the top part of the reactor building and the turbine building, is about 12 meters wide.

 

[hbjon]

Ongoing criticalities at Unit 2?

http://www.glgroup.com/News/TEPCO-D...e-Leaking-Fukushima-Daiichi-Unit-2-53751.html

I think I can safely say that that would be very distubing if that was indeed the case. I don't know if it is impossible, but getting control rods inserted into a blob of molten corium that is distilling within the rpv seems quite unlikely. Splashing borated water on it isn't going to do anything immho.

 

[Jorge Stolfi]

I get estimates that the basefloors of Unit 2-4 are 46.4 x 46.4 meter (topfloors 46.3 x 33.4 m).

My estimates, from the blueprints and sketches shown in my pevious post, are 47.8 x 47.8 m for the ground floor, 35.8 x 47.8 for the service floors (at the external wall surfaces). Total height (from ground floor to upper side of roof, no counting the parapet) is given as 45.520 m.

Unit 1 is smaller, the estimate for that using the map gave me 40.6 x 40.6 m, impressively close to the data we have in diagrams of sections of the generic sort of reactor, which unit 1 is one of.

By adding numbers from the blueprint I got 41.560 x 41.560 for the ground floor, 31.420 x 41.560 for the service floor (at the external wall surfaces). For the total height (from ground floor to upper side of roof, not counting the parapet) I get 44.400 m.

 

[Jorge Stolfi]

CORRECTION: In a previous post I gave the ground floor height of reactor #3 (and presumably #4) as OP+10.020. Per the blueprints/sketches, it should be OP+10.200 , i.e 20 cm above external ground level (OP+10.000)

 

[MiceAndMen]

Ongoing criticalities at Unit 2?

http://www.glgroup.com/News/TEPCO-D...e-Leaking-Fukushima-Daiichi-Unit-2-53751.html

GL Group seems to be legit, but...

When Units 1-3 were all scrammed on March 11, 2011 from earthquake-caused station blackout

I thought the reactors scrammed when the first tremors of the earthquake were felt and the station blackout occurred later. The unnamed source also says control rods are "dropped in", but in a BWR they are pushed up from below by a hydraulic system.

Maybe he's right about what conclusions can be drawn from the data, but two sloppy mistakes in a very short analysis are like red flags for me.

 

[razzz]

Here is the current version of my model:

http://www.ic.unicamp.br/~stolfi/EXPORT/projects/fukushima/povray/out/fig_un4_building_cut_NW.png

If I understood the blueprints correctly, there is a washer-like steel plate connecting the drywell neck (yellow) to the reactor pressure vessel (gray). That plate separates the bulb-like drywell proper from the refueling pool --- the cavity on the service floor where the shroud is located. That partition is supposed to be water-tight; AFAIK, in normal operation and most core maintenance work the refueling pool is filled with water, while the drywell is, well, dry. So if the drywell is overpressured and the shroud joint gives way, the gases should blast sideways into the refueling pool.

However, the flanges connecting shroud and drywell seem to be rather massive things held together with a zillion heavy bolts. So I would expect that the walls of the drywell will rupture at some weld somewhere else, well before the bolts do. Is that correct?

I can't find the report and I am still looking for it but was reading where a GE Mark 1 RPV was tested by overheating and it vented/failed/leaked through at a specific area of the mating surfaces between the cap and vessel flanges. Yes, while bolt and nuts are typically stronger than welds (larger amounts of metal in play) the report I was reading concluded a predicted design flaw or weak area was proven during the test (guide pin areas?? Not sure). So anyway, I was thinking if a leak occur at the cap and flange, high pressure steam and accompanying heat is capable of 'slicing' through reinforced concrete just depends where it's directed. Also, if I read the diagram correctly, there is a gasket or 'O' ring between the two flanges that would be the first thing to fail with overheating.