Japan Earthquake: Nuclear Plants at Fukushima Daiichi

[SpunkyMonkey]

Re - the mesh you mention - While it's all over the place in the SFP -
given that we do know we are looking at the end of the "Catwalk" that runs along the northern length of the FHM bridge it may be more likely it's mesh from the "catwalk" rather than the FHM proper, the trolley part.

The mesh is very close to the catwalk-railing screen we see on Unit-2 that you posted. But I don't see the same screen on the U3 crane except possibly on the trolley as I showed. Perhaps there are photos that show that mesh screening along the U3 catwalk?

Where is this mesh all over the SFP? I don't recall seeing it before.

The trolley deck I am talking about is that one the video you made of, the thing looked at from the concrete pump over the pool. It has several platforms, the main structure of most can be made out so-and-so, at least enough to make one doubt that they could be what we see in that position of the pool.

Sure it could be a lower deck of the trolley, but I believe the contents there are shaped closer to upper-deck contents.

Sorry, had this second quote attributed to westfield.

 

[MadderDoc]

<..>
Any ideas on what's the large flat object that's smashed down across the crane's bridge?

No, but it has two features which may aid an ID. Along its top, which is well sprinkled with rubble, can be seen three regular cylindrical shapes at regular intervals (wheel pins is my first impression, but those are not always right). Also at the top, but on the side of the object that is facing the camera, there is a regular pattern bright/dark/bright /dark etcetera . The dark fields could represent just cavity in the dark, but they could also be fair dinkum darker surface fields. Considering the regularity the object displays it is not likely a chunk of the wall.

closeup to top of object:

from the distance, and at a different angle:

 

[MadderDoc]

Still about the shield plug video: the spherical distortion and the shaking makes it a mess. After removing them, it's much more better (the result is under it's way up to the youtube).<..>
Ps: the video is available here:

Thank you, you done a fine job there, it so much less straining to eye and brain.

<..> the concrete there has a strange bluish colour, maybe becouse of the temperature?

I think it could be blue pigment from the peeled off paint. Some blue pigments are surprisingly resistant to harsh conditions, including high temperatures. I think we see several signs of a previously applied blue layer of paint on surfaces of the room.

Ps.: it's worth to watch the video at highest resolution and full screen.

Yes indeed, and thank you again.

 

[SpunkyMonkey]

No, but it has two features which may aid an ID. Along its top, which is well sprinkled with rubble, can be seen three regular cylindrical shapes at regular intervals (wheel pins is my first impression, but those are not always right). Also at the top, but on the side of the object that is facing the camera, there is a regular pattern bright/dark/bright /dark etcetera . The dark fields could represent just cavity in the dark, but they could also be fair dinkum darker surface fields. Considering the regularity the object displays it is not likely a chunk of the wall.

Nice images, you can really bring out the details! Right, I don't think this is a wall. I have to say the easiest thing to guess it is is what was on top of the bridge span it is rather on top of now, the trolley. But were there enough trolleys to account for all them that I'm seeing, lol!

Though as you note the trolley had a couple levels, but if that big object is the lower level and the object I suspect is the upper level actually is the upper deck, then it seems maybe they were torn apart. As odd as that sounds, I don't know what to expect at this point. The bridge is in much worse shape than I'd have expected. And how about the crane's ladder entangled with a roof girder! That mess is like a surreal bad dream!

Still about the shield plug video: the spherical distortion and the shaking makes it a mess. After removing them, it's much more better (the result is under it's way up to the youtube). I think the rusty-wet part is where the internal steel door meets its frame. The falling paint (?) chips are removed from the steel door.

Ps: the video is available here:

Nice, thanks for posting! It gives a better feeling for the actual structures as you enter the curved area.

At 1:55 it's 'climbing' upward on the door: you can see that the upper part of the image is moving strange: it's because the upper part is actually the vertical door, while a lower parts are the almost horizontal part of the circular opening.

Exactly! That's what I was trying to describe before, that the camera there is in a 'corner', at the bottom of the screen is a surface that's 90˚ to the surface on top of the screen, and the 'slime trail' is in the corner of these two surfaces. It's just an impression, and it seems stronger in your deshaked video.

 

[westfield]

Yes, 'tis, but that is not where the cask area is.

I didn't phrase that very well. I was saying that IF that image was of U3's SFP (and not some other NPP) then it would be in the SW corner of that SFP. I wasn't neccesarily saying it is unit 3 SFP - The image is from the NHK video you linked to earlier.

I have attached a marked up version of that other photo, which I believe is pointed along the west wall of the SFP3 towards its NW corner. Marked up is the outline of what I believe is the gate to the cask area. My clue was the horizontal X structure at its bottom which I see as the cask support. X always marks the spot :-)

It got clinched only recently, when we were told in no uncertain terms by Tepco that the cask area is in the NW corner -- which was where I had this photo figured out to be from, from combining information from other photos of the pool taken during the MOX refueling & a previous photo tour.

I'm fairly sure the "X" is a reflection of the roof bracing as is the almost vertical line - with the ceiling lights reflected in it.
Maybe it's me but I still can't see anything like a cask loading area in that image.

I'll come straight out with it at my peril, I may end up looking a right wally :)
I know Tepco indicated the U3 SFP cask loading area in the NW corner of the SFP in the recent debris map.
This was a surprise to me because I was sure the only other diagram TEPCO has shown us has the cask area on the Eastern side. That diagram was published by Tepco months ago and I haven't found it again yet. Does anyone remember it or have it?

It was also a surprise to me because I've never seen any alleged image of Unit 3's SFP with the cask area in the NW or NE corner.
Are lots of alleged Unit 3 SFP images wrong, mislabelled? I don't know. I have 3 alleged U3 images showing the cask area in the SE corner.

This is what I'm trying to get to the bottom of.

Finally if U3 SFP cask area was in the NW corner where did it go, it's not a lightweight structure.

For reference Unit 4 SFP - looking to NW corner, cask loading area marked - underwater view of the cask loading area - overhead view of cask area

 

[westfield]

The mesh is very close to the catwalk-railing screen we see on Unit-2 that you posted. But I don't see the same screen on the U3 crane except possibly on the trolley as I showed. Perhaps there are photos that show that mesh screening along the U3 catwalk?

Where is this mesh all over the SFP? I don't recall seeing it before.

Sure it could be a lower deck of the trolley, but I believe the contents there are shaped closer to upper-deck contents.

Sorry, had this second quote attributed to westfield.

Sorry, I said "all over the SFP" but meant all over the FHM, not literally though. - The mesh is on the lower half of all personel walkways and in some areas as machinery guard, there's a lot of it - we don't see a lot of it because images of the northern side of the U3 FHM\Bridge are few and far between.

 

[MadderDoc]

Improving the models (was Re: Japan Earthquake..)

Some interesting stuff there.
This one deals with how to improve modelling to establish core damage etc.
http://www.tepco.co.jp/en/nu/fukushima-np/roadmap/images/m120314_02-e.pdf

This appears to be saying that the model, latest ver., has been tailored to simulate the RPV pressure, and because of that, RPV pressure is well simulated.

 

[SteveElbows]

This appears to be saying that the model, latest ver., has been tailored to simulate the RPV pressure, and because of that, RPV pressure is well simulated.

My understanding of this, based on the limited info available, is that the 2nd graph is a much better fit than the first one because they have now taken account of the fact that operators were manually controlling the HPCI in a manner different to the original assumption. I think I read more about this mode of operation somewhere, will have a look when I find time.

 

[SteveElbows]

My understanding of this, based on the limited info available, is that the 2nd graph is a much better fit than the first one because they have now taken account of the fact that operators were manually controlling the HPCI in a manner different to the original assumption. I think I read more about this mode of operation somewhere, will have a look when I find time.

OK pages 14 to 19 of this document should shed light on their thinking and modified analysis.

http://210.250.7.21/en/press/corp-com/release/betu11_e/images/111222e16.pdf

So a combination of the way the HPCI was operated (e.g. flow rate), other spraying operations, and ruling out the idea of steam leak from HPCI caused them to change their analysis.

 

[westfield]

No, but it has two features which may aid an ID. Along its top, which is well sprinkled with rubble, can be seen three regular cylindrical shapes at regular intervals (wheel pins is my first impression, but those are not always right). Also at the top, but on the side of the object that is facing the camera, there is a regular pattern bright/dark/bright /dark etcetera . The dark fields could represent just cavity in the dark, but they could also be fair dinkum darker surface fields. Considering the regularity the object displays it is not likely a chunk of the wall.

closeup to top of object:

from the distance, and at a different angle:

If you imagine this on the other side of the FHM and with the "stops" torn off it's not a bad match to this.
We can't see if there's a middle wheel in this image which would have helped.

 

[MadderDoc]

My understanding of this, based on the limited info available, is that the 2nd graph is a much better fit than the first one because they have now taken account of the fact that operators were manually controlling the HPCI in a manner different to the original assumption. I think I read more about this mode of operation somewhere, will have a look when I find time.

My understanding it that whereas the model includes code to model behaviour during automatic HPCI operation, it has no code for manual operation. So, during the period of manual operation of the HPCI, for which the modellers have no model, the code is just set to roughly reproduce the actual RPV pressure measurements. Of course one can then say that the 'RPV pressure is well simulated' during that period, but it does seem a bit too pretentious for my taste, and is liable to leave the false impression that the model has been improved by the exercise.

 

[SpunkyMonkey]

There appears to be some large heavy-metal debris in the NE quadrant of SFP3 (the video is motion stabilized and so the frame moves):

"www.youtube.com/watch?v=95mB2B65ZVM&hd=1" ​

This is an area close to the object I suspect is the trolley's upper deck. One only gets the briefest glimpse, and for the life of me I can't make out what those objects are. But I'm sure I can see machined shapes, especially on a piece of debris at approximately 4 o'clock if you imagine a clock face over the video screen. The brief view is clear enough that someone who worked there or who just saw it in a photo could probably recognize it.

Here's a video clip where I attempt to outline some of these heavy-metal objects so you can see which objects I refer to.

 

[MadderDoc]

OK pages 14 to 19 of this document should shed light on their thinking and modified analysis.

http://210.250.7.21/en/press/corp-com/release/betu11_e/images/111222e16.pdf

So a combination of the way the HPCI was operated (e.g. flow rate), other spraying operations, and ruling out the idea of steam leak from HPCI caused them to change their analysis.

Yes, thank you. What I get, of what they did to their analysis is that "the quantity of water injected by the HPCI system was adjusted through the test line to prevent unnecessary halt of the HPCI system due to the reactor water level (L-8)". Which seems fair enough, seeing that was their narrative of what the operator was trying to achieve. The rub, of course, lies in the lack of definition in 'enough to prevent unnecessary halt'. How much is that? Anyway, in the same document we see the result of that flow adjustment in this figure:

Funny how the adjustment of flow needed under the stated criterium: "what it takes to prevent unnecessary halt of the HPCI system due to the reactor water level (L-8)" seems to be identical to the adjustment that would have been needed under the criterium "what it takes to bring the modeled RPV pressure down to the level suggested by the measured data".

 

[SteveElbows]

My understanding it that whereas the model includes code to model behaviour during automatic HPCI operation, it has no code for manual operation. So, during the period of manual operation of the HPCI, for which the modellers have no model, the code is just set to roughly reproduce the actual RPV pressure measurements. Of course one can then say that the 'RPV pressure is well simulated' during that period, but it does seem a bit too pretentious for my taste, and is liable to leave the false impression that the model has been improved by the exercise.

I feel you have made unsafe assumptions here. I do not claim that their model is perfect, and obviously they are looking to change code & variables so that the output of the model more closely matches the actual data. But I don't think they just threw away proper code and fudged stuff to give suitable results, I find it more likely that they did actually make changes to the model or fed a more sophisticated set of data into the model.

Although they are obviously interested in getting the model analysis to match the measured data, I believe their priority is to be able to use the model to test assumptions about the state of various facilities at the plant. They would also like to make the model better for future use. Neither of these things is well served by simply fudging stuff to get a match, as opposed to improving the code used by the model or the data that's fed into it.

We see an example of this later in the document, when they move on to issues of containment pressure of reactor 3. They still can't get their model to match the measured data, so they have to think about what other factors may be involved, with the hope of modelling these factors later on and getting better results. They did not simply fudge things at this point to give a nice clean match between model and measured data.

In another document mentioned recently in this thread, we see that they are looking at improving the core melt model, since they know that the model is vastly oversimplified compared to the realities inside the reactor pressure vessel. But since they lack much in the way of real data about the state of the cores, this is not going to be so easy. I already ranted in the past at how their model seems to give wrong results in terms of RPV damage occurring/time of this occurring, and I speculated that this may be because they have made wrong assumptions about how much decay heat and how much water cooling was available at key points for reactors 2 and probably 3. But until they actually have far more detailed evidence that state of melted fuel & RPV is much worse than model results, I doubt they really know how far away their model is from reality.

 

[clancy688]

I already ranted in the past at how their model seems to give wrong results in terms of RPV damage occurring/time of this occurring, and I speculated that this may be because they have made wrong assumptions about how much decay heat and how much water cooling was available at key points for reactors 2 and probably 3. But until they actually have far more detailed evidence that state of melted fuel & RPV is much worse than model results, I doubt they really know how far away their model is from reality.

I think they've come to the same conclusions themselves. To quote from the recently posted document: "However, calculation results show no damage on RPV."
But on the last page they seem to be convinced that there's at least some of the fuel left the RPV.
So I'd interpret the "However" phrase as "There may be something wrong with the model since it doesn't show any RPV damages, but we're convinced there are."
Of course I can be mistaken, but that's at least what it sounds like to me. Moreover, you all know the radiation readings from the Unit 2 PCV. Is it possible that there are such high readings if there's absolutely no fuel escape? That should be obvious to TEPCO as well.

 

[MadderDoc]

I feel you have made unsafe assumptions here. I do not claim that their model is perfect, and obviously they are looking to change code & variables so that the output of the model more closely matches the actual data. But I don't think they just threw away proper code and fudged stuff to give suitable results, I find it more likely that they did actually make changes to the model or fed a more sophisticated set of data into the model.

Although they are obviously interested in getting the model analysis to match the measured data, I believe their priority is to be able to use the model to test assumptions about the state of various facilities at the plant. They would also like to make the model better for future use.

Perhaps. In the present case, though, they did nothing to the model, only fed new data to it. Perhaps I am too cynical in thinking that it is not coincidental that that made the model match the measured data to perfection. Perhaps they truly just fed the data they assumed to be most likely, in an effort to validate the model and hoorah, it came out exactly reproducing the measured data, the modeller's wet dream.

Neither of these things is well served by simply fudging stuff to get a match, as opposed to improving the code used by the model or the data that's fed into it.

We see an example of this later in the document, when they move on to issues of containment pressure of reactor 3. They still can't get their model to match the measured data, so they have to think about what other factors may be involved, with the hope of modelling these factors later on and getting better results.

Yeah, right. :-) And in that process all discrepancies are equal but some discrepancies are more equal than others. In their latest model of the PCV pressure of unit 3, I see they have meticulously made assumptions of S/C vent operations matching every remaining unexplained significant change in pressure trends -- except the precipitous pressure drop at the time of the explosion, and the curious incident of the S/C vent some hours earlier.

They also seem to have gotten into trouble by assuming a lower decay heat than in previous attempt. Now the model undershoots, appears insensitive to interventions, and cannot reproduce the measured pressure increase rates in the PCV.

Although this latest version produces a much poorer fit than the previous attempt they seem quite happywith it, except they have one outstanding issue, an unexplained pressure increase in the initial phase under RCIC operation, which they seem to speculate might be fixable by assuming S/C stratification.

 

[MadderDoc]

There appears to be some large heavy-metal debris in the NE quadrant of SFP3 (the video is motion stabilized and so the frame moves):

"www.youtube.com/watch?v=95mB2B65ZVM&hd=1" ​

This is an area close to the object I suspect is the trolley's upper deck. One only gets the briefest glimpse, and for the life of me I can't make out what those objects are. But I'm sure I can see machined shapes, especially on a piece of debris at approximately 4 o'clock if you imagine a clock face over the video screen. The brief view is clear enough that someone who worked there or who just saw it in a photo could probably recognize it.

Here's a video clip where I attempt to outline some of these heavy-metal objects so you can see which objects I refer to.

I think those are parts of the roof covering, they seem to have been caught above the level of the pool, and be of low thickness. You can see them from another angle later in the video. Straight underneath them there are more parts of the FHM than you can point a stick at, here, there and everywhere there are glimpses of cross beamed greenish structures through all that stuff hunkering above. In a way it was predictable all along that there would big objects to find in the east end of the pool. Otherwise, what kept us from seeing the water surface, what dampened the heat signature. Certainly Tepco would have a pretty good picture of what is in the east end of the pool, but we have to date not been let in on it. Maybe they think we are not interested in that end of the pool? But we are, we are.

 

[MadderDoc]

If you imagine this on the other side of the FHM and with the "stops" torn off it's not a bad match to this.
We can't see if there's a middle wheel in this image which would have helped.

Right, it is not a bad match at all. It would be reasonable to think the other side is similar but that side is well hidden in all photos. Anyone got drawings of a classical Toshiba FHM? ;.)

We should also be cautious to accept those tree wheel pins, and we can certainly not exclude that e.g. the one of the middle is just a piece of rubble caught in our minds fabrication. Maybe they all are? It is difficult to not see pins after first having seen them, though.

 

[SpunkyMonkey]

I think those are parts of the roof covering, they seem to have been caught above the level of the pool, and be of low thickness. You can see them from another angle later in the video. Straight underneath them there are more parts of the FHM than you can point a stick at,

Right, but I'm referring to those underneath objects, which are clearly not roof girders or sheathing strips. I outline them here. The one object at about 4 o'clock could be identifiable had we the right pre-tsunami photo or an FHM expert on hand.

In a way it was predictable all along that there would big objects to find in the east end of the pool. Otherwise, what kept us from seeing the water surface, what dampened the heat signature.

Lol, that's exactly what I said on another forum, that in hindsight it should have been obvious that something large (and what else if not the FHM) is filling the pool given that something obscures the full heat signature and that roof debris 'floats' over the western half of the pool as can be seen with a "www.youtube.com/watch?v=zOthckEvROk#t=4s" . All that debris couldn't be floating on water, lol.

 

[SteveElbows]

I think they've come to the same conclusions themselves. To quote from the recently posted document: "However, calculation results show no damage on RPV."
But on the last page they seem to be convinced that there's at least some of the fuel left the RPV.
So I'd interpret the "However" phrase as "There may be something wrong with the model since it doesn't show any RPV damages, but we're convinced there are."
Of course I can be mistaken, but that's at least what it sounds like to me. Moreover, you all know the radiation readings from the Unit 2 PCV. Is it possible that there are such high readings if there's absolutely no fuel escape? That should be obvious to TEPCO as well.

I agree, I am sure they already thought for a long time that the model results are wrong on this, as soon as evidence came out which made no RPV damage seem implausible.

However in the early days they tried to stick to models that didn't show much damage at 2 & 3, and talk as if this was real, and this formed part of a wider picture of positivity which damaged TEPCOs credibility by being so out of whack with reality. Gradually other peoples analysis along with new facts from the ground meant they had to come onboard with public conclusions that the cores were worse than first indicated.

At this point I think they would still like to lean towards more positive assumptions until evidence kills such assumptions, but they would probably like these core melt models to show something more realistic. But how far to go? I think its fair enough that they have assumed there is still some fuel in the RPV based on temperature of RPV as water spraying from several sources was adjusted over time, but I would not care to estimate quite what percentage of the core remains there or anywhere else. They did not avoid admitting that all of reactor 1s fuel melted according to their model, but when it comes to a question of percentages in the other reactors I am not sure I would trust estimates at this point anyway, so I don't know what they should aim for if they are tinkering with input parameters.

Certainly I don't find it easy to talk much more about these models or assumptions without actually being able to see the models in detail themselves, or even run them with different data. I moan about the potential for decay heat & water injection rates & time periods for reactor 2 to be all wrong and to have resulted in the bad model results, but without actually being able to adjust some of the data to compensate for the potential flaws I speculated about, and see what the results are, I cannot test my ideas properly. Id love to know for example how much they would need to tamper with the timescale for how long they assume water was not removing heat from reactor 2 for, to get a result that indicated say 50% fuel melted and left the RPV.

 

[MadderDoc]

Certainly I don't find it easy to talk much more about these models or assumptions without actually being able to see the models in detail themselves, or even run them with different data. I moan about the potential for decay heat & water injection rates & time periods for reactor 2 to be all wrong and to have resulted in the bad model results, but without actually being able to adjust some of the data to compensate for the potential flaws I speculated about, and see what the results are, I cannot test my ideas properly. Id love to know for example how much they would need to tamper with the timescale for how long they assume water was not removing heat from reactor 2 for, to get a result that indicated say 50% fuel melted and left the RPV.

Now I doubt that timescale would be a parameter for the model. I should think to produce RPV damage one would have to change other parameters or change the set of assumptions such as to affect that time scale.

My basic issue with the Unit 2 model is that it does not assume RPV damage at about midnight between March 14th and March 15th, although RPV damage is suggested by the data: - we see the RPV pressure fluctuations come to a final halt, while drywell pressure and cams increase suddenly.

Hence the model leaves this abrupt D/W pressure increase unexplained -- in the previous, as well as in the latest version of the model. In the previous version the model didn't produce a pressure increase at all, however the latest version has in some unknown way been made to produce it.

As regards the sudden D/W pressure drop in the morning of March 15th, the latest version now assumes that the D/W pressure dropped due to D/W failure, which seems to be a reasonable assumption.

 

[SpunkyMonkey]

Btw, it's curious to note how perfectly the model of a Mark I meltdown by Ott et al happens to correspond to Unit 3's RPV pressure data:

After I graphed this remarkable correspondence I found that this sudden RPV pressure drop at Unit 3 shown above was attributed to manual activation of a pressure relief valve, so I dropped this graphic from my argument for a primary-containment explosion.

Would there be any way to falsify Tepco's claim that this perfect replication of the predicted Mark I melt-through RPV pressure data was just a coincidental artifact of manual operations?

For reference:

 

[zapperzero]

I don't see how there could be a melt-through that does not dramatically affect the pressure in the drywell.

 

[SteveElbows]

Now I doubt that timescale would be a parameter for the model. I should think to produce RPV damage one would have to change other parameters or change the set of assumptions such as to affect that time scale.

I wasn't meaning to imply that I thought there was one time parameter they would need to change.

I was talking about how I consider that a possible major reason why their core melt analysis is wrong, is that they have not used an appropriate timescale when considering how much decay heat was not dealt with by injected water. I ranted about some of the detail of this some weeks ago but cannot find my post right now, I shall dig up the relevant chart and then talk more about what I mean.

 

[MadderDoc]

I don't see how there could be a melt-through that does not dramatically affect the pressure in the drywell.

My thought too. That would force us to reject melt through, and accept as per default that the RPV pressure dropped due to the opening of a SRV, since the pcv pressure did increase but not drastically. However, a complicating factor is that primary containment vessel venting was initiated in close connection to the observed depressuring of the reactor vessel, which would have suppressed the pcv pressure increase.

 

[SteveElbows]

OK I found my original reactor 2 decay heat rant:

https://www.physicsforums.com/showthread.php?p=3837167

Since I didn't get much feedback about this at the time, and my post was rather long and perhaps badly worded, I will make another attempt now.

I have paid more attention to the situation at reactor 2 than reactor 3 so that's the one I pick on for this example. Please see the attached graphs which come from http://www.tepco.co.jp/en/nu/fukushima-np/images/handouts_111130_07-e.pdf

So my issue is with the timescale used to calculate the decay heat that needed to be dealt with at reactor 2 during the time that water level was dropping and alternative water injection could not be confirmed.

For the start time they used 13:25 on the 14th. The water level actually started to fall before this time, 13:25 is when they officially declared that there was a new incident that required notification according to the regulation, the incident being a drop in coolant level. Water level actually started dropping at 12:00. I suppose so long as the figure they used for amount of water in the reactor actually matched what was in the reactor at 13:25, and not the higher amount that was present at 12:00 or before, then this issue shouldn't be a problem. But seeing as how the amount of heat removal by water they have used in the second graph is identical for reactors 2 & 3, I have some doubts as to whether they have done this stuff correctly. I will do some more checking into the detail of this.

My bigger issue is with the end time. I went into detail on this in my previous post about the issue, but basically I have concerns that they chose the earliest possible moment for water injection beginning, and seem to have assumed that from this moment onwards there was enough water being injected to cover all heat produced from this time onwards. Given the numerous issues they encountered when injecting water at the reactor, especially in the first hours after starting this operation, I have big concerns that they are missing a load of uncooled decay heat from their model, and that's why its producing inaccurate results in regard to RPV damage etc. Its no wonder their model shows this, seeing as how the 2nd graph shows that decay heat was more than dealt with by the water that was in the reactor, leading to all of the water that surrounded the fuel being used up, but some being left below the bottom of the fuel.

 

[MadderDoc]

OK I found my original reactor 2 decay heat rant:

https://www.physicsforums.com/showthread.php?p=3837167

Since I didn't get much feedback about this at the time, and my post was rather long and perhaps badly worded, I will make another attempt now.

I have paid more attention to the situation at reactor 2 than reactor 3 so that's the one I pick on for this example. Please see the attached graphs which come from http://www.tepco.co.jp/en/nu/fukushima-np/images/handouts_111130_07-e.pdf

So my issue is with the timescale used to calculate the decay heat that needed to be dealt with at reactor 2 during the time that water level was dropping and alternative water injection could not be confirmed.

For the start time they used 13:25 on the 14th. The water level actually started to fall before this time, 13:25 is when they officially declared that there was a new incident that required notification according to the regulation, the incident being a drop in coolant level. Water level actually started dropping at 12:00. I suppose so long as the figure they used for amount of water in the reactor actually matched what was in the reactor at 13:25, and not the higher amount that was present at 12:00 or before, then this issue shouldn't be a problem. But seeing as how the amount of heat removal by water they have used in the second graph is identical for reactors 2 & 3, I have some doubts as to whether they have done this stuff correctly. I will do some more checking into the detail of this.

My bigger issue is with the end time. I went into detail on this in my previous post about the issue, but basically I have concerns that they chose the earliest possible moment for water injection beginning, and seem to have assumed that from this moment onwards there was enough water being injected to cover all heat produced from this time onwards. Given the numerous issues they encountered when injecting water at the reactor, especially in the first hours after starting this operation, I have big concerns that they are missing a load of uncooled decay heat from their model, and that's why its producing inaccurate results in regard to RPV damage etc. Its no wonder their model shows this, seeing as how the 2nd graph shows that decay heat was more than dealt with by the water that was in the reactor, leading to all of the water that surrounded the fuel being used up, but some being left below the bottom of the fuel.

Sorry I did not pay attention when you first raised the issue. As regards evaluating the potential for fuel damage this method is of course deeply flawed. Standing alone it looks like a typical piece of corporate science, suggesting, albeit not claiming a particular conclusion that does not quite follow from the information presented. The question here is of course: so what happened after?

I think the model we were looking at would be highly sensitive to the assumptions made as to how much water reached the core right after the rpv was depressurised, and sea water injection was being attempted. A few m3 here and there could mean the difference between having a melt-down and having a melt-through. In the latest version of the model which has no rpv damage, adjustments seem to have been made to produce a subtly higher water level in the RPV during those fateful hours.

 

[MadderDoc]

SRV question

I read somewhere in a Fukushima accident analysis, that a manually opened safety relief valve would shut due to the sheer weight of the valve, once the pressure drops below about 0.3-0.4 MPa.

What status would that leave the valve in? Would it just reopen on renewed pressure build-up, or would it need another manual actuation to become open ?

 

[SpunkyMonkey]

My thought too. That would force us to reject melt through, and accept as per default that the RPV pressure dropped due to the opening of a SRV, since the pcv pressure did increase but not drastically. However, a complicating factor is that primary containment vessel venting was initiated in close connection to the observed depressuring of the reactor vessel, which would have suppressed the pcv pressure increase.

Though the Ott et al melt-through model I just posted says RPV pressure drops to equalize with the containment. Isn't that's pretty much what happened (albeit containment pressure was a bit higher thereafter)?

When the SRV was opened, was it never closed again? If so, is that usual as a risk-mitigation strategy under the circumstances? Wouldn't it just be opened for a period? Is there a standard protocol for this action? RPV pressure never rebounded, but was like a flat tire forever thereafter.

 

[MadderDoc]

Though the Ott et al melt-through model I just posted says RPV pressure drops to equalize with the containment. Isn't that's pretty much what happened (albeit containment pressure was a bit higher thereafter)?

I see what you mean, yes, but I think that effect could be produced also by a leak in the reactor above the fuel level.

When the SRV was opened, was it never closed again? If so, is that usual as a risk-mitigation strategy under the circumstances? Wouldn't it just be opened for a period? Is there a standard protocol for this action? RPV pressure never rebounded, but was like a flat tire forever thereafter.

Yes, it seems we ended up eventually with a RPV that couldn't build up pressure, however, when that state was reached is not immediately clear, hence my interest in the inherent properties of a manually opened SRV.

 

[MadderDoc]

From the descriptions in the accident analysis report, I get the impression that the SRVs talked about would be of a type which is called dual function direct acting valves. They are called dual function because they are able to be opened in two ways: a) automatically by the pressure reaching a set value (pressure mode), or b) by manual or remote actuation (actuation mode). They are called direct acting, because actuation operates directly on the valve body, not through an intermediate system.

I have attached a technical diagram of a dual function-direct acting valve cropped from
NRC report on SRV performance at US nuclear plants

In pressure mode, the valve would open automatically on exceeding the set point, which would in the case of Fukushima be around 7-8 MPa.

For manual or remote actuation, compressed air is needed to drive a pneumatic piston at the valve, and power is needed to open a valve that let compressed air flow to that piston. Then, the piston through a lever applies an opening force to the valve body, sufficient to open the valve -- on the condition that it can be assisted by an inlet steam pressure at or above about 0.7 MPa. If inlet pressure is below about 0.7 MPa. actuation will fail due to lack of opening force.

However, once successfully opened due to actuation -- (my question is, do I understand this correctly?) -- there would seem to be nothing that keeps the valve from closing except the pressure from the inlet steam. The piston-lever arrangement seems designed only to flip the valve open, not to keep it open. Once pressure falls below a certain level (about 0.3-0.4 MPa), the valve is meant to reseat, and to stay closed, and to open again only on actuation or by reaching its high pressure set point.

 

[SteveElbows]

A couple of possibly silly questions about detected level of hydrogen in reactor 2.

Firstly how come they don't seem to have increased the rate of nitrogen injection into the reactor beyond 5.0N m³/h for reactor 2? I am under the impression that they originally reduced it when they were getting things ready to do the endoscope investigation, and I think they turned the injection to PCV back up since then, but why they haven't done this with the reactor too?

Secondly the measured level of hydrogen seems to have ten increasing lately, and the 5am report for today seems to indicate its reached 0.42 vol %. What could explain this, and how high should the level reach before I express concern?

(Current data I am looking at is http://www.tepco.co.jp/nu/fukushima-np/f1/images/2012parameter/12042805_table_summary-j.pdf )

 

[zapperzero]

Does "cold shut down" also imply no active cooling required?

Nope.

 

[mheslep]

And? How far does the decay heat have to drop to eliminate the need for active cooling? Or more to the point, what is the Watts/deg C heat flux path to ambient?

 

[jim hardy]

@ madderdoc -

However, once successfully opened due to actuation -- (my question is, do I understand this correctly?) -- there would seem to be nothing that keeps the valve from closing except the pressure from the inlet steam. The piston-lever arrangement seems designed only to flip the valve open, not to keep it open.

did something written lead you to that thought?
It would be a surprise to me. It looks like a direct operated valve with a pneumatic cylinder to actuate the lever, which pulls directly on the spindle


If steam opens it it'll close when pressure comes down.
But if the handle opens it it'll stay open until handle is released. i think.


http://www.tycoflowcontrol.com/valves/Images/CROMC-0292-US.pdf

That bellows is not a pilot, it's there to make the valve insensitive to back pressure. see 2.4.3 of link

old jim