Talk:Prompt criticality

Latest comment: 2 years ago by Gah4 in topic stability

Game?

edit

It seems that the "game" at Prompt Critical Video Game has nothing to do with nuclear physics, and just uses a cool-sounding word. Does anyone object to deleting it?

68.57.72.229 (talk) 19:01, 25 January 2009 (UTC)Reply

Disputes

edit

The statement "and the increase in the reaction will be extremely rapid and uncontrollable, causing an explosion within a few milliseconds" is also untrue, as the reaction is not uncontrollable, just rather difficult to control (it can be controlled using careful design, material properties, etc) and an explosion is only a potential result, not an assured one (again, ref. GODIVA) —Preceding unsigned comment added by 199.46.196.231 (talk) 16:06, 13 July 2010 (UTC)Reply

The statement "Prompt-critical assemblies are only used in nuclear weapons." is provably false as experimental facilities such as the Godiva series at LANL performed numerous experiments wherein the reactivity was well in excess of prompt criticality (see Wiki page on Godiva, and external links referenced in that page) —Preceding unsigned comment added by 199.46.196.231 (talk) 16:00, 13 July 2010 (UTC)Reply

Previously a disputed article -- comments read as follows -- This page currently needs a complete rewrite. For example, it confuses the terms prompt critical and supercritical, and talks of a delayed neutron flow control system which sounds impressive but is I suspect meaningless.

See critical mass.

However there's lots of good stuff here as well. Andrewa 02:26, 29 Nov 2004 (UTC)

See this external link for a good definition of supercritical (and I think they mention prompt-critical too). Andrewa 02:38, 29 Nov 2004 (UTC)

Listing on Wikipedia:Accuracy dispute is optional AFAIK, and I don't think necessary in this case. I'll attend to the problems in a few days' time hopefully, including renaming to prompt criticality and deleting the resulting redirect. Andrewa 02:45, 29 Nov 2004 (UTC)

OK, I've attempted a rewrite. Our coverage of the whole topic is still not terribly good. Critical mass is still a worry. But I've removed most of the overgeneralisations and fundamental misunderstandings I think. Andrewa 18:13, 29 Nov 2004 (UTC)

Thanks for the rewrite. I'm by no means an expert in nuclear engineering, but when I saw the words "went prompt critical" in the PL-1 article, I thought it would be nice to explain the term and did so to the best of my ability. Unfortunately in the process I confused myself thinking supercriticality was more powerful than prompt criticality. As for delayed neutron flow control system, that's what a graphite rod assembly is, though admittedly the name is too pompous and inappropriate for it. --Azazello 14:03, 1 Dec 2004 (UTC)

Sorry if my words above were a bit harsh, you seem to have the spirit of the Wiki well in hand! I think the results in this case are getting there.

I'm not a nuclear engineer either, but I have one in the family, worked in the industry for some years myself (in the computer department) and have a bee in my bonnet about some of the issues.

Not convinced about the graphite rod assemblies. Control rods are generally cadmium or boron neutron absorbers. In theory you could have a graphite rod assembly that made the pile more reactive when inserted, and use it to shut the pile down when it was removed, but I don't think that's ever been done for a power reactor, at least not deliberately. Something like it was done unintentionally with the RBMK, that was one of the design errors and was what actually triggered the first explosion, which was possibly inevitable by that time anyway. Andrewa 18:48, 1 Dec 2004 (UTC)

Cleanup and linguistic edits, plus incorporation of the concepts of prompt/delayed neutrons to distinguish supercriticality and prompt criticality (based on other versions floating around the web), otherwise without (I hope!) substantial content change. I have not removed the cleanup tag, since I notice that the article is scheduled for renaming to "... criticality", not "... critical" (and I don't know how to do that), and I have left the stub tag in place. -- Rudolf Cardinal, 12 June 2005.

Further additions, defining prompt criticality formally. Cleanup/stub tags tentatively removed. -- Rudolf Cardinal, 13 June 2005.

Reversion of part of opening paragraph - a previous edit made it sound as if a prompt critical assembly could be critical, or supercritical, or go bang. A prompt critical assembly, as I understand it, is a supercritical assembly that is supercritical needing only the contribution of prompt neutrons, and goes bang. You can't have a "merely critical prompt critical assembly", as I understand it - so a prompt critical assembly can't cause either a self-sustaining fission reaction or an exponential increase and an explosion. I've added a further explanation of prompt criticality as a subset of supercriticality, before the main explanation of subcritical/critical/supercritical/prompt critical. -- RudolfCardinal 17:26, 1 February 2006 (UTC)Reply

Prompt critical means k=1, no exponential growth. Mostly it is used as the dividing line between prompt subcritical and prompt supercritical. The nice thing about delayed critical is that mechanical systems (control rods) are fast enough to keep it there. It would be difficult, but maybe not impossible, to keep a system exactly at prompt critical. Even more, though, what is important is the reactivity of the whole system. Some parts will be above, and some below, keeping the average where you want it. (For both delayed and prompt.) In theory, you can have a "merely critical prompt critical assembly", practically, it is hard to say. Gah4 (talk) 17:43, 24 August 2018 (UTC)Reply

Some rewriting

edit

I did some major rewriting to try to keep the article on-topic and make it easier to understand for those not already familiar with the topic. I may have lost some relevant bits in the process so please do add them if you feel they should remain. I will try to split the article into sections as well. J.Ring 22:45, 8 September 2006 (UTC)Reply

I will review over the weekend. Thanks for the cleanup efforts! Georgewilliamherbert 00:27, 9 September 2006 (UTC)Reply

Significant errors

This paragraph is abiguous/wrong: "In a prompt-critical (k > 1) assembly, the neutron activity increases exponentially by the factor k, and will cause an explosion if kept prompt-critical for long enough . . . "

This describes a supercritical condition, not a prompt-critical one (k(prompt) > 1 for prompt criticality). Supercriticality is not very relevent to prompt-criticality. Casual readers will be confused in the difference between prompt and delayed neutrons and their impact on criticality. It would be better to describe prompt-criticality here in terms of fast neutrons. (The section above is anonymous)

Errors in the reaction rate

There are some errors in the description of the reaction rate.

In a prompt-critical (k > 1) assembly, the neutron activity increases exponentially by the factor k, and will cause an explosion if kept prompt-critical for long enough. In contrast, in a subcritical assembly, each fission event triggers, on average, less than one new fission event (k < 1) and the activity decreases exponentially by the factor k. For example, if 2.4 neutrons are released per fission event, then if the probability of a neutron causing a further event is less than 1/2.4 = 0.42, the assembly is sub-critical and the neutron activity decreases exponentially with time.

It is not true that for a sub-critical assembly that the activity decreases exponentially with time; that would imply that the activity will drop asymptotically to zero. There is a steady-state activity whose level will increase with increasing k. This is due to the fact that spontaneous fission neutrons will trigger other neutrons; the total number of induced fission neutrons induced by a single spontaneous event is a function of k. When k>1, then this number is infinite and that results in an uncontrolled chain reaction.

If you inject a burst of neutrons into a sub-critical assembly, you will get a burst of activity that drops exponentially with time back to that steady-state level of activity.

I would therefore suggest rewriting this paragraph thus:

In a supercritical (k > 1) assembly, the neutron activity increases exponentially with time. If the assembly is supercritical but not prompt-critical, the increase will be fairly slow (e.g. double every few minutes). This was the case for the first controlled nuclear reaction at Chicago_Pile-1 built by Enrico Fermi. If the assembly is prompt-critical, the increase will extremely rapid and will cause an explosion if kept prompt-critical for long enough (meaning a few millionths of a second or less).

In contrast, in a subcritical assembly, each fission event triggers, on average, less than one new fission event (k < 1). While the level of neutron activity increases with increasing values of k, it will approach some steady state value and stay there, rather than increasing without limit like a supercritical assembly.

DavidGauntt 22:39, 14 June 2007 (UTC)DavidGaunttReply

I would agree with most of your recommended changes but disagree with citing the Chicago-Pile-1 as the only example of a supercritical reactor. ALL reactors must go supercritical to raise power. It is generally done in a controlled manner with k only slightly greater than 1.

AdamGott 14:45, 24 October 2007 (UTC)AdamGottReply

I added this without the Chicago_Pile-1, and clarified the exponential decrease.--Patrick 14:31, 24 October 2007 (UTC)Reply

Confusing prompt critical with supercritical

edit

The last two paragraphs of Prompt critical accidents say:

There had also been prompt criticality accidents during the early development of nuclear weapons, with the two most notable cases occurring in 1945-46.
On June 16, 1958, a supercritical portion of highly enriched uranyl nitrate was allowed to collect in a drum causing a prompt neutron criticality in the C-1 wing of building 9212 at the Oak Ridge National Laboratory Y-12 complex in Oak Ridge, Tennessee. It is estimated that the reaction produced   fissions. Eight employees were in close proximity to the drum during the accident, receiving neutron doses ranging from 30 to 477 rems. No fatalities were reported.[1]

But above, it says "If the assembly is prompt-critical, the increase will be extremely rapid and will cause an explosion if kept prompt-critical for long enough (meaning a few millionths of a second or less)." In neither case was there an explosion; the only time the Demon Core actually went prompt critical was when it was used in a nuclear bomb. Likewise, the accident at Oak Ridge didn't produce an explosion, or even enough radiation to kill a person. Of all the accidents on the cited page, I don't know why this one was singled out.--Prosfilaes (talk) 01:54, 11 April 2009 (UTC)Reply

No, the issue is that for the SL-1 huge reactor plant with 14 kg of U-235 inside, the total fission output estimate (from May 1961 report) was   fissions, which resulted in a 45 MW-sec estimate. This was a preliminary estimate, and they finally estimated 133 MW-sec of energy in 1962 after tearing apart the reactor. So perhaps SL-1 emitted 1.8 x 133 / 45 =   fissions during that excursion (the final report of 1962 does not say how many fissions).
I want you to now compare the order of magnitudes of the 2 situations... the first is a massive nuclear reactor plant rated for 3 MW. And then you have an Oak Ridge accident that generates 20% of the fissions that destroyed SL-1? How is that not prompt critical?!? What else would explain a few quintiillion fissions to you? A "supercrititcal" core, yes, but it didn't take 2 hours to generate that many fissions. Things that slowly get hot tend to overheat and blow themselves apart. The Oak Ridge event was a criticality excursion, meaning it did its thing very fast. You have to realize that nothing bad happened at Oak Ridge because there was not a container which pressurized itself until it blew itself apart. SL-1 "jumped" 9 feet into the air due to Newton's laws of motion and it only had 5 times more fissions than Oak Ridge.
If Oak Ridge had been anything else but a prompt critical core, it would have just heated up, set off some detectors, melted some things, etc. But people received nearly deadly levels of gamma/neutron doses in a split second. And you are right about the other incidents reported at the reference cited... lots to choose from. Those other excursions, like the others at Oak Ridge, were certainly prompt critical, or otherwise 15 grams of uranium would not have emitted  . See List of criticality excursions I like to saw logs! (talk) 09:19, 16 April 2009 (UTC)Reply
The Oak Ridge accident wasn't a prompt critical core; the description is "a 55-gallon stainless steel drum" filled with contaminated water. Unshielded nuclear reactions are dangerous, and don't have to be prompt critical to give off lethal levels of radiation. Harry Daghlian received a lethal dose of radiation from a core that was cold enough to disassemble by hand. The Richland, Wash., Apr. 7, 1962 excursion (same document) took 37 hours to self-terminate. Criticality excursions can be slow. There's no reason to think this one was fast, and the report gives no times. If we generously assume one second, then let's compare the real orders of magnitude of the 2 situations. One was 5.3 * 10^18 fissions in 4 milliseconds, or 10 ^ 20 fissions per second. The other was 1.3 * 10^18 fissions in one second, which is two orders of magnitude less.
There are three events described as prompt critical in that document; one is hardly described, one is under controlled circumstances, and the third Oak Ridge, Tenn., Feb. 1, 1956: "considerable fuel was displaced from the reactor". If anything is meant by prompt critical as opposed to supercritical, it's to be measured by the speed and explosiveness of the reaction, not the number of fissions. --Prosfilaes (talk) 17:15, 16 April 2009 (UTC)Reply


Let's look at energy rather than power for a moment. From Radiochemistry and nuclear chemistry By Gregory R. Choppin, Jan-Olov Liljenzin, Jan Rydberg, page 519, found at Google Books, [1] we see that prompt neutrons and gammas per fission amount to 12 MeV. Those would typically flee the scene of a barrel or a vat filled with water. The fission products, on the other hand, amounting to 165 MeV per fission, remain in the vat.
So a cursory examination of the amount of heat generated in these accidents would reveal that SL-1 produced about 140 MJ if we use the 5.3 E18 figure I concocted above (using 165 MeV/fiss and 1.6 E-13 J/MeV). Let's say I have a frozen pizza that takes 140 seconds to cook in a 1000 W microwave, or 140 kJ per pizza. so SL-1 could have cooked 1,000 frozen pizzas with the energy of its fission products.
Now, if we scale that down to some of the other "accidents" reported earlier, we can see that we quickly stop being able to cook so many pizzas. The 1.3 E18 fissions? Only 245 pizzas. This amount of heat is not all that big of a deal, even if it emits deadly neutron and gamma radiation. We also would note that the proximity to such an excursion would be the most significant criteria as far as exposure, so obviously people nearby wouldn't always die... there would be little danger from the radiation at 5 meters, and essentially no danger from the heat.
Let's look at the 245 pizza excursion. What I call 245 pizzas is 34 MJ of energy. If we took 50 gallons (call it 189 liters or 189 kg) of pure water at 25 °C, added 34 MJ, what would be the final temperature? The answer is a 43 °C temperature rise... no big deal. It doesn't even boil the water... 68 °C. And the water shields a lot of the radiation. It's not a major industrial accident.
So it comes down to whether the 55-gallon container was prompt critical or whether it needed moderated, delayed neutrons to do the excursion. But it doesn't seem to me that the excursion lasted very long. I like to saw logs! (talk) 06:47, 18 April 2009 (UTC)Reply
"...it is possible that the system reactivity slightly exceeded prompt criticality before the first excursion..."[2] --Quote from "A Review of Criticality Accidents," Los Alamos National Laboratory, 2000, page 14. I am changing the article to include this once again. I like to saw logs! (talk) 08:23, 18 April 2009 (UTC)Reply

Fukushima Dai-ichi #3

edit

I have added in the Fukushima accident as a potential prompt criticality accident. I noted that it had previously been added but removed without explanation. The velocity of the detonation at Fukushima Dai-ichi #3 is some evidence for there having been a prompt criticality as is the lack of confinement and the damage to #4. While these are not definitive, I believe that the statement should be included with proper clarification unless there is strong evidence to the contrary.


  1. ^ Walker G. "Criticality Accidents". Trinity Atomic Web Site. Retrieved 2007-06-17.
  2. ^ http://www.csirc.net/docs/reports/la-13638.pdf

References

edit

I noticed that the two unnumbered references are broken. One is a 404 and the PDF link redirects to a list of links at the Department of Energy Technical Standards Program and not to a file. Also, in general, there aren't very many inline references in this article for the amount of information presented. I don't know enough about the subject to work on it though.Xblkx (talk) 18:51, 31 May 2012 (UTC)Reply


Does that mean it explodes?

edit

So, if one were to stack up a prompt-supercritical mass of enriched uranium or plutonium, does it mean that you'd get a nuclear explosion without employing any additional "helpers", such as neutron reflectors chemical high explosives? Does this mean that anybody in possession of a prompt-supercritical amount of fission material can cause a nuclear explosion? If so this information could be included. And if it's not the case, it might be said why not.

Another question, would it be possible to build a prompt-critical stack from (a very large amount of) unenriched uranium, or from thorium? Thanks. -- 77.21.99.8 (talk) 11:45, 6 February 2012 (UTC)Reply

It is not just the quick release of energy (see my discussion above about frozen pizzas and barrels full of water) that causes an explosion. You also have to contain the energy in some way to give it a reason to explode. You can confine it in some way, as in SL-1, which "exploded." But the SL-1 explosion was so limited in scope, no one heard it a few miles away. A nuclear reactor contains and confines nuclear excursions, but great amounts of pressure aren't built.
So we need to look at pressure. An enclosure of steel can contain energy, and when it bursts, there is an explosion. But to obtain blast effects (the purpose of a military weapon), you will need enormous levels of pressure that can confine the energy. This is the principle of a basic "Fat Man"-style explosion in which shaped charges provide confinement.
There are two major principles that will make for a nuclear explosion: a supercritical assembly and pressure confinement. A few other factors are at work in minor roles. See them at Critical mass... I like to saw logs! (talk) 02:06, 30 January 2013 (UTC)Reply
On the first question, just because a system is prompt-supercritical does not mean it will explode. The system must remain prompt-supercritical for a long enough time the power increases to the point that it will explode. There are reactors (eg. TRIGA reactors) that were designed to be able to be put in a prompt-supercritical state, and then because of large negative coefficients of reactivity, stop increasing in power and actually shut themselves down. There have also been a number of prompt-supercritical experiments and accidents (eg. the Demon Core, SPERT tests) which did not cause nuclear explosions.
On the second question, there is no critical mass for thorium, it can't go critical alone. Thompn4 (talk) 16:07, 20 May 2016 (UTC)Reply
If you just put together, slowly, a prompt supercritical mass, it will heat up slowly, melt, and then stop the reaction. That is why gun design (like little boy) doesn't work for Pu. (Faster than a speeding bullet, is still too slow.) To make an effective bomb, you have to be prompt supercritical for a long enough time. And no, an infinite mass of unenriched Uranium won't explode. Too many neutrons first hit U-238 without enough energy to fission it. The design of power reactors has the rods sized and spaced such that neutrons hit moderator, slow down below the U-238 absorption energy, before they reach U-235 or U-238. In addition, it would be difficult to speedily assemble a very large mass of natural uranium. Gah4 (talk) 18:08, 24 August 2018 (UTC)Reply

PWR and prompt criticality

edit

"Many reactor designs succeed in making prompt criticality practically impossible. Some pressurized water reactors, for example, do not contain enough fuel of high enough enrichment to make a prompt critical assembly with the materials in the core." I agree that you cannot take out the core put it into a big lump and make a critical assembly. However, this fact is irrelevant when talking about reactor safety. If control rods / chemical shims are not used then there can be a very large and rapid increase in power. The difference between delayed and prompt criticality is about the ability to control the nuclear reaction not whether dangerous power surges can occur.

"Such reactors can still overheat and even melt if the ability to cool them is lost (a loss-of-coolant accident)" The main safety feature of a PWR is that you hopefully have a negative void coefficient.

"but they are unlikely to explode" Prompt criticality is not the only thing that can make reactors explode - Fukushima. The paragraph seems to suggest that if you can get prompt criticality it could cause a 'nuclear explosion'. Plunk502 (talk) 11:28, 29 January 2013 (UTC)Reply

I have now removed the paragraph Plunk502 (talk) 11:01, 10 February 2013 (UTC)Reply

Erm

edit

"an assembly is prompt critical if for each nuclear fission event, one or more of the immediate or prompt neutrons released causes an additional fission event. This causes a rapid, exponential increase in the number of fission events."

I think this isn't strictly true, because if, for each nuclear fission event, exactly one of the prompt neutrons released causes an additional fission event, then the rate of fission events simply stays constant, doesn't it?

So it should read "if for each nuclear fission event, more than one of the immediate or prompt neutrons released causes an additional fission event" --Cancun771 (talk) 08:47, 15 April 2013 (UTC)Reply

Yes, but you are riding the knife edge to end all knife edges. If the drunkard's walk of the amount of reactivity going on starts to increase, you'll have a run-away situation going to saturation in less then a microsecond.SkoreKeep (talk) 01:36, 27 August 2015 (UTC)Reply
It does seem that critical is often enough, maybe even the WP:COMMONAME, for supercritical. Depending on exactly how big k is, the exponential growth can be pretty slow, or very fast. It is commonly believe to be 2 for bombs, I suspect partly because the calculation is easier. In a moderated reactor, the mean free time is pretty long, so growth can be pretty slow. Gah4 (talk) 03:54, 24 September 2019 (UTC)Reply

The wrong k. While it is usual to use k for the growth constant in exponential growth,  , the k use here is instead:  . In the former, it is supercritical greater than zero, and subcritical less than zero. There should probably be a reference for this use, though. Gah4 (talk) 18:17, 3 October 2016 (UTC)Reply

all in moderation

edit

There doesn't seem to be much discussion for moderators, and their contribution to fission rate, and growth rate. For a bomb, you need unmoderated prompt criticality. With enough moderator, and especially with a temperature sensitive moderator, you should be able to achieve steady state prompt criticality. Gah4 (talk) 18:22, 3 October 2016 (UTC)Reply

I was wondering the same thing -- it seems there are three different tiers of supercriticality: delayed-supercritical, moderated/thermal-supercritical, and unmoderated/fast-supercritical. Each of these has a typical time scale, ranging from seconds-minutes (reactor startup - delayed supercritical) down to milliseconds (reactor accident - thermal supercritical) and then down to nanoseconds (nuclear bomb - fast supercritical). The latter two both get classified as "prompt criticality" since they're too fast for a mechanical system to react to, but besides this the behaviour should be quite different. --Nanite (talk) 17:43, 23 September 2019 (UTC)Reply
There are two relevant factors that determine the characteristics of criticality: (1) slow vs. fast neutrons (related to the presence of moderator materials) and (2) prompt vs. delayed neutrons. Prompt neutrons are those released immediately in the fission event. Delayed neutrons are those that result from the subsequent decay of fission products. A small fraction of neutrons are delayed by times on the order of seconds. Bombs rely on prompt fast neutrons. Reactors (whether fast or thermal) rely on delayed neutrons for their stability/controllability. NPguy (talk) 15:46, 28 September 2019 (UTC)Reply

reactivity

edit

A recent edit exchanged k-effective and reactivity, which I agree with. There are many other places where k-effective is used that, for the same reason, could be changed to reactivity. I believe that they should be changed. Gah4 (talk) 18:11, 24 August 2018 (UTC)Reply

CP-1

edit

Reading Chicago_Pile-1 again, it isn't obvious that it ever went (delayed or not) critical. Consider that they had a neutron source, maybe a pretty strong one. They could then measure the overall k for the whole assembly, even when it was much less than one. Consider the case where  . One neutron from the source will, on average produce   more. Those, then, will then produce  , and from those produce  . Summing the geometric series then shows that for each source neutron, you get   neutrons. With a good source, they might get the 1/2 watt with k of 0.9999 or so. It seems that Fermi did understand delayed neutrons enough to know that they could build it where they did, but still had various manual and automatic safety systems. Gah4 (talk) 03:47, 24 September 2019 (UTC)Reply

poison

edit

Should the criticality section mention neutron poison fission fragments? As written, it seems to be true at the instant a reactor starts up, but after it is running, there are fission fragments that can also absorb neutrons, or decay products of those. Most famous is xenon poisoning, that makes restarting reactors soon after shutdown difficult or impossible. At delayed critical, enough neutrons have to be produced to balance along with those absorbed by such neutron absorbing reaction products, in addition to the non-fission absorption by uranium. Gah4 (talk) 01:32, 10 February 2020 (UTC)Reply

unsafe test

edit

I think I agree with the recent edit, removing unsafe describing the Chernobyl test. It wasn't the test that was unsafe, but powering up a reactor with xenon poisoning. The reactor had to be running at an appropriate energy to do the test. Getting to that energy caused the accident, not the test itself. As well as I know it, reactor operators are supposed to understand xenon poisoning, and its effect on starting up reactors. Gah4 (talk) 09:01, 9 October 2020 (UTC)Reply


Slotin Accident

edit

The deadly Slotin accident is missing, 21 May 1946, wheras the comparable Daghlian accident is the first in this list. https://en.wikipedia.org/wiki/Louis_Slotin#Criticality_accident — Preceding unsigned comment added by 2003:C3:2F03:B800:F425:CE2E:274C:44CC (talk) 20:21, 22 November 2020 (UTC)Reply

It's in the report. so I'll put it in.--Hieronymus Illinensis (talk) 06:18, 17 September 2021 (UTC)Reply

average

edit

The article notes that it is, on the average that each fission causes another one, presumably meaning time average. But there is also a spatial distribution. Reactors are designed, and refueled, to keep the power distribution somewhat equal throughout. An assembly might be way above critical on one side, and way below on the other side, so critical on average, but that isn't usually desired. The Oak Ridge drum described is likely such a case, where some fairly small part went prompt critical. Since the mean-free-path is fairly large, the size of the critical region is important. Gah4 (talk) 23:11, 22 November 2020 (UTC)Reply

stability

edit

There is a recent edit with the edit summary saying: made it sound like nuclear reactors could be controlled with just prompt neutrons, that is untrue. I believe that isn't true. That is, it is possible to keep the reaction stable without delayed neutrons, though it is harder. Control rods are a big part of stability, but they move at mechanical speeds, too slow to keep things stable. The main effect is thermal, and based on slow neutrons. As the fuel heats up, there atoms move faster, and the neutrons, too. The cross section is very sensitive to neutron energy (speed), and so decreases with temperature. Different reactor designs are more, or less, sensitive to the thermal affect. It seems that it is possible to build one where you can pull the control rods all the way out, and it won't melt down, just warm up a bit. On the other hand, the RBMK is not so stable, and gets worse at low enrichment. (Including after years of operation.) One of the fixes after Chernobyl is to increase the enrichment level for more stability. Well, also, thermal neutrons are slow, so even worst case power can't grow as fast as bombs. Gah4 (talk)

But another thing. A reactor is not homogeneous. Specifically, they are much larger than the mean free path. One part can be supercritical, and another part subcritical, for exact critical on the average. One side will be hotter, and the other cooler. So they move fuel rods around to equalize the reactivity. Gah4 (talk) 20:38, 17 June 2022 (UTC)Reply