[HN Gopher] A simplified analysis of the Chernobyl accident (2021)
       ___________________________________________________________________
        
       A simplified analysis of the Chernobyl accident (2021)
        
       Author : _vk_
       Score  : 15 points
       Date   : 2025-01-22 11:13 UTC (2 days ago)
        
 (HTM) web link (www.epj-n.org)
 (TXT) w3m dump (www.epj-n.org)
        
       | colechristensen wrote:
       | The technical reasons why Chernobyl exploded when AZ-5 was
       | pressed were only half of the reasons for the disaster.
       | 
       | The other half of the reasons Chernobyl failed were human
       | reasons.
       | 
       | It is very important when analyzing large systematic failures to
       | not leave these things out.
        
         | SoftTalker wrote:
         | Yes, the RBMK was not an inherently safe design (in fact was
         | deeply flawed) but it could be operated safely if you didn't
         | deviate from procedures. Which they did, because they were
         | never told about the design flaws (it was a state secret).
        
           | timewizard wrote:
           | The local engineers were effectively scapegoated by the KGB.
           | It was disappointing to see HBO follow the same lead.
        
             | hencq wrote:
             | Huh, did you stop watching after episode 1? The HBO series
             | was all about how it was the system that caused the
             | accident to happen. The fact that it was a known failure
             | mode was a major plot point.
             | 
             | It can hardly get more explicit than this scene:
             | https://www.youtube.com/watch?v=jBwSuSuGhyk
        
           | kqr wrote:
           | Procedures are written under the assumption that the actual
           | system behaves like the theoretical system the engineer has
           | in their head. It never quite does. There's always a gap, and
           | this gap nearly always requires deviating from procedures to
           | ensure safe operation.
           | 
           | Deviating from procedures prevents as many accidents as it
           | causes. Safety cannot be based on adherence to procedure.
           | Safe systems must be designed to take advantage of (and be
           | protected against, I suppose) human ingenuity.
        
             | colechristensen wrote:
             | This is nonsense, especially this line.
             | 
             | >Deviating from procedures prevents as many accidents as it
             | causes.
             | 
             | And they weren't doing some small deviation from procedure.
             | They were doing something that was expressly forbidden and
             | they knew why it was forbidden. In the time leading up to
             | the accident it would be difficult to distinguish between
             | what they were doing and trying to intentionally cause a
             | meltdown. In a reactor experiencing xenon poisoning instead
             | of shutting down for 24 hours (procedure) they removed
             | every nuclear reaction moderating mechanism they could.
             | 
             | This isn't a smart little deviation, it's pouring gasoline
             | on a fire and hoping for a good outcome. It is hard to
             | describe how stupid this was.
        
               | Retric wrote:
               | They are referring to a wider range of situations than
               | this very specific incident. There are many accidents
               | that have been avoided by modifying procedures on the
               | fly. Pilots discovering letting go of the yoke/stick
               | solved some very specific issues is one example.
               | 
               | It generally requires operators to know how the systems
               | work in great detail which would have avoided the
               | specific Chernobyl incident.
        
         | dylan604 wrote:
         | To me, Chernobyl is an example of the classic conundrum that
         | engineers cannot possibly think of every single weird thing
         | that could possibly happen so that a perfectly safe anything
         | can be made. It applies to software design just as much as a
         | nuclear reactor. Sometimes it takes a failure to actually
         | happen before something can be made safer. Somethings are just
         | more consequential when they fail making the learning from
         | failure much more expensive.
        
           | colechristensen wrote:
           | Not really. Corners were cut, the ultimate issue which caused
           | the explosion was known beforehand, and the operators
           | violated several points of standard procedure as well as
           | doing several basic unwise things. It was not at all a case
           | of unknown edge cases but stupid piled on top of stupid until
           | the damn thing exploded.
           | 
           | The biggest, stupidest action was trying to operate a reactor
           | very clearly experiencing xenon poisoning including the many
           | unsafe things they did to try to overcome the poisoning. I'm
           | pretty sure modern reactors still shut down for 24 hours to
           | avoid the xenon issue. This was well known, even without the
           | design flaws this was a huge risk, and anyone with an ounce
           | of sense would have known not to do what they did leading up
           | to attempting to scram the reactor.
        
         | orbital-decay wrote:
         | There were two high-level causes, basically:
         | 
         | 1. The failure to scale the education quickly enough. Nation-
         | scale nuclear energy was new when the RBMK line was introduced.
         | The demand for nuclear engineers skyrocketed, and it was
         | impossible to train the required amount of professionals to the
         | same standards as nuclear scientists in just a few years.
         | Meanwhile, the RBMK assumed deeper knowledge of its design than
         | they had.
         | 
         | 2. The system that made academicians (the official Academy of
         | Sciences title) equivalent to mid-to-large caliber politicians
         | within their area of expertise. As a result, Dollezhal's pride
         | ran unchecked and prevented him from addressing well-known
         | design flaws (that already caused the 1975 accident).
         | 
         | Both reasons are not unique to USSR at all and can be learned
         | from. (something that is often ignored because "it can't happen
         | here")
        
       | Terr_ wrote:
       | > When there was xenon poisoning in the upper half of the core,
       | the safety rods were designed in such a way that, at least
       | initially, they were increasing (and not decreasing) the core
       | reactivity.
       | 
       | I wonder if any reactor-design groups do "fuzz testing" on
       | simulated models, checking that they can recover from very weird
       | states even if it's not clear how the state could have been
       | reached in the first place.
       | 
       | For example, having one section just arbitrarily xenon-poisoned,
       | while another is arbitrarily too hot.
        
         | _n_b_ wrote:
         | Yes, essentially this happens. PWRs and BWRs have operating
         | limits on their power shape derived from doing those kinds of
         | analyses.
         | 
         | They're tend do be more physical than "arbitrarily xenon-
         | poisoned" but represent a variety of extreme and nominal states
         | to form an operating envelope, and then healthy margins are
         | applied on top of that.
        
         | xk3 wrote:
         | The disaster was a result of fuzzy testing :-)
         | 
         | https://en.wikipedia.org/wiki/Chernobyl_disaster#Safety_test
        
         | cyberax wrote:
         | Yes. Reactors now are designed to never have positive feedback
         | loops that can result in uncontrollable power spikes. They do
         | the worst case simulations to prove that.
         | 
         | Russian atomic regulator downright shut down the project to
         | design a light water breeder reactor, saying that it's not
         | going to be licensed. Light water breeder reactors are barely
         | theoretically possible, but they require trade-offs to limit
         | the amount of water within the reactor core. So the trade-offs
         | result in a positive void coefficient. It's supposed to be
         | offset by other safety features, but better be safe than sorry.
        
         | timewizard wrote:
         | Yes, it wasn't intentionally, but it was at Ingalina.
         | 
         | https://en.wikipedia.org/wiki/Ignalina_Nuclear_Power_Plant
        
       ___________________________________________________________________
       (page generated 2025-01-24 23:01 UTC)