Hacker News Reader

Comment by regularfry | Hacker News Reader

So here's a question that I'm sure a good answer exists to: given that an untested safety mechanism is guaranteed not to work when you need it most, how are SCRAM systems tested?

View on HN

There are various ways you can test it. But it obviously depends on what the mechanism is. If you’re cutting physical wires used to raise or lower something then hopefully it is obvious the mechanism will work (but maybe it goes wrong if the cut end of the wire is caught somewhere.)

In a computer system, you ought to be able to test systems to make sure they behave correctly under the consequences of a switch press. You could also test them—there may be times when it is ok to press the switch, or where you have redundancy such that two switches must be pressed but you only press one to test.

But I think another alternative is to not have the mechanism used extremely rarely. I think it can be good to accept cases where things go wrong a bit from time to time so that people have practice dealing with them as the alternative is to only have things go wrong under some enormous catastrophe when no one will have any idea what to do.

For nuclear power your third paragraph isn’t anywhere in the same universe.

SCRAM drills are held regularly, but under no circumstances is any intentional level of risk in critical safety systems accepted on the plants I worked on.

I was present in a data center during a scheduled complete shut-down so that some significant power and equipment moves could be done. At the designated time, the manager ceremoniously pressed the Big Red Button by the door and.... nothing happened.

So an electrician was sent to throw some switches elsewhere in the building.

I never got a clear explanation of what went wrong, only that it had been "fixed" and would work if needed in the future.

I think that's probably why they went with cutting the rope with an axe instead of some mechanical mechanism like moving the pawl from the ratchet that lifts the rods.

Then it's as mechanically as simple as possible

Incident report: shortly before the meltdown, we believe the safety supervisor was knocked unconscious by an improperly secured axe head that became detached from the handle swung by the junior safety officer. We recommend increased testing of axe head attachment mechanisms in future nuclear safety devices.

Typically they are tested at very low/no power levels when the reactor is first started after any kind of significant overhaul. We are talking staying subcritical, retracting your control elements a tiny bit in banks, and dropping them until every section is tested.

Additionally, the system and crew are tested on a regular basis with live drills. Sometimes surprise live drills. Its one of the most commonly-tested systems on the ship. Not to the point of actually causing a reactivity excursion, but the physical act of rapid rod insertion using the scram control switches (or breakers) happens all the time.

Please note my precise use of language; as stupid as it sounds if you were involved in the program (your language is precise enough that I’d lay a bet you were in some fashion) you signed documents indicating your inability to utilize certain words.

It’s stupid and I’m calling attention to it by this, and Wikipedia uses those words, but we sign NDAs and it’s the Government. No one ever claimed it all makes sense.

I’m curious why the downvotes but no comments?

I’m in no way judging the other commenter. I think that particular restriction is stupid since the information is in the public domain. However it IS still a binding legal restriction. My comment was as much about the stupidity as about any possible breach.

> I’m curious why the downvotes

'cause you came across as an asshole. It doesn't bother me that much. Nuclear safety relies on ~~assholes~~ attention to detail to work, to be honest.

> but no comments

Its not exactly a jumping off point for discussion, is it? I've been out for 15 years and still regularly fail to recognize when I'm being a dick. Look back at what you wrote and think about for a minute.

Fair enough; I work Security now so I don’t get confronted with my dickishness often enough I suppose, since it’s an asset there.

I appreciate the call out. We can always learn and grow!

I don’t know why the downvotes but I’m scratching my head trying to figure out what words in the OP would be forbidden. But I guess you wouldn’t be able to answer that question.

Suffice to say the reason you can’t identify it is why it’s stupid. Everyone knows them and yet the Navy requires we not reference them because that gives away design and operational characteristics of our plants.

Reply on news.ycombinator.com