[AusNOG] Lessons Learned From SA Blackout

[Joshua D'Alton]

https://www.youtube.com/watch?v=slDAvewWfrA  is an interesting watch along
these lines.. its a massive sudden increase in load, vs a sudden drop in
supply, but core failure scenario is the same.

Comparing this SA failure to some large cloud outages in the last decade
due to cascading load shifts.. very similar to electricity in the
fundamentals as you suggest. Highly interesting regardless!

On Fri, Sep 30, 2016 at 7:22 PM, Ben Buxton <bb.ausnog at bb.cactii.net> wrote:

>
>
> On Fri., 30 Sep. 2016, 18:44 Mark Newton, <newton at atdot.dotat.org> wrote:
>
>>
>> It shouldn’t have happened like that: The network was working before the
>> transmission lines fell over, and it’s working now while they’re still on
>> the ground, unrepaired. The question is why it couldn’t work during the
>> state transition, and I expect that’s what most of the upcoming
>> investigation and resulting post-mortem will be concentrating on.
>>
>
> When a whole bunch of power suddenly stops being fed into the network by
> say a toppling tower, the rest of the network has to supply it. Some
> inter-network feeds trip because they aren't provisioned to provide the
> sudden relative demand. Remaining generators that could normally scale to
> handle the load were running at lower power but suddenly found themselves
> with a much bigger relative demand (ie within a second).
>
> Just like a car engine, they immediately start to slow down from the
> increased load. The slowdown caused a frequency shift which turns into a
> phase shift which is Really Bad News. Protection circuits trip.
>
> No matter how fast operators react, it always takes some time to turn up
> the thermal source. That time will be greater than the protection trip
> time. Thus, state goes on the blink.
>
> The reason power is restored now despite the towers still being down is
> that they would have made sure the available generators were cranked to
> supply the full demand before turning the switch back on.
>
> (The above is my armchair EE hypothesis but based on casual studying of
> power systems in general. Plus much of the theory is fundamentally the same
> as IP networking)
>
>
> Bb
>
>
>
>
> _______________________________________________
> AusNOG mailing list
> AusNOG at lists.ausnog.net
> http://lists.ausnog.net/mailman/listinfo/ausnog
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ausnog.net/pipermail/ausnog/attachments/20160930/c2d50409/attachment-0001.html>