Comment by EEGuy | Hacker News Reader

Comment by EEGuy | Hacker News Reader

Until recent news such as this, I didn't know it was the "conceptually simple" but practically difficult ability to switch (interrupt) a GW or so of DC energy flow preventing the implementation of better than point-to-point HVDC (i.e. multiple access point) power grids.

Evidently interruption's easier with AC because of the zero-crossings of current every half-cycle.

Three alternate solutions to HVDC circuit interruption are described in this article. Two are electronically integrated with the output circuitry of the bidirectional AC-DC converters (bidirectional energy flow AC-DC converters being an interesting, but off-topic subject in their own right).

I find it an ingenious use of an IGBT bridge surrounding a capacitor to throw the cap's charge into reversed polarity so as to stop the current flow -- effectively opening the circuit. This tells me the cap would seem to be in series with the DC side somehow. Were it in parallel, well, it wouldn't make much sense to reverse the cap's polarity as you'd only get to do it once.

I also find it fascinating that the mathematics it takes to describe AC circuits is far more complex than for DC -- one has to accept that there genuinely exists a square root of negative one, "imaginary" though we may call it.

As in programming, good nomenclature is often late to the table. I wouldn't try (and I'm certainly not qualified) to "refactor" centuries, maybe millennia [1] of mathematical terminology.

-----

[1] https://en.wikipedia.org/wiki/Imaginary_number

Very interesting article indeed!

>I also find it fascinating that the mathematics it takes to describe AC circuits is far more complex than for DC

Why? The math is exactly the same as for DC once you substitute complex numbers for current, voltage and impedances.

Things get more complicated in the time domain, which becomes relevant once you want to look into problems like switching HVDC on and OFF. The nonlinearities of the switching devices lead to differential equations that can only be solved numerically.

Thank you kken for your clarifications.

I don't know the physics modelling switching devices or arcs, but I do recall reading long ago that once an arc starts, its resistance drops dramatically, continuing to drop as more current flows through it (within limits).

So an arc becomes a runaway situation if current flow isn't ballasted or elsewhere interrupted. But if current flow is controlled, an arc has practical application in fluorescent lighting and arc welding.

A visual demonstration: http://www.youtube.com/watch?v=9S64r-MjBN4

You don't have to use imaginary numbers to model AC circuits. The math is just easier if you do.

Pie in the sky: if the ability to efficiently and rapidly switch DC is acquired, will we see the rise of packet switched energy networks? Data networks made the transition from circuit based to packet based networks, so why not energy networks?

If centralised generation is replaced with distributed generation there will be the need to route energy between arbitrary sources and sinks, located at any point in the network. One can envisage a network analogous to a packet based data network, where "routing algorithms" are used to direct fixed amounts of energy (a packet) from "A" to "B". A process analogous to flow control is used to govern the number of packets, and so the average power, being transferred between a source and sink. A device analogous to a buffer (capacitor?) could be used to average the power at the entrance and exit to the packet network when smooth power is required, just like running a "smooth" audio stream over a packet data network.

For DC chains, you might find yourself in need of Operational Analysis, especially when you plan to switch the flow. That beats the "simple" imaginary math just right.

http://en.wikipedia.org/wiki/Operational_calculus

Reply on news.ycombinator.com