SoylentNews is people
SoylentNews
The Australian Energy Market Operator (AEMO) says wind-powered electricity generation's "intermittency" had nothing to do with the blackouts following South Australia's catastrophic storms in late September.
In its ongoing investigation of what caused the "system black" event on 28 September, the AEMO says the storm caused a loss of 445 MW of wind output (out of a total loss of 900 MW), rather than the 315 MW of its original estimate.
However, AEMO's updated preliminary report (PDF) makes it clear that the political assertions made about wind energy were false.
"The most well known characteristic of wind power, variation of output with wind strength (often termed 'intermittency'), was not a material factor in the events of 28 September 2016", the report states on page 21.
Only 20 MW of wind power disconnected due to excessive wind speed. Instead, most loss of wind output was down to how the wind farms' control systems were configured – in particular, the settings for "voltage ride-through" events, which the report notes are proprietary settings in the control software.
For more information on the storm, see coverage by CNN, The Guardian , The Conversation and ABC (Australian Broadcasting Corporation).
"Bad Bespoke Software." Software shouldn't have spokes. That's the problem right there.
(Score: 3, Insightful) by martyb on Thursday October 20 2016, @09:13PM
Your story reminded me of one I'd heard. Back in the day (late 70's or early 80's), IBM's big mainframes ($millions) ran in BC (Basic Control) mode. For a $VERY_LARGE fee, a technician could visit your site and upgrade your machine to EC (Extended Control) mode. IIRC, this unlocked extra memory addressability and extra hardware features.
One curious sysop watched closely as one of his systems was upgraded. He saw the IBM tech show up, take the back off one of the cabinets, poke around for a few minutes, put the cover back on the box, and leave. How could he do that so quickly? So, he calls up a buddy at another shop, tells him to take notes of everything inside the cabinets, and find out what changed. Sure enough, it was just the moving of a... jumper cable. It controlled a bit in the PSW (Program Status Word) which toggled whether or not the machine was in BC or EC mode.
Word soon got out that in 5 minutes, you could upgrade your systems for free and save lots of $$$.
And then the inevitable happened. A system would crash, and you could not bring a tech onsite to debug it, unless you reset the jumper. And, of course, the bug would not reproduce. Little by slow, customers became dependent on the extra capabilities. So, they begrudgingly accepted they needed the support and shelled out the $$$ to get an official upgrade.
It seems to me that the moral is: Make upgrades better, but not perfect.
I'll leave it to the reader to determine if this is the path that the systemd developers have chosen.
Wit is intellect, dancing.
(Score: 2) by LoRdTAW on Thursday October 20 2016, @11:28PM
Before he quit, my predecessor swore that a computer controlled glove box (we no longer have) had a time bomb in it. Every year during autumn it would inexplicably just stop working. No physical problems. No bad sensors, loose wires, broken anything. The DOS (yes, that DOS) application would just freeze up and stop responding. Reboots didn't fix it, scan disk, different I/O card, nothing. So they had to call the only tech for that machine, a guy who worked for the manufacturer, to come in and fix it. He would run a "diagnostic" tool from a floppy, type in a few things, and magically the machine would come alive again. All for the low price of a $2000 service call. When asked if the diagnostic software could be purchased we were told it his was proprietary software he would not sell it. When also asked about the error he gave some lame BS answer about program memory corruption or some nonsense. They scraped that machine before I started working there thankfully.
Oddly enough, that machine which replaced it also runs DOS and still does. Written in QBasic and ISA I/O cards. That guy also wouldn't sell the source code after he told me we were the last people in the world he knows of using that machine (yea right). He instead would charge us $10k to build a new system using visual basic and PCI I/O cards. No thanks. He also sold us a replacement part he said he had lying around in his garage for $500 which didn't work. He then had the nerve to make us pay for the return shipping and refused to ship it over night out of pocket as were behind on work. Those niche service techs have a nice scam going. I have already eliminated another guy who charged a $1500/month retainer. He was in every 3-4 months fixing the same fucking problems because he never actually fixed them. Just band aided it until it failed again warranting his retainer. Scam artists.
(Score: 2) by butthurt on Friday October 21 2016, @02:38AM
a related comment:
/comments.pl?sid=15695&cid=406548 [soylentnews.org]