@ieee_pes AI to write a tweet? Are you serious? You are a highly respected Electrical Engineering forum, not a mom and pop shop. YOU ARE THE GOLD STANDARD. Put a keyboard in a humans hands.
Prototype GDA took me a long weekend to write, i have spent 2 weeks working on "Production" and i have come to the inescapable conclusion, building in a Manic Period its how GDA was built. #MentalHealth#TalkToSomeone
On 28 August, GB power ran from 40 g/kWh at 10:30 to 174 g/kWh at 23:05 — a 4.3x swing the single 'grid average' hides. When you use power matters as much as how much. #UKEnergy
On 28 August, GB grid frequency swung from 49.803 to 50.187 Hz, and spent ~6 hours outside the tight ±0.1 Hz band. The 50 Hz 'constant' is anything but. #UKEnergy
On 28 August, gas met 18% of GB demand and never once switched off — running every half-hour, never dropping below 1,714 MW. The 'renewable' grid leans on it around the clock. #UKEnergy
On 28 August, GB imported 75 GWh over the interconnectors — 13% of everything the grid used — power, and its carbon, we count as someone else's. #UKEnergy
Grid-clock check for 28 August: the grid averaged 50.004 Hz against the 50 Hz standard — running slightly fast. On the mains-clock measure, @NESO_ENERGY ended the day 46 MWh in credit. #UKEnergy
@gridmatters@StephenNCC1701@dorfman_p Further, synthetic inertia is not real.
It was not present at T=0, it is not part of the response in the first 200ms, and in a datacentre, i have already collapsed the inverters output.
@gridmatters@StephenNCC1701@dorfman_p As a multi MW data centre operator i can assure you sir because we have tried, more than once, carrying a datacentre on pure inverter is not viable yet, without inertia, when a load ramps from 0 to 120KW per rack, hundreds of them, inverters collapse. We ramp 0 to 50MW < 100ms
@therohanparmar Indeed, and its of course caveated under the real one runs in production, with real data, and very real imperfect users.
But to get this close from a spec sheet, I am becoming impressed. Not something I thought I would be saying.
I handed Claude a 150 page spec sheet for an app. The code works, its clean, its not bad at all. It looses out to the real thing, the gap has closed, under opus it was 18%, under fable it's 4.2%
For the first time I'm going to say this about AI.
Parity is near...
On 28 August, @NESO_ENERGY made 10,824 balancing interventions to hold the grid at 50 Hz — on average one every 8 seconds, all day long. 5,166 to ramp generators up, 3,226 to ramp them down. This is what 'stability' takes. #UKEnergy
On 27 August, GB power ran from 80 g/kWh at 03:15 to 218 g/kWh at 21:20 — a 2.7x swing the single 'grid average' hides. When you use power matters as much as how much. #UKEnergy
On 27 August, GB grid frequency swung from 49.809 to 50.158 Hz, and spent ~5 hours outside the tight ±0.1 Hz band. The 50 Hz 'constant' is anything but. #UKEnergy
On 27 August, gas met 31% of GB demand and never once switched off — running every half-hour, never dropping below 3,239 MW. The 'renewable' grid leans on it around the clock. #UKEnergy
On 27 August, GB imported 65 GWh over the interconnectors — 10% of everything the grid used — power, and its carbon, we count as someone else's. #UKEnergy
@StephenNCC1701@dorfman_p Cant run a datacentre on pure inverter and grid forming tech. The load swings are far to brutal to survive without true inertia.