Quote:
Originally Posted by paperplane
If they require 380MW to run, are going to generate 190MW themselves with their NG power plant, and get 190MW from the grid, I don't understand how they at times might be able to put energy back into the grid from what they generate.
As well as what was previously said, how can they throttle down their power needs if NBP requires them to based on grid demand? Would they really want to sign an agreement to not be able to operate fully, especially when they're beholden to whoever their large corporation user is?
There seems to be a lot of questions and wishy-washy details so far.
|
It is important not to see the power consumption as linear or constant, like a manufacturing facility. A simple example would be a data centre for business software applications, let's say accounting.
During the day, the servers would be using lots of power while everyone is working, processing transactions and running reports. However, after hours or overnight, usage will drop off, and so will the power requirements.
During that lull in activity, in theory, there might be excess power generation that isn't required, though I am pretty skeptical it's practical or economical to put it back into the grid.
If this is for AI, there are two options: either it's used to train models, which takes a long time and requires a lot of computing power, or it's used to respond to user queries, which is more of a real-time application. People aren't asking ChatGPT questions at a consistent rate throughout the day; there are busy and quiet times.