How do national grid forecast demand acturately

July 31, 2008 2:33 AM

How do national grid forecast demand acturately?

How do national grid forecast demand acturately? I know they look at the weather and history/patterns. But does anyone know more precisely how they do it, from day to day and year to year?

posted by henry.oswald to science & nature (7 answers total)

Do you mean the energy grid?
posted by 26.2 at 3:54 AM on July 31, 2008 [1 favorite]

I don't think it works quite like that, certainly not in the UK anyway. It's up to the supply companies (i.e. companies that buy electricity from generators and sell it to consumers) to agree contracts for sufficient power to meet the demands of their customers at any given time. This is carried out through bilateral contacts between generators and suppliers; if either fails to deliver - either supply or demand - when they have agreed to do so they will then be the one which bears any additonal system costs. The main costs will be if the supply company underestimates the demand of their consumers in any given period or if the generators can't deliver as agreed. National Grid has the responsbility for 'balancing' the system, and for ensuring supply = demand, to meet the unexpected demand (either additional or unmet) National Grid will go to a secondary electricity market (the balancing market), which consists of generators catering specifically to that market, they will generate less often than generators catering to the open market but will get paid more per unit. There is a 'gate', which from memory is one hour in the UK, this is the minimum notice that the generator and supplier can give National Grid as to what their arrangements are, if they can't sort out arrangements before this period, or their is some problem after the gate has closed then National Grid sorts it out via the balancing market.

As I say, the additonal costs of meeting this demand are passed on (i)
to the supply company who underestimated the demand of their customers, effectively the supply company bears the risk of estimating demand and of having in place sufficient contracts to meet that demand. The idea is that whichever supply company is best at managing that risk will have a competitive advantage in the marketplace.

(ii) to a generator who fails to deliver (for example, if the plant has to shut down unexpectedly), then it is they who must pay the additional costs if National Grid has to go to the balancing market. Again, the economic argument is that the generator with the least costs should be able to capture more of the market.

The regulatory regime relevant to balancing is important in relation to the intermittent renewable energy sources, e.g. wind, wave, tidal, PV. Effectively if generators can't deliver when they have contracted to do so then they will bear costs. Because output will be intermittent in comparison with sources of 'firm' generation then National Grid is more likely to have to go to the balancing market and renewable energy generators are more likely to bear additional costs. This is a key reason why the RE sectors (especially wind) have been trying to improve predictability of weather conditions in order to improve predictability of output. The length of the gate can also have a signficiant impact on the costs associated with intermittent generation, since a shorter gate requires less prediction ahead. There may also be problems with the additional balancing costs for renewables not actually reflecting the actual costs, with specific impacts on comparative costs for renewable energy generators.

There is a third market for electricity generators, ancillary services, which isn't really relevant to the balancing market, but exists to provide additional services to the grid, such as 'black start', voltage control, etc.
posted by biffa at 4:19 AM on July 31, 2008

I should point out that the bilateral trading system is very UK specific but the responsiblity for delivery in most competitive privatised electricity supply industries will have the suppliers and/or generators bear the risks and costs rather than the transmission network operator.
posted by biffa at 4:23 AM on July 31, 2008

26.2 National Grid is the name of the company who owns and operates all the electricity lines in the UK. And the gas mains IIRC...

biffa looks like he knows what he's talking about so I'll only add thoughts not definites, but I believe that we use hydroelectic plants as capacitors to provide power in a hurry if needed. In theory this will damped (aha!) the problem of brownouts, although in my part of the UK (Cambridgeshire) brownouts are incredibly common.
posted by twine42 at 5:19 AM on July 31, 2008

This link to a BBC story has a good explanation of what National Grid do in terms of forecasting. I found it while checking whether twine was right about hydro being the balancing generator of choice, the BBC story seems to say its a contributor but fossil fuel is more important, which will effectively be gas I suspect, which can follow load a bit better than coal (we don't really use oil for electrical generation any more in the UK). You have to remember we don't have a huge amount of hydro in the UK, and some of it is base load to maximise profits.
posted by biffa at 6:00 AM on July 31, 2008

You need a very complex model essentially of the whole grid in detail and the demands on it (you can't just treat the total output of the power stations as one number - you need to know where the power goes in and where it goes out and how it gets there to figure out losses etc. properly).

Basically there's teams of people writing the computer models and fine-tuning it based on how things actually play out. I'm not sure how much more detail you should put into an AskMe answer about this stuff.
posted by edd at 6:12 AM on July 31, 2008

Complex models are the answer here. But even then it's not simply a matter of knowing which generators are where and who is going to be consuming what when. Although there are methods to forecast loads (see ARIMA, for one of the more basic ones), utility companies have to have much longer term projections with respect to loads so that they can come up with long term (~20-25 year) plans for their generation, transmission, and distribution systems. Often these are separate utilities, but it depends on where you're located.

These methods are based less on time-series and weather patterns, and more on things like economic factors. One-industry down in decline? Probably won't need new transmission lines. Booming city with no sign of slowing down? Likely going to have to put something in there soon.

But yeah, complicated growth models are the answer to this.
posted by sah at 2:08 PM on July 31, 2008