Data centres consume a significant amount of energy. In the 1990’s and the first decade of this century, the Government, influenced by public opinion and the highly laudable aims of the Green movement, put off politically unpalatable decisions about what would happen to energy supplies later this decade. Mostly the politicians hoped that the decisions surrounding building new nuclear power stations would go away, either because new green generation technology would come along and prove itself or that it would become someone else’s problem after they had left office. Such technologies take many, many years to work their way into mainstream usage, and it takes twenty years to build a nuclear power station, assuming that it has planning permission to start with.
The result is that the UK faces an energy shortfall in the later part of this decade, as the older nuclear stations are retired because they are worn out and no longer safe, and many of the old coal-fired stations simultaneously go out of service because they can no longer meet modern pollution regulations and no-one will invest in a declining technology. Gas will only fill part of the gap, making the UK dependant on local politics in Siberia, let alone all the places in between. Although people talk lightly about it today, there really will be brown-outs (where the supply voltage drops) and switch-offs (hmm, we don’t seem to have enough electricity – shall we switch off Southampton or Leeds today?). So this begs the question, what can data centres do to protect themselves?
There are several ideas for data centres to create their own electricity, but not all are practical. The easiest is to use diesel generators, and almost any decent existing data centre already has at least two (being a Tier 3 data centre requires at least N+1 generators). This works well in terms of ensuring electricity supply, but the cost of electricity generated is around double that of buying it, and it depends on diesel fuel being available at a time when everyone else also wants it. Diesels cause pollution and noise, and may not be acceptable for long-term running in urban areas. Perhaps a more sensible option is one where the distribution network authority takes control of local diesel generators and switches them on to offload the network at times of high demand. That will allow “peak-lopping” and reduces the big power surges when everyone goes to make a cup of tea during the advert break in the middle of Coronation Street.
Another option is solar power, but current solar technology would generate only a small fraction of the power needed by a data centre in Britain. Plus obtaining planning permission for large solar arrays may prove to be very difficult, especially in urban areas. Wind suffers from the same problem as solar – it only generates a fraction of the power needed by a data centre in Britain, there’s no guaranteed supply of wind, it’s hard to get planning permission and is not usable in a City.
Fuel Cells are another possibility and one US Bank built a fuel-cell based data centre as far back as 1999. But they also have disadvantages – high cost and the availability of a guaranteed supply of hydrogen. And, overall, they aren’t very green. Operating them is extremely green, because they produce only water as a waste product, but manufacturing hydrogen is not at all green or energy-efficient. Most hydrogen today is made from natural gas or crude oil and a lot of heat, which rather negates the greenness of its use. The greenness comes from operating in a City, but the disadvantage is a potential giant bomb in the City – all a terrorist needs is a little bomb as a detonator.
Burning biomass, such as waste straw or discarded organics from food processing is another possibility, but limited by the guaranteed availability of fuels – and, of course, you can’t burn vast quantities of rancid old fats in the middle of a City. Something like a Doxford engine (with an exhaust scrubber), designed for low-cost cargo ships and burning almost anything that contains carbon and hydrogen molecules and can be liquefied, might work very well in a non-urban area, generate a substantial amount of very low-cost energy and gobble up a lot of waste. Perhaps someone would like to develop that for a genuinely green data centre?
Local micro-generation using natural gas is very practical. It is also cost-effective, as the released calorific energy in gas costs about a third of that from delivered electricity. A data centre burning gas, and with no connections to the national electricity grid, is entirely practical, but probably still needs to be backed up by diesel generators for security of supply. For large data centres, aircraft gas turbines are very practical. An engine from a Boeing 737 will generate 20MW or so continuously, and gas-burning applications of aircraft engines are well-established in areas such as pipeline pumping. Plus they are small and very reliable, although not cheap.
So what is a data centre to do? A very few may move to gas micro-generation, backed up by diesels. But almost everyone will rely on their existing diesel generators, buy big fuel tanks, accept the costs, and put in place preferential supply arrangements to ensure that when their area gets switched off for the day they get their fuel first. There’s certainly a cost to the politicians’ wilful indecision ten years ago and data centres will have to bear a part of it.
Tags: Power & Cooling, Green Tech