Energy savings and environmental benefits

We argue that alternative desktop devices offer significant energy savings and environmental benefits when compared to the approach which puts a PC on every desk. On this page we'll expand on that (but see the footnote for a warning).

The case for environmental benefits is relatively easy to make, but the case for energy savings is complicated by the fact that not everyone wastes energy in the same way. Let's start with the easy bit first.

There is a choice of alternative desktop technologies, with a useful distinction being made between recycled PCs and built-for-purpose devices.

Reycled PCs

Old PCs are an interesting case. Cutter can make use of recycled PCs as desktop devices after their hard drives and other redundant components have been removed. This is probably the most environmentally friendly approach, since it minimises landfill waste and no new component manufacture is involved. Recycled PCs tend not to be as power-efficient as special-purpose alternate devices but overall that's more than offset by not replacing them with newly manufactured equipment. They are also likely to be slightly noisier and obviously, having already served for several years, are liable to have unpredictable reliability. Where cost or environmental issues are paramount they do offer a compelling opportunity.

After three or four years of use as a PC, we've seen some of these gutted devices go on to deliver years of service in their new guise with minimal maintenance. In organisations that aren't obsessed with projecting a 'new and shiny' image it can be fun to paint or decorate the cases of the old PCs and celebrate their longevity instead of being embarrassed by their bulk and styling. Cannibalised power supplies and cooling fans will keep a fleet of them running for a surprisingly long time, though most users just scrap them if they fail.

The power consumption of stripped-down PCs is variable. Twenty to eighty watts, perhaps, depending on what's inside them.

Built-for-purpose Desktop Devices

Built-for-purpose devices come in a range of forms. Overall they differ somewhat in terms of performance, manageability and ease of use but for this discussion they can be grouped together. Typically they consume in the region of five to twenty watts of power, compared to the eighty or hundred and twenty of a generic desktop PC. Since they contain few or no moving parts and are designed for a long service life, they should last at least ten years which contrasts well with the the usual three or four years of a desktop PC. These are clearly substantially better figures both from a power and waste perspective. They are silent or near-silent in operation. Pricing varies and although they should cost less than a full desktop PC, lower production volumes means that they aren't yet as low in cost as you might expect them to be.

Energy Savings

Energy savings are difficult to predict because so much depends on patterns of usage. Alternative devices are much less likely to be left running overnight than office PCs (if the users are encouraged and trained to turn them off). Organisations that are strict about power savings can ensure that idle PCs are switched off or put into power savings mode, but in practice this varies greatly from user to user, department to department and organisation to organisation. It should also be noted that the power-saving mode or even the 'off' mode can use almost as much power as some alternative devices at full-power. The 'off' state is a surprise to many people but a modern PC even when apparently switched off is not in fact really off, it's just in its lowest-power state and still consumes a few watts. To really turn it off, you have to unplug it or use the off switch (if there is one) on the power supply at the back of the case.

From power-on to full availability should be a matter of a few seconds to at worst a minute for any desktop device in a Cutter system. Typical desktop PCs will often take substantially longer than this, in the extreme cases we have heard of ranging from several minutes to a quarter of an hour depending on the environment they are operating in. People being people, this can be a powerful disincentive to switch them off at night and for some it gets compounded by a lingering fear that they might not reboot properly anyhow. Slow bootup times are not great for productivity either.

It is therefore quite hard to predict the power usage of of desktop PCs since for this aspect of their usage there is so much variation. Probably the nearest to 'typical' you can find is that nobody turns anything off and just goes home leaving everything switched on.

We then have to consider the application servers. These are usually permanently on and consume around twice as much power as a desktop PC whilst providing application services to between thirty and sixty simultaneous end-users each. From that point of view they are clearly very power-efficient especially if unused desktop devices are routinely switched off and that is compared to a fleet of permanently-on PCs. The industry in general is now turning its attention to power management of applications servers and we can expect the no-load power consumption figures to improve with newer models. Ideally they would be shut down when not in use and brought back up to power when they are needed but that's not currently a practical option. The servers will consume a bit less power when idle than when fully loaded with users.

Further complicating the equation is air conditioning. A room full of servers is almost certainly going to need substantial air conditioning and this adds noticeably but unpredictably to the power budget, having to be evaluated on a case-by-case basis.

A room full of alternative devices is much less likely to need air conditioning than one full of PCs but may in fact need more heating during cold periods since so little heat is being produced. So that's hard to give accurate predictions for either.

Because there are so many uncertainties surrounding the patterns of use and the likely power consumption experienced, we are cautious about trying to predict particular power savings when switching from one type of system to another, other than to wave our hands and use rather meaningless phrases like 'substantial'. It IS possible to produce an energy saving prediction as part of an overall system design and we do have customers who are very happy about the outcome; we just don't feel it's very useful to provide swathes of unsubstantiated figures as a sales pitch.

Framework Figures

Having said that, here's the classic 'back of a cigarette packet' calculation for a medium scale 400 desktop deployment based on nothing being turned off ever. The screen or display devices are left out of this calculation as they are likely to be roughly the same in either case.

PC-based system:

  • 400 desktops at 100 watts = 40kW, or over a year nearly 350 megawatt-hours

Alternative system:

  • 400 desktops at (say) 10 watts = 4kW
  • 16 application servers at 250 watts = 4kW
  • Total is 8kW, leading to 70 megawatt-hours per year

In the UK if you are getting your power at 10p per unit (that's a kilowatt-hour) for the PC-based system that's £35,000 in electricity costs, for the alternative desktop system, £7000. We'll leave it to you to do some similar estimations based on your own needs. You can entertain yourself by working out what that is in terms of tonnes of carbon dioxide too.

Footnote

Warning:

Figures given on power consumption and life expectancy mentioned below are based on general experience and observation backed up at times with occasional measurement. This discussion is for guidance only, not a firm promise of performance of any particular installation.