Right Sized (Chloride, UPS)

15 March 2010

Energy costs, the CRC Energy Efficiency scheme and the changing nature of electrical loads make UPS specification a challenge. Rob Tanzer proposes a series of technical and performance benchmarks for any big UPS project.

AS CAPITAL INVESTMENTS GO, selecting and costing a UPS project can be deceptively simple – even when we’re talking about big ‘three phase’ systems, rather than the commodity single server or desktop systems well known to users. Power, autonomy and availability requirements can appear to translate fairly easily into an apparently correct specification to be met by a contractor or supplier – but fast pace regulatory and technological change can have a huge impact on the business case for any given technology. This difference between a just adequate and a best-fit solution can equate to a big slice of your electricity bills – millions of pounds in some cases.
Among the challenges for FM is that from April 2010, the CRC Energy Efficiency Scheme imposes a carbon tax on big power users, with additional penalties for anyone who fails to improve. The scheme is likely to extend eventually to lesser users of power, with all businesses likely to pay more for their power in the near future.
UPS equipment is a long-term investment. The latest generation of flywheels, for example, are designed to last for twenty years, with switchgear and UPS equipment not far behind. At a basic level, therefore, it’s worth specifying the most efficient available technology. A direct comparison between equivalent units from two top manufacturers reveals an up-front efficiency difference that could easily amount to a £million saving over a ten year period, for a user with a 1000 kW UPS.
It is now possible for a new UPS, even in the most demanding applications for the most critical loads, to have a total efficiency not far short of 98 percent. To put this in perspective, there are still contractors around who opt for traditional technologies that simply waste up to 10 percent of critical power throughput – at a potential annual cost to the end user of over £90,000 for every 1000 kW of UPS capacity.
Over specification
With this long view in mind, FMs and UPS suppliers have often opted for significant overspecification to maximise the life of UPS by anticipating future loads. However, radical overspecification is now regarded as poor practice.
Over-specified UPS units, by definition, will run at partial capacity. At low loads UPS efficiency decreases dramatically, a major headache for managers facing periods of reduced load or uncertain future growth in demand. FMs will doubtless be mindful that – especially in new build facilities - one effect of the Energy Using Products Directive is that new electrical devices consume far less energy in ‘standby’ mode than previously, with corresponding impact on the difference between peak and trough demand.
One solution is to specify a modular UPS. Being composed of several smaller units, these systems should automatically switch off excess modules when power is not required, eliminating the problem. These systems have the added advantage of allowing users to buy only as much capacity as initially required, minimising up front capital costs. Nevertheless, where partial loads are anticipated UPS can and should be specified so as to deliver more than 95 percent efficiency down to loads as little as 20 percent.
A second cause of over-specification is that future loads are difficult to predict. For example, blade server power supplies running at light loads usually draw power at a significantly leading power factor which can dramatically increase the inefficiency and ongoing costs of running a conventional UPS. Such blade servers and nonlinear IT loads need to be supplied from a source that can match the current waveform and power factor at which they draw it, and it is important that the UPS can support this. A traditional UPS, faced with a typical leading power factor load, as drawn by blade servers, often has to be significantly derated, and in some cases will support as little as 60 percent of the load that it could otherwise support with a conventional load. Selecting a UPS that can support any leading or lagging power factor load will eliminate any need for such derating.
Rising energy prices and the additional costs imposed by the CRC Energy Efficiency scheme can and should lead organisations to ask - Is the cost of double conversion technology really necessary? Should the load be protected all of the time?
For many, especially those responsible for large data centres, the answer up to now has been an unequivocal ‘yes’ to both questions. Indeed, the standard for the most secure Class 1 UPS systems IEC standard EN 62040-3 – which is in turn a requirement adopted for tier III and IV data centres - dictates that: “The inverter always supplies the power to the load and it takes power from either the AC input via the rectifier or from the battery.
The rectifier has to be controlled so as to recharge and maintain the battery in a charged condition”. Whilst the traditional double conversion process – whereby the battery always supplies the load – is one way of fulfilling this requirement, it is not the only way. Between 5-10 percent of total power throughput is typically lost in a traditional double conversion system and so there is a compelling argument for more efficient alternatives to be considered.
One answer is to bypass the batteries and inverter altogether in what is usually termed an ‘ECO’, Maximum Energy Saving (VFD) or standby mode, whereby the critical system load will be supplied with whatever voltage the AC supply happens to be at the time. This is a common solution for low power IT applications.
However, the time required for a UPS in ‘ECO’ mode to detect a mains voltage deviation and then transfer to double conversion mode is short but measurable to a few milliseconds. Because of this time delay many data centre managers, view ‘ECO’ modes with considerable suspicion. The voltage variations permitted for a UPS supplying the most critical facilities, such as tier III and IV data centres (those complying with the UPS standard EN62040-3 Class 1), are much tighter – in fact, the maximum permitted deviation is +/-30% for a similar very short time period.
Another solution is to use double conversion only when absolutely necessary – when irregularities in power supply occur that are of such magnitude that only batteries can realistically compensate. In the UK, despite the fact that every single user can expect around 7 - 10 power disturbances every single day, the need for UPS systems to operate on batteries for any significant period is most infrequent. Manufacturers have realised that for the rest of the time it is possible to compensate for lesser irregularities – which would nevertheless bring a server room down – using the UPS inverter alone to condition the power delivered to the critical load.
If UPS is able to access this third mode of operation (VFD), the power of the inverter is used to modify the AC supply waveform using just the amount of power required to bring it within the parameters required by the system load. In this Power Conditioning mode the inverter operates in parallel with the bypass supply to regulate and condition the AC voltage waveform delivered to the system load. The inverter also monitors the current drawn by the system load and provides precise power factor correction and harmonic correction to ensure that power drawn from the incoming AC supply is sinusoidal and unity power factor at all times. The actual efficiency achieved could typically achieve an efficiency of 98 percent.
Rising energy prices and, for larger organisations at least, the need to buy carbon credits, makes ignoring ongoing costs simply unaffordable. UPS in the most mission critical applications should not represent the cost that it too often does: it’s well worth ensuring that any supplier can apply creative thought to addressing needs in the most cost effective way possible.
● Rob Tanzer is Chloride’s Technical Support Manager