UPS systems - Correctly sizing UPS to meet energy efficiency

Typography
Articles

Rising costs and a very real danger of future energy rationing are pushing the need for UPS systems to be correctly sized further up the agenda says Robin Koffler of Riello UPS (right)


Thermal imaging has typically been the exclusive domain of specialists in the field, and has required the use of costly and difficult to operate equipment. With recent advances in sensor technology, this non-contact, versatile measurement technique has become available to a wider audience through dramatic reductions in price and in developments of user-friendliness.
UPS customers are being urged to be energy efficient and play their part in tackling climate change. They cannot afford to have equipment, including UPS systems, running inefficiently nor can they risk an overload situation from ‘undersizing’ that would render equipment unprotected.
This is particularly true in power-hungry data centres where energy consumption in the business world is at an all time high. Not only do new, high-end servers require more power to operate, they also demand greater cooling resources. Every megawatt required to power hardware takes another 1.5Mwatts to cool it according to some reports. The need to prepare for the future is vital as electricity costs over the next five years are set to double, say analysts: In a recent report, consultancy BroadGroup found that the average energy bill to run a corporate data centre in the UK is around £5.3m/year. The company predicted that this would double to £11m over five years.
But is sizing UPS really down to simple maths? There has been a tendency, historically, to oversize UPS to ensure that, when everything is working at full capacity, the system itself is not overloaded (a situation to which some UPS respond by shutting down after a specified period of time, if it’s a double-conversion design, or at best by switching into bypass mode until someone notices, thus leaving critical computer systems vulnerable to cuts in the power supply or problems associated with raw, unfiltered mains energy).
Oversizing UPS leads to higher initial installation and on-going maintenance costs. Undersizing, especially in a busy data centre where new equipment is being continuously added and fluctuation in usage runs up and down the scale like fingers on a Cello, a customer will very soon be in trouble if they attempt to save costs in this way.
Whilst an On-Line UPS has a built-in automatic bypass, running close to its design limits with regular overloads is never considered good practice. The bypass is for emergency situations only and if overloads are a frequent and regular occurrence, it is always best to oversize the system slightly.
According to business continuity supplier Sungard, the proportion of business continuity plans invoked by companies experiencing power failures jumped from 7% in 2005 to 26% in 2006. The reasons for this may be three-fold: either energy supply is becoming less reliable, or existing UPS installations are unsatisfactory or - worse still – non-existent!
So, what needs to be done to ‘rightsize’ UPS? Firstly, the equipment that is being protected (defined as the ‘load’) has to be categorised into critical, essential and non-essential loads.

Critical Loads
Are defined as all the IT and electrical components and equipment that make up the business architecture and without which business continuity would be lost. Servers, routers, computers, storage devices, telecommunications equipment, security and building management systems usually fall into this category.
In this instance, UPS protection will probably require some form of extended runtime to keep equipment running continuously, more often than not it will also require redundancy so that if a power failure occurs and one UPS goes out of action, the other would take over powering the load.

Essential Loads
These are essential to the business but in their absence some semblance of functionality can exist. Essential loads are things like lighting (other than emergency lighting), air conditioning and heating. Some essential loads may need some form of redundancy built into the UPS system but in many instances back-up does not need to be as robust as with critical loads.

Non-essential loads
Are those that the business can survive without for the time it takes to reinstate power, such as printers and canteen facilities for example.
There can be significant differences between the power ratings recorded on rear panel labels and in operating manuals, and the true values drawn by electrical equipment. This is because hardware manufacturers use power supplies rated for maximum, worst-case conditions which are often far in excess of the actual power drawn. Loads can typically be seen running at only 50-60% of this total capacity. In addition, any ratings given may be in amps or watts to further complicate matters and there can be quite a difference between actual in-rush (start-up) and running power requirements.
Factors that must be considered when sizing UPS loads include:
• Apparent power (VA)
• Active power (W)
• Power Factor (pf)
Other factors that need to be considered if the right UPS is to be installed include expected response to overloads, which should only be intermittent, battery runtime required, fault tolerance (resilience) levels to be reached, the type of electrical installation in terms of the supply and load voltage and frequency requirements and the potential for future system expansion.

Apparent Power
Volt-ampere (VA) is a unit of measure for apparent power drawn by an electrical device. Once known, this figure can be matched to an appropriately-sized UPS. VA is calculated by multiplying the RMS source voltage (V) by the current drawn in amps (A). apparent power (VA) = volts (V) x amps (A)
For example, if an electrical device is connected to a 230Vac single-phase supply and the current drawn by this device is 10 Amps, the resulting VA value would be: 10 x 230 = 2300VA or 2.3kVA.
For a three-phase load, the calculation is slightly different. A 15kVA three-phase UPS will supply a maximum of 5kVA per phase (15