For most data centres, deciding on power protection requirements for today is fairly straightforward, but factoring in what you may need in the future can be hit and miss.
Over-sizing a UPS has been the traditional route but it carries with it a higher capital outlay and delivers poor efficiency. Right-sizing for today – without considering future requirements – may result in the disruption and cost associated with ripping out and replacing the system before the end of its design life.
Neither scenario is acceptable for today’s DCMs (data centre managers) who are facing increasing pressure to cram in more capacity while cutting costs. Their problems are compounded by having to meet ‘green’ initiatives from the European Union, Energy Star, The Green Grid and others; designed to increase efficiency and decrease the environmental impact of electrical equipment.
In October 2011, international standards agency Energy Star issued a new version of its UPS Specifications and Test Method standard, which will become effective by April 2012. The specification defines new requirements for UPS including minimum efficiency and input Power Factor.
Deciding what’s important – energy efficiency or business resilience, cost or reliability – is the first step in the decision process. This is not to say that DCMs cannot enjoy all four elements but a hierarchy of priorities will determine which is prime concern and how much it will cost.
The criteria that will determine a data centre’s power protection requirements are: criticality, availability, load rating, autonomy, maintenance and installation.
What about tomorrow?
Who knows, beyond a best guess, how much their load rating will have increased in five years time? Criticality and availability may not change but at what pace will load rating rise as the business grows and more equipment is installed? How about battery autonomy? Will the need for longer back up time increase? All of these questions add to the challenge of specifying a UPS.
Data centres almost always run mission critical applications. Any downtime caused by cuts or glitches in mains electricity are intolerable and a continuous source has to be assured. This is done by installing UPS and an alternative energy source such as a diesel generator or additional battery strings. There is also a secondary, but equally important, role for UPS, which is power conditioning.
Grid electricity is prone to disturbances such as power sags, surges, brownouts, over-voltages, harmonics and the like. Such problems can also be generated from within the data centre. The UPS acts as a buffer, conditioning mains electricity and protecting downstream loads. In selecting UPS systems for data centres, criticality will likely be priority number one - over and above cost and energy efficiency. How critical power protection is will dictate the type and topology of system that will be suitable.
Data centres running mission-critical applications will require the highest levels of availability and this will greatly influence the type and topology of UPS that is chosen.
Today’s DCMs have a choice between static and flywheel UPS and both types offer designs that provide redundancy, are highly efficient and deliver high-levels of reliability and scalability specifically for mission critical data centres. As for static UPS, the topologies available are online VFI (Voltage and Frequency Independent), line interactive VI (Voltage Independent) and offline VFD (Voltage and Frequency Dependent). Online VFI UPS topology offers the highest levels of reliability and availability and is recommended for mission critical application.
To accurately size a UPS, DCMs must take into account the types of load that the UPS system will be powering and whether they are inductive, capacitive or resistive. This will decide whether they have a leading power factor or lagging power factor. Some loads require tight voltage and frequency regulation while others (usually essential but non-critical loads) do not. SMPS loads, for example, are normally inductive, whereas Blade servers will be capacitive. When selecting UPS for mission critical applications, a product with a high output power factor should be chosen as it will be able to protect both new and legacy computer loads. Other issues affecting load-rating calculations are real power verses apparent power. This is the difference between VA (Volt-Ampere) and W (Watts). Confusion happens when electrical equipment manufacturers opt for different units when displaying nameplate power ratings – some display in VA and others in W. When calculating the total power rating of all equipment that will make up the UPS load in a data centre, users need to ensure they have accurate calculations for both and that they have confused one with the other.
Battery autonomy (also known as back up or discharge time) is a measure of the time for which the UPS battery (or battery banks) will support critical loads during a mains power failure. Autonomy is a function of battery charge state, capacity and load size and for data centres, the need for additional autonomy (in terms of time, say, from two hours to four and in terms of capacity, from 300kVA to 600kVA) may increase over time. This will have a bearing on future space requirements and UPS load rating.
How a UPS system is maintained over its lifetime will greatly affect reliability, how it performs and the period of its useful life. A static UPS does not require air-conditioning but a battery room does. UPS battery life must be maintained at a constant 21-25 degrees centigrade to preserve it. Having to replace banks of UPS batteries because of poor temperature control can be costly. UPS maintenance should be regular, routine and carried out by specialist third-party providers or the UPS manufacturer’s own team of engineers with provision for spare parts (as this can affect overall availability).
It’s worth mentioning this because data centres often house UPS systems in specially constructed rooms, away from areas of heavy people traffic and adjacent to computer rooms. Thought must be given to how the system will be expanded over time. Has space provision been made for additional UPS battery strings, for example, or additional UPS modules in either parallel-redundant N+X systems or further modules in a modular solution? Construction work is expensive but it is the one area in which over-sizing to future-proof requirements could be cost-effective in the long run.
In order to address cost and efficiency, alongside a data centre’s need for resilience and high-levels of availability, DCMs need to consider the future now. Obviously, none of us has a ‘crystal ball’ but it is possible to formulate an idea of how much (and how quickly) your requirements will expand over time. Be clear about criticality and availability. Choose a UPS topology to suit requirements, installation and future expansion. Be accurate about load rating and type. Choose your level of autonomy - and build to accommodate.
Tags: Power & Cooling