How to size your UPS needs in 5 simple yet powerful steps
Most IT professionals understand what a UPS is, but fewer are confident knowing exactly how big it needs to be to ensure their datacentre or comms/server room has enough power in case of emergency.
What is a UPS?
Why are UPS Systems Important?
How much do UPS systems cost?
What kinds of UPS are there?
How do I go about choosing a UPS that’s right for my datacentre?
What happens if the UPS fails?
How can I prevent UPS failure?
Glossary of terms
The perils of undersizing or oversizing your UPS
The last thing you want when you’re contingency planning for continuous datacentre uptime is an undersized UPS. By the same token, you don’t want the wasted expense of a supersized UPS either.
Follow these steps to right-size your UPS to cost-effectively match your unique datacentre requirements so that you have the power protection you need in case of emergency.
The 5 Steps for UPS Right-Sizing
How to calculate your power load with some simple maths
Read the power rating information glued onto the nameplate of each piece of datacentre equipment. This is the simplest starting point for determining your power load.
The important information to look out for is the operating voltage (V) and current (amps, or A). Multiply these together to calculate volt-amps, or VA (VxA=VA). This figure is essentially the same as watts, at least for DC power anyway. AC is slightly different but unless your datacentre power requirement is particularly large, not a critical consideration at this stage.
These ratings can’t be entirely relied upon to calculate your UPS requirement because they denote the maximum possible power consumption of the device concerned – with all of its capabilities turned on and operating at 100% utilisation – rather than telling you the real-world usage.
- Write down each piece of equipment your UPS needs to protect
- Find out how many amps and volts each uses
- Multiply the two numbers together to calculate VA
- Add up all the VA figures for the devices on your list
Many data centre professionals will take it upon themselves to scour their environments with a clipboard, calculator and magnifying glass, adding up the overall power load. The problem with this approach is that the nameplates typically provide a range of figures rather than a single, absolute number. Manufacturers are also at liberty to round values up to the nearest unit so that, for example, 3.5 becomes 4. That doesn’t sound much, but is actually a double-digit percentage increase!
A better starting point is to use the online power load configurator tools provided by the major manufacturers, like this one from Dell. These enable you to approximate your power load based according to a range of factors, all without having to manually search high and low for nameplates. Another approach is to extract load data from existing UPS platforms, either by manually recording readouts or running and aggregating automated reports. UPS platforms that are governed by overarching DCIM software will give the most accurate information in a simple and automated fashion.
How to zero-in on kVA rating without getting distracted by kW
VA is volt-amps, while W is watts (k is just the multiplier of 1,000 e.g. 1kW = 1,000 watts). Watts is a measure of real power while VA is a measure of ‘actual’ or apparent power. The difference is basically between what you are promised and what you receive.
This is important in the world of UPS because of what is known as the power factor (PF). For electrical systems that are able to convert their promise of power into reality, the PF is exactly 1.0 and never a dot higher. Few devices achieve this, and typically only small, simple items (the classic example is a light bulb). For everything else, the PF is less than 1.0.
The latest generation of UPS solutions have a fairly high PF of 0.9 or higher, in line with the characteristics of the data centre components they protect. More ancient varieties of UPS drags behind on 0.8, 0.7 or even lower. This explains why, in order to achieve 500kW you need a UPS rated to around 550kVA. Looking on the bright side, an old UPS would have needed to be rated at more like 700kVA.
Aim for high UPS utilisation for the most efficiency
The temptation to over-spec UPS capacity hits a brick wall with the reality of how inefficient these systems become when not fully utilised. UPS solutions operate best at the upper range (80%+) of their usable capacity because they lose so much unutilised energy as heat at lower bands. Running a UPS at 20-50% load will result in a noticeable drop in energy efficiency, often as much as 5–10%. That inefficiency can easily translate into thousands of pounds in wasted energy bills. And the double-whammy is that you have to overcompensate with additional cooling load too! Yet more reasons why, whatever UPS you select, it needs to be as close to your requirement as possible.
The optimum utilisation range for UPS solutions is 80%+
This phenomenon can really bite when setting out your UPS redundancy requirements. Deploying two UPS platforms for 2N redundancy effectively splits the total load right down the middle. The redundancy is there so you can call upon a single UPS to run at 80% utilisation, but this necessitates running both at 40% utilisation at all other times – impacting your energy efficiency. Modular UPS systems that enable N+1 redundancy typically overcome this issue (see Step 5).
Add sufficient headroom into your power load calculations
The biggest challenge with knowing your power load is the dynamic nature of data centre environments. Your current data centre is drawing power 24/7, but the precise amount will fluctuate in line with the changing processing requirements placed on each component. For example, a dense server array churning through intense processing cycles, or a cooling unit working harder in response to higher ambient temperatures. This gives rise to the notion of peak power loads that may be considerably higher than the average.
DCIM solutions give an accurate view of true power load
Calculating your power load is one thing, calculating having enough to sustain data centre operations in the event of a main power outage is something else. So as well as anticipating the peak loads that a standby UPS may be required to deliver, you should also consider the short-term growth that may impact your environment. Going to the trouble of deploying a 100kVA UPS might satisfy all your immediate needs, but not if 30 days later your requirement has jumped to 120kVA.
TIP: Add 20% to your total kVA calculation to allow for future growth
Ultimately an experienced UPS professional is best placed to use their experience and specialised toolsets to determine your prevailing power load requirement.
Look at modular UPS solutions when planning for the long-term
According to analysts at Frost & Sullivan, the adoption of modular UPS solutions is growing at over 20% as new data centre demands require optimum flexibility.
Modular UPS soltuions enable you to call upon extra UPS capacity incrementally, either through the design of the system allowing for the plug-in of more physical modules at some later date, or with the system being provided fully-loaded but with unwanted capacity disabled until you need to call it on stream. Modules range in size, but are frequently 10kVA or 20kVA for smaller UPS deployments and 250kVA for larger systems, and can scale up to very large systems.
The economics make sense from both an energy efficiency (only using what you need, and running UPS solutions to higher utilisation) and a capex perspective, which is why modular UPS solutions are so popular in modern data centre environments.
Modularity also assists in the planning of additional redundancy so that UPS power protection can be all but guaranteed even in the event of a component failure. This also enables UPS maintenance to be carried out without taking the entire UPS solution temporarily out of commission, thereby avoiding the data centre being exposed to significant risk.
Again, an experienced UPS professional is best placed to advise upon the selection of UPS solution in accordance with redundancy requirements.
Comtec Power’s award-winning engineering team has been right-sizing UPS solutions for UK datacentres for nearly 20 years and is accredited to the highest standards with market-leading UPS manufacturers including Schneider Electric and Riello. Recent customer deployments include a top 20 UK university, Britain’s biggest social housing landlord and a global banking corporation.
Glossary of Terms
A level of redundancy equating to a fully mirrored system. In a 2N configuration, the entire infrastructure requirement of multiple components can be fulfilled by an entire secondary infrastructure waiting to take over.
Amps or Amperes are the unit of measurement for electrical current.
Direct Current (DC)
Distinct from AC (Alternating Current), DC denotes electrical current that flows in a constant direction.
Datacentre infrastructure management (DCIM) solutions are used to monitor and control both IT and facilities management metrics within a single console.
1,000 volt-amps. Volt-amps (VA) are the unit of measurement for actual or apparent power. The VA of an appliance is calculated by multiplying its voltage to its operating current.
1,000 watts. Watts are the unit of measurement for real electrical power.
A marginal level of redundancy enabling an infrastructure to continue operating despite the failure of an essential component. In an infrastructure with four components, N+1 would operate five.
The miniature physical record of electrical values for a given appliance found embossed or attached to its surface.
Power Factor (PF)
The ratio of the real power used by an appliance to do its work as a proportion of the apparent power supplied. Typically applied to a kW figure to determine the required kVA rating of UPS solutions.
Uninterruptible Power Supply. A battery-based hardware platform that provides a reliable and appropriate level of electrical power – typically to IT systems / datacentres – in the event that mains power is lost.
The track record of availability performed by IT systems over a given period. Uptime is expressed in percentage terms (e.g. 99.999% uptime) and normally covers one year.
Volts are the unit of measurement for electrical pressure, otherwise known as ‘voltage’.