Guide to UPS Selection for Critical Applications

Escalating energy costs coupled with a drive to reduce carbon footprints have resulted in a
number of changes in the UPS industry. An initial development was the introduction of modular,
scalable UPS solutions which enabled higher system efficiency through ‘rightsized’ power
protection. However, recent product introductions boast up to 99% efficiency. While such
numbers may be eye catching to potential purchasers, ‘high efficiency’ options can increase
potential risk to the critical load.

There has been a general assumption all UPS designs and technologies are born equal. It’s not
difficult therefore to see why there has been a certain tendency to conclude all UPS topologies
afford similar levels of protection. Essentially there are two generic types of UPS available at
present - Static and Rotary covering various topologies (VFI,VI & VFD)- although it is important
to understand that Static UPS enjoy the lion’s share of a global market forecast to grow in 2010 to
$7.3b (IMS Research, Uninterruptible Power Supply Sourcebook, July 2010).

Static units range from 200 VA to 1600 KVA as single modules which can be connected in parallel
to achieve larger outputs and or redundancy of N+1 or N+N. Rotary UPS are mainly available
from 500 KVA to 1600 KVA as single modules and suitable for paralleling to reach larger ratings.
There are a number of key areas to consider when selecting which type of UPS system will
provide optimum protection at a competitive price. Criteria include criticality, UPS type,
configuration/ availability, load rating/ battery autonomy, maintainability & installation/ structural
considerations.

Criticality

The very first question is one of mission criticality, in other words what are the consequences for
an organisation if a mains disruption or outage crashes its computer system or renders IT services
unavailable. Depending on the application consequences may extend from a temporary
inconvenience to the paralysis of the entire business. Apart from just continuity of power, quality
of power is also very important for IT equipment.

Appropriate UPS selection should therefore provide protection from grid related disturbances
including blackouts, brownouts (sags), dynamic overvoltage, overvoltage, undervoltage, transients
(surges), frequency variations, voltage distortion Hf (burst) and harmonics. In addition, power
disturbances generated within data centres, caused by lifts, HVAC, inrush current from printers or
even a loose neutral connection must also be mitigated.

UPS types and codes

Industry standards have been developed to describe the technical characteristics and therefore
protection afforded by various UPS topologies. IEC 62040-3 provides very clear definitions for
each UPS type complete with typical block diagrams :

1. VFI – the UPS output is independent “of Input mains supply Voltage and Frequency
variations”, i.e., the primary power path is the inverter and not the AC mains for true double
conversion on-line topology.

2. VI - the UPS output is dependent “on Input supply frequency variations, but Mains supply
Voltage variations are conditioned (independent)...”.

3. VFD - the UPS output is dependent “ on Mains Input Voltage and Frequency variations”.

Ensuring the correct balance between efficiency and resilience.

Double conversion (VFI) UPS systems protect the load continuously (with no switching risks)
against voltage/frequency variations from the utility. However, in the recent past a number of UPS
manufacturers are trying to reach unrealistic high (99%!) efficiency by sacrificing electrical
power quality demanded by Servers and adding risk of unwanted source switching right at crucial
moment of UPS operation. In other words such UPS systems are expected to switch between
various modes of operation –VFD, VI and VFI. This kind of operational switching can impact on
system resilience (due to prolonged switching periods) and reduce MTBF resulting in very low
Availability values due to increased MTTR.

In addition to availability implication, high efficiency UPS also sacrifice power quality demanded
by IT equipment. It’s essential never to expose the IT load to raw mains by running the UPS in
bypass mode. Even if the bypass path has a harmonic conditioner and choke, it does not provide
clean, conditioned and isolated power required by IT equipment. A series choke within the UPS
bypass path will limit fault clearing capacity, increase losses due to load harmonics and make
discrimination more difficult.

A typical diesel rotary UPS is not double conversion and does not correct frequency swings
during normal operation. Furthermore, Rotary units have a very high component count resulting in
very low MTBF. When utilised for IT loads, the units need a 150% oversized neutral to handle a
high level of Triple ‘n’ harmonics. Blade servers impose a leading power factor load on UPS and
therefore this must be a key factor during selection of larger UPS systems. However, UPS
systems can be highly underutilised making it necessary for input harmonic filters to hold the
distortion level (THDI) at 5% right across the load range. There is a trend to utilise a PFC type
active charger resulting in very low (4%) input harmonics and almost unity input power factor.

From a scalability point of view, static units are easily retrofitted within a building, but the same
cannot be said for rotary UPS, mainly due to structural issues and noise problems. Further
complications may have to be addressed due to vibrations, fuel storage, fire risk, wet stacking
(caused due to light load running) and exhaust fumes.

Load rating / battery bank

Sizing of UPS needs to account for Crest Factor and harmonics from the PCs and servers, not
ignoring blade servers which have leading power factor and very high inrush currents. Generally
all UPS units are rated for specific KVA of load at 0.8 lagging power factor. Battery autonomy
must be adequate to protect the load during blackout and batteries need to be monitored as they
are the weakest link in any power protection system. For larger systems it is worth sizing the
battery for end of life and to meet ENBS6290 Pt.4 1997. Multiple battery strings are advisable for
critical applications.

It’s common practise to have up to 15 minutes autonomy backed by diesel generators capable of
handling leading power factor loads presented by blade servers. Sites without secondary
generators may need a minimum of 30 minutes’ autonomy plus software to ensure a graceful IT
shutdown and to protect against data corruption or loss.

Configuration / availability

In order to achieve high levels of availability, built-in system redundancy (N+1 or greater) and fast
service from manufacturers are recommended. It may be worth using static transfer switches
near the PDU since this improves both availability, maintainability and limits fault propagation.

Installation / maintainability

These disciplines are subject to an article in their own right as the siting and installation of
equipment is a complex subject. Static UPS does not require air-conditioning but battery room
needs to be maintained at temperature range - 10 to 25 C for extended life. From a maintenance
point of view, this should be carried out on a regular basis either by the manufacturer’s own team
to protect product liability insurance cover, or at the very least by factory-trained personnel if
using a third party for this function. Maintainability and spare parts provision are important as
these have a profound effect on availability.

Conclusions

When selecting UPS systems for critical applications, the specifier must consider the impact of a
loss of IT services on the organisation, since this will almost certainly cause financial loss or
damage to brand. Newer, high efficiency static UPS may achieve 99 % efficiency, but only
momentarily and with all the associated risks to the IT load posed by VFD operation. Rotary UPS
pose a number of problems, limited scalability, the noise they create and their long maintenance
period requirements.

Double conversion (VFI) UPS systems are up to 97% efficient and will protect an IT load
continuously (with no switching risks) against mains events and outages. A small trade-off,
perhaps, but compensated by a great deal of peace of mind where IT continuity and resilience is
vital.

Guide to UPS Selection for Critical Applications