Journal of Power Sources, 23 (1988) 211 - 220 211

BATTERY REQUIREMENTS FOR UNINTERRUPTIBLE

POWER-SUPPLY APPLICATIONS

G. D. BROWNLIE
Kabanos Pty Limited, l/6 Gladstone Road, Castle Hill, N.S. W. 2154 (Australia)

summary
There is an expanding market for small-to-medium capacity secondary
batteries for use in uninterruptible power supplies (UPSs). UPSs are com-
monly used to provide protected power to computer installations, point-
of sale terminals, and essential telecommunications equipment. The capacity
of currently available UPSs is, typically, from 500 VA to several hundred
kVa, with power-backup from ten minutes to one hour at full load. All
UPSs incorporate secondary batteries; usually sealed lead/acid cells. Battery
requirements include: high reliability; extended operating life; tolerance to
extended float charging; and good, deep-discharge characteristics at high
discharge rates. Recent research and development experiences in the design
of a range of UPSs is described and the need for batteries optimised for UPS
applications is highlighted.

Introduction

Uninterruptible power supplies (UPSs) have been in use for decades

in critical power-supply applications. Traditionally, their use has been
limited to power supplies for essential telecommunications equipment,
medical equipment, and emergency lighting. A variety of rotary and static
power supplies, all including secondary batteries, has been used, generally
providing power for the ten to fifteen minutes required to start, and bring
a standby diesel generator on load.
The application of solid-state power electronics has enabled the devel-
opment of compact, efficient, uninterruptible power supplies which are
now finding a wider range of applications, particularly in the protection
of computer-based equipment. Such equipment (including personal com-
puters, minicomputers, and the disk drives of major computer installations
and point-of-sale terminals) can be damaged by power interruptions, or data
stored either in volatile memory or on disk can be irretrievably lost. UPSs
provide a cost-effective means of protection against power failures.
The battery requirements for UPS applications are unique in that
batteries must have long life and be suited to extended float periods while
providing good, deep-discharge characteristics at high discharge rates. The

Elsevier Sequoia/Printed in The Netherlands

212

author’s company is currently designing a range of UPSs with ratings from

500 VA to 20 kVA, for both 2201240 V and 110 V power systems. Research
into battery performance characteristics has highlighted the need for bat-
teries optimised for UPS applications. This paper describes the architecture
of typical UPSs and standby power supplies as well as the general charac-
teristics of the requisite batteries, and discusses in detail the requirements of
the ideal UPS battery.

UPS architecture

Two distinct designs are commonly used for power backup applica-
tions, as shown in Figs. 1 and 2. The most common architecture for non-
critical applications is the standby power supply (SPS), in which the mains
is normally connected to the load via a solid-state ‘static’ switch and the
battery/inverter system is switched to the load only in the event of mains
failure. The mains-powered battery charger is rated to recharge the batteries
within a period of, typically, 4 - 10 h, and normally float charges the bat-
teries. Advantages of the SPS include relatively low cost, low stress on the
inverter (which is on load only on mains failure), and low charger rating
(typically one-fifth of the inverter rating). Disadvantages are the lack of
mains regulation and the mains-to-inverter transfer time of 3 - 10 ms, which
may seriously affect certain loads. In addition, the SPS output is not isolated
from the mains and the only transient and radio-frequency filtering is that
provided in the mains-line filter.
The architecture of a true UPS is shown in Fig. 2. In this case, the
inverter is always on line and the full load of the UPS is also carried by the
charger, in addition to the power required for battery charging. There is
normally no direct mains link between the input of the UPS and the output,
unless a bypass line to protect against failure of the charger or inverter is
provided. The UPS provides continuous, isolated, regulated and filtered

I
CONTROL

‘Fig. 1. Stand-by power supply architecture.

MAINS * CHARGER
-L- DC/AC , PROTECTED ~
INPUT INVERTER OUTPUT
-t
Fig. 2. True UPS architecture.
 213

power to the load. The system is most suited to applications where any
mains interruption, such as that during the transfer from mains to inverter
in an SPS, could result in a disruption of the load operation. The UPS is
also the appropriate source of backup power where good voltage regulation
is required. While the UPS provides truly uninterruptible power and no
discernible output transients on mains failure, its disadvantages include
higher cost and the need for greater circuit reliability, as both charger and
inverter must continuously carry the full load of the UPS.
The technology used in backup power systems has evolved over the
period of their application. Thyristor circuits have traditionally been used in
battery chargers, but switched-mode power supplies are now being used at
charging capacities up to 10 kW. High-power, solid-state switches have been
employed in inverters for some years, normally driving the load through
a 50 Hz isolating transformer. Recent developments in inverter design
include the use of transformerless circuits using high-frequency, pulse-width
modulated inverters, commonly using power MOSFETs, and the use of
high-frequency cycloconverters to generate sinewave outputs.
In general, high-frequency switching technology allows more efficient
chargers and inverters to be implemented. Circuit efficiency of the inverter
may be as high as 85%, allowing much more efficient usage of the energy
stored in the batteries. Other benefits of high-frequency switching tech-
nology include lower power dissipation (allowing circuitry to run cooler) and
more compact UPS systems.

Typical UPS performance requirements

The requirements for UPS and SPS batteries are determined by the per-
formance schedules of the backup supplies. A power backup time of 10 min
is generally considered to be the minimum acceptable for most applications,
though some systems may require considerably longer backup times. The
shorter backup times are more demanding on UPS batteries, as the batteries
must operate at higher discharge rates. Given the UPS power rating, the
circuit efficiency and the battery voltage, the battery capacity can be deter-
mined.
High reliability is an essential requirement for UPSs. The latter must be
intrinsically more reliable than the available mains power, otherwise the UPS
is of no benefit. With developments in switching power circuitry, it could be
argued that the reliability of the UPS is largely determined by that of the
batteries. It is considered that a mean time between failure of at least three
years is required. With UPS battery rail voltages between 48 V and 240 V
being typical, a large number of batteries may be used in a UPS, resulting
in a greater probability of failure, so good battery life is essential.
Other UPS performance requirements are important for particular
applications, such as output waveform quality and transient response under
214

changing load conditions. However, these do not affect battery requirements

and are not discussed in this paper.

Battery types

UPS battery characteristics are summarised in Table 1. The most com-

mon battery used in UPS systems is the lead/acid battery. Large UPSs
require large numbers of batteries; vented wet cells are normally used and are
racked according to normal battery practice. Small UPSs commonly use
sealed lead/acid cells which allow flexibility in operating position and
minimal maintenance. In general, nickel/cadmium batteries are not used,
though the claimed reliability of these batteries makes them suitable for such
application. Problems with nickel/cadmium batteries include the wide range
of cell voltage over charge and discharge operation (which can cause inverter
design problems in the UPS architecture), and low energy density.

TABLE 1

UPS battery types

Lead/acid batteries (2 V per cell)

(i) Wet cells
- 25 W h kg-’
- 40 W h 1-l

(ii) Sealed cells

- 30 W h kg-’
-75 W hl-’

Nickel/cadmium batteries (1.2 V per cell)

- 10 W h kg-’
- 15 W h 1-l

Battery requirements

From the above, it is seen that the essential requirements of UPS

batteries are as follows:
(i) high reliability, with a minimum life of three years;
(ii) good, extended float-charge characteristics;
(iii) good, high discharge-rate performance;
(iv) minimal maintenance requirements.
Other battery requirements that are important in some applications include:
(i) tolerance to a wide range of operating temperature;
(ii) high energy density;
(iii) ruggedness and good resistance to damage in transport;
(iv) operation in any orientation (sealed batteries only).
These requirements are discussed in more detail below.
 215

Battery life
Battery reliability is essential for satisfactory UPS operation. While
the reliability of typical UPS circuitry is the subject of some differences of
opinion, many in the power-backup industry, and many users, consider that
the batteries determine the system reliability. This is particularly so in the
larger UP%, which may contain up to 100 batteries.
A minimum life of three years is required. Proven reliability is a factor
affecting choice of batteries by the UPS manufacturer, though it is difficult
to establish the validity of life figures published by battery manufacturers,
particularly for new technology batteries which may offer considerable
benefits in UPS service. For this reason, some UPS manufacturers take the
conservative approach and use only batteries of known reliability.
The nature of batteries also makes quality assurance for the UPS
manufacturer difficult. The life and capacity of a battery is affected by its
history, including factors such as time since manufacture, storage tempera-
ture, and charge state. It is both expensive and time-consuming for UPS
manufacturers to screen batteries by subjecting cells to full charge/discharge
tests, though selected samples are normally subjected to full testing. The
UPS manufacturer must be confident that installed batteries will last for a
minimum of three years in service.
Battery life is also determined by conditions in service. Most UPS
systems are continuously on line and the batteries are on continuous float
charge. The design of the battery charger and the float performance of the
batteries are therefore critical to UPS reliability, as discussed below. In
addition, some batteries may require periodic discharge to ensure extended
life and to test the discharge performance of the batteries. Operating tem-
perature also affects battery life and charging requirements. Batteries also
exhibit ageing effects, and capacity may be reduced by operation at sus-
tained high temperatures. It is essential for UPS applications that the decline
in capacity does not seriously reduce the UPS backup time. In specific terms,
expert guidance from battery manufacturers spelling out the operating
conditions required for maximum battery life in UPS service would be
appropriate.

Battery charger requirements

Charger requirements differ for SPS and UPS applications. In the
SPS, the charger is required only to charge the batteries and does not carry
any of the output load. In normal service, the batteries are fully charged
and the charger is merely used to provide the necessary float charge. In the
true UPS, the charger must carry the full load of the inverter (see Fig. 2),
as well as charge the batteries, all from the same charger output. A more
complex and reliable high-power charger circuit is therefore essential.
Normally, the charger is rated for the full UPS output load, plus losses in the
inverter, together with a margin for battery charging of 20% of the rated
UPS output.
216

For lead/acid batteries, charger requirements are well established. In

larger UPS installations, a charger which allows full control of boost and
float charge modes is normally required, and it is possible to optimise the
charger output characteristics to meet the battery demands. A boost charge
voltage of 2.50 V is usually specified and a float voltage of 2.25 V per cell is
standard. Some critical applications may call for in-circuit monitoring of the
voltage of each cell in the UPS battery string. In smaller UPSs, cost pressures
may result in non-optimal charger characteristics, in which a boost mode is
not available. However, it is essential that the float-charge characteristics are
optimised for long battery life. The charger is normally designed to provide
a constant-current output for battery charging, with output voltage limiting
at the desired float voltage. The charger output voltage will then vary by,
typically, 10% over a full charge cycle, as shown in Fig. 3, and the inverter
control circuitry must compensate for these changes to maintain a regulated
a.c. output.
A potential problem with battery charging is the temperature depen-
dence of the battery terminal voltage, which, in turn, results in a tempera-
ture dependence of the optimum float voltage. For UPS installations designed
to operate over a wide range of ambient temperatures, temperature com-
pensation of float voltage may be required to ensure long battery life.
Overcharging must also be avoided, particularly in UPSs using sealed bat-
teries. Gassing of cells can then affect capacity, and dangerous gases may not
be able to dissipate safely.
Nickel/cadmium batteries are also used in UPSs. As noted above, the
cost and relatively low energy density of nickel/cadmium batteries tends to
limit their application, and the recommended charger characteristics also
present problems. While the nominal cell voltage of a nickel/cadmium cell
is 1.2 V and does not vary significantly during cell discharge. the recom-
mended float voltage is 1.40 V - 1.42 V and the charge voltage is 1.7 V -
1.75 V. A more complex charger is therefore required to provide optimum

2 4 6 8 10 12 14 16 18 20’

TIME/h

Fig. 3. Typical constant-current lead/acid battery charge characteristics.

 217

charge and float conditions. Nickel/cadmium batteries give rise to a problem

in the true UPS architecture (Fig. 2). The battery rail voltage could be any
voltage between 0.85 V and 1.75 V, depending on whether the batteries are
at the end of their discharge or are on charge. This poses a design difficulty
for the inverter circuitry, which must provide regulated a.c. output over this
range of input voltages. This problem is much less severe for lead/acid
batteries, for which the operating cell voltage range is normally 1.8 - 2.5 V.

Battery capacity and discharge characteristics

Typical lead/acid discharge characteristics are shown in Figs. 4 and 5.
Figure 4 gives the voltage of a typical sealed lead/acid cell, as a function of
discharge time, for a range of discharge rates. Normally, small, sealed bat-
teries are rated at a 10 h (C/10) or a 20 h (C/20) rate, and batteries are
typically operated at 20 times their rated current for a 10 min backup time.
The problem of battery de-rating under such severe discharge condi-
tions is apparent in Fig. 5, which shows the percentage of available C/10
capacity for different discharge rates. It can be seen that typical sealed
batteries give less than 50% of their rated capacity at 20 times their rated
discharge current. The performance of nickel/cadmium batteries is mar-
ginally better at, typically, 5596, for the same discharge conditions. Obviously,

1.6

0.6

\ /’
0.6 .

dec _ min h *
I __
5 10 20 401 2 4 6610 20 401 2 4 6 610 15 20 40 60 100
DISCHARGE TIME
Fig. 4. Typical lead/acid battery discharge characteristics.
218

05 1 2 3 5 10 20 35 70 150
MULTIPiE OF DISCHARGE CURRENT (C/10)

Fig. 5. Lead/acid battery de-rating with load

more of the rated battery capacity is available if the UPS is required to

provide a longer backup time, but then a greater ampere-hour battery rating
is still required.
There is clearly a need for batteries optimised for short discharge
periods for UPS applications. This requirement is in conflict with the re-
quirement for good, extended float performance as, traditionally, batteries
have been optimised for either deep discharge or extended float operation.
One factor in favour of UPS applications is the relatively low number of
deep-discharge cycles during the life of a UPS. Typically, there may be only
four deep-discharge cycles per year, and then some of these may be battery
tests to ensure correct operation of the system.

Battery maintenance
Low battery maintenance is necessary for UPS applications. In smaller
UPSs, sealed lead/acid batteries have become standard equipment, and with
developments in this technology, the use of high-capacity sealed batteries is
now possible. Sealed batteries are essential for UPSs mounted in
enclosed cases, offering flexibility in battery mounting, as they can be used
 219

in any orientation. For small UPSs, a test backup cycle is recommended at

periods of three months to verify battery capacity. A problem common to
all small UPS systems is that the UPS is generally used in a non-technical
environment and specialist maintenance staff are not available. Therefore, all
battery maintenance must be minimised.
Larger UPS installations generally require conventional wet-cell bat-
teries. These are installed either in special ventilated cabinets or in special
battery rooms. Regular maintenance of wet-cell batteries is required, and in
some UPS systems it is possible to isolate and remove part of the battery
string for maintenance while the UPS is on line. Many large UPS applica-
tions are in telecommunications where technical staff are normally available
to perform battery maintenance, and the need for regular battery attention
is not seen as a disadvantage. While sealed or low-maintenance batteries can
be used in large UPS systems, the additional capital cost is not normally seen
to be warranted.
As in any battery application, periodic test discharges may be required
to ensure correct battery performance and capacity. Occasional boost
charging may also be necessary to equalise cell charges. Some means of
monitoring cell condition on-line is also of benefit in large installations. It
would also be useful to monitor automatically cell conditions in small UPS
systems, so that the user can be confident that the desired power backup
will function when necessary.

Requirements for special batteries

In addition to the requirements for batteries optimised for UPS appli-
cations, current developments in UPS technology have indicated the need
for special UPS batteries, particularly for smaller UPS systems.
Traditionally, a low battery voltage has been used for smaller UPS
systems. This results in less efficient inverter design as the currents in the
power switching devices are high and the internal voltage drops are high re-
lative to the battery voltage. A step-up transformer is also required to ob-
tain the required mains voltage. There would be considerable benefits if the
battery voltage could be made closer to the mains voltage. This would
involve the development of fractional ampere-hour batteries with voltages of
up to 240 V. The battery output could then be switched directly to the
mains by the inverter circuitry and highly efficient, small, UPS systems could
then be developed for special, dedicated applications. It would also be pos-
sible to incorporate a UPS system into electronic equipment at present
sensitive to mains failure, avoiding the need for an external facility.

Conclusion

A sizeable market has always existed for batteries for standby power
systems. This market is now expanding as the benefits of uninterruptible
power supplies are being appreciated in providing protected power to critical
220

equipment. Major applications for UPS systems include: computer installa-

tions, from single microcomputers through to full mainframes; computer-
based equipment such as point-of-sale terminals and electronic funds transfer
terminals; essential telecommunications equipment; and security systems,
The unique requirements of batteries in UPS service include the need for:
high reliability; good, extended float performance; and good, deep-discharge
characteristics at high output discharge rates. While currently available
batteries provide satisfactory performance in UPS service, there is a clear
need for batteries optimised for UPS applications.