Increased power protection with parallel UPS configurations
Increased power protection with parallel UPS configurations
Making the selection between Distributed Bypass and Centralized Bypass systems
Janne Paananen
Large Systems Group Manager, Three Phase Solutions
Eaton Power Quality
Sini Syvänen
Product Applications Engineer, Three Phase UPS Products
Eaton Power Quality
Executive summary
Organizations such as large data centers, banks and hospitals depend on reliable electricity to safeguard
their critical data. A parallel UPS system continues to maintain power to the critical loads during
commercial electrical power brownout, blackout, overvoltage, undervoltage, and out of tolerance
frequency conditions.
Paralleling provides an excellent solution for matching an organization’s growth needs while extending
the value of existing UPSs. This white paper discusses the main differences and typical concerns about
Distributed Bypass and Centralized Bypass systems to help you determine the suitable solution for your
Table of Contents
The need for parallel UPS systems ......................................................................................................................... 2
How paralleling technologies enhance the reliability of electricity ........................................................................ 2
Understanding Distributed Bypass and Centralized Bypass systems ...................................................................... 3
What are the effects on reliability - or are there any? ..............................................................................................4
How about operating multiple static switches simultaneously under fault conditions? ..........................................5
Configuration of the static bypass switch for load support .................................................................................... 6
Centralized or Distributed? Choosing the parallel UPS system ............................................................................... 8
Concluding thoughts .............................................................................................................................................. 9
About Eaton........................................................................................................................................................... 9
About the authors ................................................................................................................................................. 9
May 2012
Page 2 of 9
The need for parallel UPS systems
Uninterruptible Power Supplies (UPSs) provide continuous power to electronic systems to safeguard
business-critical data. If the UPS needs to switch offline for some reason, it switches to an internal bypass
path, and critical loads run off utility power until the UPS can be brought back online.
Parallel operation extends the normal operation of a UPS by offering increased capacity and/or redundant
capability. The parallel system continues to maintain power to the critical loads during commercial
electrical power brownout, blackout, overvoltage, undervoltage, and out of tolerance frequency
conditions. The architecture of such a power protection system is designed to prevent the loss of valuable
electronic information, minimize equipment downtime, and minimize the adverse effect on production
equipment due to unexpected power problems.
Organizations such as large data centers, hospitals and banks are increasingly finding that using straight
utility power is risky, even if used for short periods of time. Because the cost of downtime and the risk of
losing data are too high, organizations deploy redundant UPS systems to ensure electrical supply even in
cases when one UPS ceases to operate.
How paralleling technologies enhance the reliability of electricity
In paralleling, two or more UPSs are electrically and mechanically connected to form a unified system
with one output — either for extra capacity or redundancy. In an N+1 redundant configuration, there
would be at least one more UPS than needed to support the load. As a conjoined system, each UPS
stands ready to take over the load from another UPS whenever necessary, without disrupting protected
A redundant UPS configuration is designed to ensure that critical workloads remain protected even if one
or more of the UPSs within that configuration becomes unavailable. Parallel redundant configurations,
including N+1 and N+N architectures, are among the most common and effective varieties.
Figure 1: A parallel Eaton 9395 UPS system. In this example, each UPS includes two power modules
(UPM) and one static bypass.
In the parallel redundant system, the electrical failure of any UPS power module (UPM) results only in the
affected module isolating itself instantly, and not shutting down the entire system. The remaining UPMs
continue to support the critical load, with conditioned power, and thus the mission reliability is enhanced.
The reliability benefit is due to the redundancy in the protected power. If the system operates as intended,
it is extremely unlikely that the user would have to operate from the straight utility power. Any equipment
failure is handled by the redundancy of the system, by isolation of the failed component, and the transfer
to bypass is only used as the very last resort. In essence, mains bypass power would only be used due to
UPS-external factors such as overload, over temperature or short-circuit. Routine maintenance of the
UPS system should not require transfer to bypass.
May 2012
Page 3 of 9
Understanding Distributed Bypass and Centralized Bypass systems
There are typically two different types of parallel UPS systems: distributed bypass systems and
centralized bypass systems.
In the distributed bypass system, each UPS has its own internal static switch, rated according to the UPS
size. Each UPS monitors its own output, and if the UPS system needs to transfer to bypass, each static
switch in each module turns on at the same time, and they share the load current amongst themselves.
The distributed bypass system is shown in Figure 2.
Figure 2 The distributed bypass system, with each UPS having its own static switch.
In the centralized bypass system, there is one large common static switch (also known as System Bypass
Module or SBM) for all paralleled UPSs, rated according to the size of the entire system. If the UPS
system needs to transfer to bypass, the load current is then fed through the System Bypass Module. It
should be noted that in this case the UPS units do not include internal static switches. These UPS units
without internal bypass are sometimes referred to as Input-Output Modules, IOMs. The centralized
bypass system is shown in Figure 3.
May 2012
Page 4 of 9
Figure 3: The centralized bypass system with IOMs (UPS 1, UPS 2 and UPS 3) in grey and the SBM
module in blue.
What are the effects on reliability - or are there any?
Reliability and availability are key components in any organization’s IT systems, and therefore UPS
reliability is one of the most important design factors. Measures such as MTBF (Mean Time Between
Failures), MTTR (Mean Time To Repair) and concurrent maintenance capabilities can be used to
estimate the availability of a UPS installation. In addition, more sophisticated methods, such as Markov
modeling, can be used to estimate the critical mission reliability of a fault tolerant system such as a
redundant UPS.
This chapter will concentrate on the static bypass behavior and reliability of a parallel UPS system. When
comparing a centralized bypass with a distributed bypass system, the two most common failure types of
the static bypass should be examined; a static switch thyristor might fail to open-circuit and cannot
operate when it should, or it is short-circuited and remains conducting when it should be off. Let’s
examine the system reliability in these two failure modes.
Static switch open-circuit failure:
In a centralized bypass system, if the static switch has an open-circuit failure, the only static switch in
a system simply fails to operate when needed, thus making the bypass unavailable.
In a redundant distributed bypass system, when one of the static switches fails to operate, the rest of
the switches are still functioning and can support the load when needed. Therefore, the system has
enhanced reliability against this type of fault as the system is not dependent on the operation of a
single static switch.
However, if the distributed bypass system does not have redundancy and is fully loaded, a static
switch open-circuit failure will lead to unavailable bypass, similarly to the centralized bypass system.
May 2012
Page 5 of 9
Static switch short-circuit failure:
The short-circuit failure is a less common failure mode of the static switch. If a thyristor in the static
switch suffers a short-circuit failure, the faulty bypass line connects the system output to the
incoming mains. This creates a safety hazard as it enables feeding power back to the utility from the
UPS system. Therefore backfeed protection is a mandatory safety feature of a UPS installation.
During a short-circuit failure of the static switch, the backfeed protection device can be opened and
the inverters can remain on-line to support the load.
Backfeed disconnect devices together with backfeed detection is implemented internally in all Eaton
3ph UPS equipment as standard. Therefore in a redundant distributed bypass system, each static
bypass can detect the short-circuit failure and isolate it independently, thus leaving enough bypass
capacity to support the load for the system if needed.
In a centralized bypass system the SBM unit also includes an internal backfeed disconnect device as
standard. Therefore static switch short-circuit failure in an SBM can be isolated and the load is
protected in double-conversion. However, the static bypass will be unavailable until maintenance
work is carried out.
How about operating multiple static switches simultaneously under fault conditions?
A common concern regarding distributed bypass systems is the operation of the multiple static switches
under fault conditions. The worry is that the system could not perform simultaneous transfer to the static
bypass under fault conditions, and the static bypasses might become unavailable. This has even led to
some experts having doubts about using a distributed bypass system in their designs for datacenters. To
understand how the distributed bypass system operates during fault situations, let’s look at two cases:
normal transfer to bypass and emergency transfer to bypass.
Distributed bypass system - Normal transfer to bypass
A normal (or planned) transfer takes place when the user commands the UPS system to bypass
from the front panel of the UPS or via an external signal. The transfer to bypass occurs also due to
certain fault situations such as overload, overheating or similar. Whatever the reason, the system
has detected a need to transfer to bypass. In this case, one of the UPSs makes the decision to
transfer and energizes its static switch.
At the same time, the UPS transmits the transfer request for other units over the communication line.
Other units receive this request and transfer to bypass as well. Processing the data to be sent and
received causes only minor delays, maximum around 2 milliseconds, in static switch turn-on times.
This delay is negligibly small since during the normal transfer to bypass the inverters are still able to
support the load. The current levels in system output are on moderate levels, thus not risking the
power devices in static switches.
Distributed bypass system - Emergency transfer to bypass
Emergency transfer to bypass (ETB) occurs when the inverters are not capable to maintain the
system output voltage within normal limits. Most critical situation would be when there is a short
circuit in the UPS system output side. In this case, the inverters are feeding as much current to the
fault as they can to maintain output voltage and will possibly reach their current limit used to protect
the power components.
If the downstream protective devices between the UPS and fault aren’t sized small enough or aren’t
fast enough, the UPS system output voltage will drop and become out of limits. Hence ETB occurs
immediately resulting in a high level of fault current through the bypass to clear the fault. For such a
case, it is very important that all static switches turn on simultaneously to share the high current
among them.
May 2012
Page 6 of 9
In the Distributed Bypass system, each UPS will individually monitor its own output, as well as the
system output, and transfer to bypass if they are out of limits. Each unit will detect the fault in output
The detection of proper output voltage is fast, and the need for ETB is detected approximately at the
same time in all units. They will turn on their static switches independently without the delays of any
communication lines. The possible delays between the units are the result of running the program
loops for output voltage detection. The resulting delay is fractions of a millisecond. Thus they enable
simultaneous transfer to properly share the fault current between static switches.
It is important to understand how a UPS system operates under different fault conditions. As described
above, both bypass configurations are equally reliable since different fault scenarios within and outside
the UPS system have been taken into account in product design. A UPS system with proper fault
detections and backfeed protection devices can operate static switches simultaneously or isolate the
faulty static switch allowing the inverters to operate normally. In Eaton UPSs and SBM, the backfeed
protection comes built-in as standard and the UPS internal static bypass and the SBM system bypass
both utilize same fault detection methods to enable highest critical mission reliability.
Configuration of the static bypass switch for load support
The main difference between the distributed and centralized bypass systems is the static bypass switch
configuration for the UPS system. This affects the input and output switchgear configurations as well as
the bypass cabling requirements.
In the distributed bypass system, each UPS has its own bypass switch rated for the UPS power.
These bypass switches are connected in parallel.
In the centralized bypass system the system static switch in the SBM is rated to support the entire
UPS system load.
Input and output switchgear configuration
With distributed bypass system, there is a need for multiple UPS rated breakers on input switchgear and
switches on output switchgear to feed and isolate the UPSs with parallel static switches.
With the centralized bypass system, there is a need for an additional full system power breaker on input
switchgear and an additional full system power switch on output switchgear to feed and isolate the static
switch. Also, there is a need for multiple UPS rated breakers on input switchgear and switches on output
switchgear to feed and isolate UPS modules. The breakers and switches have been drawn in Figure 3.
However, the Maintenance Bypass Switch (MBS) does not differ in these two configurations, as it always
needs to be dimensioned for full system power. The MBS will be part of the input and output switchgear.
Also the breakers for UPS or IOM units on input and output switchgears will be the same in both
Typically a UPS has separate feeders for the rectifier and static switch, as shown in Figure 2. In a
redundant distributed bypass system, however, a common feeder for rectifier and static switch may be
used without compromising the system reliability and to save installation costs.
Table 1 Differences in input and output switchgear configuration.
Centralized Bypass
Distributed Bypass
Input Switchgear Breakers
Maintenance Bypass Switch
Static Switch
UPS units
1 pc Fully Rated
1 pc Fully Rated
1-5 pcs IOM Rated
1 pc Fully Rated
1-6 pcs UPS Rated
1-6 pcs UPS Rated
May 2012
Page 7 of 9
Output Switchgear Switches
Maintenance Bypass Switch
Static Switch
UPS units
Centralized Bypass
Distributed Bypass
1 (+1) pc Fully Rated
1 pc Fully Rated
1-5 pcs IOM Rated
1 (+1) pc Fully Rated
1-6 pcs UPS Rated
Bypass cabling requirements
When using a distributed bypass system, there is a need to consider that each UPS unit cabling
connected between the input switchgear to the UPS static bypass, and connected between the UPS
output to the output switchgear must have equal length and impedances. This is needed in order to have
equal load sharing between UPS static switches when the system is on bypass. If these impedances are
not equal, it will cause unbalanced load between the static switches and may lead to an overload of one
or several static switches. Figure 4 shows the basic principle of the wiring.
Required parallel system
wiring length must be equal
to ensure approximately
equal load sharing when in
bypass mode. For proper
operation, the following must
be apply:
1a = 2a = 3a = 4a … = Na
1b = 2b = 3b = 4b … = Nb
In order to save cabling
costs, the following is also
considered to be sufficient:
Any difference in wire length
will result in decreased
capacity of the UPS system
while on bypass. For
example, a 10% difference
between the longest and the
shortest wire lengths will
decrease bypass capacity by
10%. This may disable the
transfer to inverter.
Figure 4: Overview of the required parallel wiring principle and wiring length.
The cabling requirements are the main reason why distributed parallel systems are typically used for N+1
redundancy or up to maximum 90% capacity (as specified in Tier III classification). Dimensioning the
distributed parallel system for close to 100% capacity is not an optimal solution due to the difficulty to
match the bypass cabling to enable full bypass capacity. The centralized bypass system is more tolerant
for cabling impedances between UPSs.
However, the distributed bypass system will offer more flexibility for system dimensioning as it can be
expanded by similar rating UPSs in parallel to add redundancy or capacity. With centralized bypass
system, the power rating is limited to the power rating of the SBM module. On the other hand, the
distributed bypass system allows using different IOM ratings in parallel since the static switch
dimensioning is not an issue.
May 2012
Page 8 of 9
Centralized or Distributed? Choosing the parallel UPS system
Large organizations need tailored configurations that meet the set requirements for availability and
manageability. The choice of the configuration is also affected by the existing situation, whether the
customer is getting a new UPS system or is upgrading or changing an existing setup.
In the table below are listed some of the pros and cons that can be considered when evaluating the right
setup for a parallel UPS system.
Table 2 Pros and cons of the two paralleling UPS system configurations.
Central Bypass Paralleling
User controls the system from a central point.
Installation is not impacted by cable impedance
or length.
Less mechanical components and switching
Maintenance Bypass may be integrated into
the System Bypass Module (SBM).
Dependence on a single static switch.
Dependence on a single bypass breaker.
Dependence on a single motor operator if a
single static switch is momentary.
SBM unit adds cost and footprint.
Maintenance and service costs are slightly
Distributed Bypass Paralleling
No need for SBM: savings in footprint and upfront cost.
No dependence on a single static switch or
bypass breaker.
Tie cabinet doesn’t contain any intelligence: it
is simple, reliable, and vendor-independent.
Better scalability.
Multiple static switches must operate in unison.
For example, all of them must turn on and off at
the same instant.
Installation must include consideration of the
bypass wiring impedance (+/- 10%).
All UPSs must be identical. Not possible to
parallel units with mismatched ratings.
When comparing the cost of the systems, the distributed bypass system will be less expensive since SBM
unit is not needed. The cost difference is greatest with the smallest parallel systems. However, as the size
of the parallel system increases, the difference in overall system cost, including the switchgear and
installation, decreases.
Eaton solutions use unique Powerware Hot Sync paralleling technology in both distributed and
centralized parallel systems. Most paralleling technologies on the market can meet the needs for
synchronization, load-sharing and selective tripping by requiring control wiring and load share signals
between UPSs and the bypass cabinet. A failure in the communication will result in the parallel system
transferring to bypass, which is exactly what customers want to avoid when purchasing a parallel
redundant UPS.
With Eaton’s patented Powerware Hot Sync paralleling technology, the above-mentioned problems are
eliminated. Hot Sync technology is based on a load share algorithm that accomplishes the
synchronization and load sharing of multiple parallel UPSs independently of any communication between
the UPS units. In fact, using the Hot Sync load share algorithm, each inverter in the UPS system is able
to regulate its own output and to load share independently based in its own output measurement data.
Both systems can use Eaton’s Energy Advanced Architecture (EAA) to save energy and therefore the
operating costs of the IT system. The two complementary proprietary technologies, Variable Module
Management System (VMMS) and Energy Saver System (ESS), maximize UPS efficiency and
significantly reduce the energy consumption and environmental impact.
May 2012
Page 9 of 9
Concluding thoughts
Organizations obtain parallel UPS systems to prevent the loss of valuable electronic information, minimize
equipment downtime, and minimize the adverse effect of power outages on production equipment.
Already in the 1970s, extensive UPS installations were made in various military, industrial, commercial,
government and healthcare facilities.
In modern systems, the requirement for no single point of failure is essential. Using Eaton’s Hot Sync
technology, each UPS module operates independently without an external master controller or intermodule control wiring. When choosing Eaton’s parallel UPS system, you can be sure that your critical
load is protected by the most reliable system on the market.
Whether distributed or centralized bypass system is optimal for each installation is dependent on the
entire solution, and needs to be considered specifically for each UPS project. In fact, both paralleling
methods have their pros and cons, but most importantly both are solutions that guarantee reliable
protection for critical loads. Enhancing the reliability of the UPS system, and therefore the availability of
the datacenter itself, has two main reasons: the UPS system redundancy and the ability to perform
concurrent maintenance on any UPS or SBM while the system continues to provide conditioned, batterybacked power.
About Eaton
Eaton Corporation is a diversified power management company with more than 100 years of experience
providing energy-efficient solutions that help our customers effectively manage electrical, hydraulic and
mechanical power. With 2011 sales of $16.0 billion, Eaton is a global technology leader in electrical
components, systems and services for power quality, distribution and control; hydraulics components,
systems and services for industrial and mobile equipment; aerospace fuel, hydraulic and pneumatic
systems for commercial and military use; and truck and automotive drivetrain and powertrain systems for
performance, fuel economy and safety. Eaton has approximately 73,000 employees and sells products to
customers in more than 150 countries. For more information, visit
About the authors
Janne Paananen is Large Systems Group Manager in EMEA for Eaton Corporation, specializing in large
UPS system solutions for datacenters and special applications. He has more than 10 years of experience
with large three phase UPS products and has been working in after- and pre-sales organizations
providing support and in-depth product trainings for Eaton’s personnel and customers worldwide.
Sini Syvänen is a Product Applications Engineer in Three Phase UPS Products in EMEA for Eaton
Corporation. In this role, she provides pre-sales support and training to local sales teams on products,
technologies and applications. She also contributes to product development programs and product
evolution projects.
The staff at Eaton is committed to creating and maintaining powerful customer relationships built on a
foundation of excellence. Decades of experience in paralleling large UPS systems are incorporated in the
reliable product offering and customer-based tailoring.
Downloads on demand
Download Eaton white papers to learn more about technology topics or explain them to customers and
contacts. Free white papers can be found from our online library:
Read more about Eaton 9395 UPS and the System Bypass Module (SBM) from
Learn more about Eaton’s offering towards data centers at
May 2012
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF