Diesel Rotary UPS (DRUPS) vs.

Static
UPS: A Quantitative Comparison for
Cooling and IT Applications

White Paper 222

Revision 0

by Han Ming Kuang

Hng Yong Way
Sanjeet Sandhu

Executive summary
A number of published comparisons indicate
that DRUPS is a superior solution for data
center cooling and IT applications. The inaccu-
rate assumptions and mistakes found in these
comparisons are explained with support from
third party sources. In this paper, we present a
detailed quantitative comparison from a design
architecture-level perspective between low
voltage DRUPS and low voltage Static UPS.
We consider capital expenses, energy losses,
maintenance costs, footprint, and TCO using a
common 2N architecture. The analysis shows
the Static UPS is less expensive to purchase,
install, operate, and maintain while having a
slightly larger footprint.
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 2

Introduction Some published comparisons indicate that low voltage DRUPS is a superior solution
for data centre cooling and IT applications. When you consider the details of these
comparisons, however, it becomes apparent that they are often misleading and do
not provide a factual comparison. These comparisons often make the mistake of
using very old static UPSs with very poor efficiency curves and long battery auton-
omy, and the comparisons do not address how redundant-type design architectures
typically used in mission critical facilities today affect the analysis.

The following section lists common myths and explains them away.

Myth #1: “Higher system efficiency”

• It is often claimed that the efficiency level of DRUPSs is much better than the
best Static UPSs available in the market. However, these comparisons are
typically made against a UPS using the old SCR-based switching technology
with a very poor efficiency curve of 92% (highest value on curve) at best.
• Common Static UPSs available in the market today can achieve up to 97.6%
efficiency, which is on par or even better than the DRUPS. For example, the
efficiency level for a Schneider Electric Symmetra MW, even at a partial load
level of 25% is as high as 95% which is better than common DRUPS used to-
day.
• In these misleading comparisons, efficiency at partial loads is not shown.
Since UPSs in data centres almost always operate at far less than full load,
having partial load comparisons would be more realistic and useful.

Myth #2: DRUPSs are equally scalable

• While a DRUPS can be scaled by adding additional units, they are usually
oversized significantly to accommodate for future unknown load requirements.
This is mainly due to its lack of flexibility and limited range of capacity options.
In addition to being neither modular nor easily scalable, the upfront cost is also
higher than a comparative Static UPS system.
• Static UPSs are more scalable and flexible given their wider range of available
capacity ranges. A unit with an installed capacity much closer to the design IT
load of each phase of a project can be selected. Adopting this “right-sizing”
strategy, it enables data centre owners to “pay-as-you-grow” and the ability to
much more easily adapt to market changes.

Myth #3: 40 - 60% space savings

• “A Static UPS takes up much more space than a DRUPS” is not entirely a true
statement. And the percentage of additional space used by Static UPS vs.
DRUPS is also overly inflated in most cases.
• This space comparison is often based on 15 minutes (or longer) of battery
runtime for the Static UPS, which would result in a higher space requirement.
This is an unfair comparison since a DRUPS can only provide about 8 – 12 se-
conds of ride-through time. Also note that the space calculations are usually
based on VRLA type lead-acid batteries. Lithium-ion batteries use 50-80%
1
less space than VRLA .
• Depending on the application and design topology used, there are cases
where it is possible that the space required by DRUPSs is more than what is
required for the Static UPSs.

1
White Paper 231, FAQs for Using Lithium-ion Batteries with a UPS

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 3

Myth #4: No air conditioning is needed; therefore, it uses less

energy
• Yes, Static UPSs need air conditioning to maintain the indoor environmental
conditions needed to maximize battery life. However, the claim that using
DRUPS can, therefore, save more energy than a Static UPS is misleading.
This is because the comparison needs to be done in the context of the entire
system used, and not simply on one element of the system. Considering the
whole system, a DRUPS-based system uses MORE energy than a Static UPS
system due to its higher power losses under (realistic) partial load operating
conditions (< 50%).

Myth #5: “Thermal buffer” is highly complex and expensive

• It is common to use a thermal storage tank system to ensure continuous cool-
ing when a static UPS is used. This “thermal buffer” is actually quite simple.
Having a thermal buffer means having enough chilled water in the supply pip-
ing and/or in a storage tank to prevent the supply air temperatures from going
too high during an outage before the backup power supplies come online to
supply the cooling plants. In terms of cost, you should obviously consider the
entire system cost and not just some of the system components as some of
these suspect comparisons have done.

The analysis that follows in this paper addresses each of these common myths. It
assumes the DRUPs and Static UPS system is being used to back up the entire
data center with a 2N electrical infrastructure supporting both the IT and cooling
loads, although the static UPS architecture also uses a thermal buffer system to
ensure continuous cooling.

This paper offers a fact-based, “apples-to-apples”, and more realistic analysis that
quantifies the capital expense, total cost of ownership (TCO), maintenance costs,
and footprint to ensure better informed decision-making.

Methodology and A common 2N redundant architecture was used for this analysis (see sidebar). The
IT load density is assumed to be 6 kW/rack (average) for 500 IT racks. This yields a
assumptions total IT load of 3 MW.

In today’s data centres, a challenge that some face is an increase in rack power
density and higher density applications. In addition to this rise in density, ASHRAE
has expanded its recommendations for thermal limits in recent years encouraging
some to raise inlet air temperatures to increase economization hours. With increas-
ing densities and/or higher temperatures in the white space, an effective thermal
ride through strategy that keeps air temperatures from going too high in the event of
an outage becomes more critical. This strategy ensures the data centre and critical
plants are able to “ride through” the plant restart times and chiller loading time after
a power failure as the generators come on line, and vice versa, when the generator
is off line and grid power then returns.

The two most common strategies for maintaining cooling air during an outage
adopted by data centre designers and used in this analysis are:

• Thermal storage system (buffer tank of chilled water and static UPSs for fans/
pumps/ controls)
• Entire cooling system on DRUPSs

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 4

For the thermal storage system, the buffer tank is sized to offer 8 minutes of ride-
Design assumptions
through time before the chilled water supply temperature will begin to rise above its
summary set point. The Static UPSs provide backup to the chilled water pumps, chiller control
• 3MW IT load, 500 racks at system, CRACs and fan coil units. This ensures this equipment continues to deliver
6kW/rack average chilled water and cold air to the data centre as well as the M&E plant during the time
• Calculated cooling load - before backup power systems (e.g., generators) have started and begin to provide
945RT power.
• Water cooled chilled water
system at N+1
The battery runtime for the Static UPSs is 5 minutes which is typical for most data
• 8 - 15°C CHW temperature
• Canopy/ containerized type centers. Note that depending on the UPS model selected, it is possible to spec a
of generator shorter runtime to reduce capital expense and save space. This 5 minute runtime
• Static UPS: specification is more than sufficient to bridge the total duration required for the
o 2 paralleled 1600 kVA generator to start, ramp up to speed, and also to account for the timing delay for the
Symmetra MW UPSs at ATS switching.
2N for IT loads
o 400 kW Symmetra PX
In the DRUPS architecture, the entire cooling system and all IT equipment are
UPSs at 2N for cooling
loads supported by DRUPS. Remember that the Static UPS only needs to support some
• DRUPs: of the M&E plant to support the thermal ride-through process: chilled water pumps,
o Canopy/ containerized chiller controls, CRACs, and fan coil units (FCUs). It typically takes between 8 and
type of low voltage 12 seconds to go from a loss of utility power to completing the startup of the diesel
DRUPS generator within a DRUPS system.
o 2 paralleled 2000 kVA
DRUPs at 2N for IT
loads
See sidebar for all the design parameters and assumptions used in this analysis.
o 1670 kVA DRUPs at 2n Each subsection below details the comparison results on capital expense, UPS
for cooling load system efficiency, maintenance costs, footprint, and the 10 year TCO.
• Comparison and TCO
calculation focus only on The schematic diagrams for the 2N architectures are shown below. Figure 1
the electrical plant equip- illustrates the architecture where the entire facility’s IT and cooling loads are
ment supporting the whole
supported by DRUPS.
data centre
• Original calculations were
done in Singapore dollars
and were converted to USD
using a conversion rate of 1
Singapore $ to 0.7135
USD.

Figure 1
2N architecture using
DRUPS for IT and cooling
application (i.e., the full
data centre)

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 5

Figure 2 shows the architecture where a Static UPS system is supporting the IT and
cooling equipment (i.e., chilled water pumps, chiller controls, CRAC and FCU).
Buffer tanks are also present for storing chilled water to ensure there is sufficient
cooling before the alternative power source takes over in the event of an outage.

Figure 2
2N architecture using static
UPS for IT and cooling
application (i.e., full data
centre) with buffer tanks for
thermal ride through

The quantitative UPS system energy loss analysis

analysis Given the total IT load of 3 MW, the UPS utilization factor is 47% for both DRUPSs
and Static UPSs under normal operating conditions. For cooling loads, the UPS
utilization factor is 43% for the DRUPS and 44% for the Static UPS under normal
operating conditions as shown in Table 1.

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 6

Figure 3 illustrates the UPS efficiency curves based on one of the top DRUPS in the
market against Schneider Electric’s Symmetra MW Static UPS for the IT load and
Symmetra PX Static UPS for the cooling load.

100%

90%

System Efficiency
DRUPS 2000kVA
(One of the top DRUPS in market)
Symmetra MW 1600kVA
80%

70%

Figure 3 60%
UPS efficiency as a 0% 10% 25% 50% 75% 100%
function of % load Load

100%

90%
System Efficiency

DRUPS 1675kVA
(One of the top DRUPS in market)
Symmetra PX 400kVA
80%

70%

60%
0% 10% 25% 50% 75% 100%
Load

As you can see from the above figure, the efficiency of Static UPSs has improved
significantly over the years especially at low load conditions. The gap between
Static UPSs and DRUPSs widens substantially when the load drops below 50%,
which is very common in data centres today and is, indeed, required for a 2N
system.

In assuming a 2N redundant architecture and a realistic UPS utilization factor under

normal operation, an efficiency comparison (total system energy losses) between
DRUPSs and Static UPSs yields the following results:

The Static UPS has 68% less energy losses than the DRUPS. This has a major
impact on TCO which will be shown later in this paper. Table 1 shows the details.

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 7

UPS UPS UPS Part load Total UPS

UPS type
load capacity loading UPS eff. system loss
3000 kW 4 x 2000 kVA
DRUPS – IT (1600 kW each)
47% 93% 226 kW

DRUPS – 2 x 1670 kVA

Table 1 1160 kW 43% 91% 115 kW
Cooling (1340 kW each)
2N topology UPS system
energy loss analysis Overall UPS system energy loss for DRUPS 341 kW
results 4 x 1600 kVA
Static UPS 3000 kW
(1600 kW each) 47% 97% 93 kW
– IT Symmetra MW

Static UPS 2 x 400 kVA

353 kW (400 kW each) 44% 96% 15 kW
– Cooling Symmetra PX

Overall UPS system energy loss for STATIC UPS 108 kW

Footprint analysis
There is a common market perception that Static UPSs always take up much more
space than a DRUPS system due to the battery storage space requirement.

However, our more detailed analysis which was based on commonly used design
architecture and more realistic assumptions regarding the application showed that
the difference in footprint between the two systems is not nearly as significant
as commonly thought.

Let’s take a look at the below comparison. Apart from the actual equipment dimen-
sions, the electrical plant layout below also takes into consideration the space or
access necessary for maintenance.

DRUPS System – approximately 1,128 m2 (12,142 ft2)

Figure 4
DRUPS – 2N topology
overall plant layout (IT
and Cooling loads)

Static UPS System – approximately 1,176 m2 (12,658 ft2)

Figure 5
Static UPS with 5 min of
runtime and thermal
storage tanks– 2N
topology overall plant
layout (IT and partial
Cooling loads)

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 8

In this real world example, the Static UPS system requires only 4.3% more
space than the DRUPS system. Note that this calculation even includes the area
required for the thermal storage tanks.

Notes on TCO TCO analysis

calculations The total cost of ownership (TCO) analysis was based on a 10 year cycle, which
offers a sound basis for determining the value -- cost vs. ROI -- of an investment;
• Maintenance and cyclical
replacement cost included
more than just the capital cost alone. TCO provides a more complete picture of the
for main equipment - overall ownership cost.
generator, Static UPS and
DRUPS This TCO analysis includes capital expenditures, comprehensive maintenance, and
• Maintenance cost includes electricity costs over 10 years.
a 5% CAGR
• For Static UPS, battery For Static UPS, the capital cost For DRUPS, the capital cost includes
replacement is every 5
years, so there are 2
includes the following equipment: the following equipment:
replacements calculated in
the 10 year TCO analysis • 2250 kVA prime rated canopy/ • 2000 kVA canopy/ containerized
• For DRUPS, 1 engine containerized generator diesel rotary UPS
overhaul is included in the
• 1600 kVA prime rated canopy/ • 1675 kVA canopy/ containerized
10 year TCO
• Potential rental gains as a
containerised generator diesel rotary UPS
result of space savings is • Main switchboards /w automatic • Main switchboards
excluded transfer switch
• Decommissioning and • Main distribution boards
depreciation costs were not • Main distribution boards • Power Distribution Unit (PDU)
considered • 1600 kW Symmetra MW static UPS
To make a simple “apples-to- /w 5 min of battery runtime
apples” comparison, a discount • 400kW Symmetra PX static UPS
rate was not applied to create
“net present values” for future c/w 6 min battery
OPEX-related costs. The focus • N+1 CRAC for UPS and battery
of this analysis is on the
relative difference between the room
two architectures. 3
• Thermal buffer tanks 2 x 28.6 m
• Power Distribution Unit (PDU) with
isolation transformer (2% isolation
transformer loss taken into ac-
count)

Table 2 Capital Maint. & cyclical Electricity TCO for 10

UPS type
10 yr TCO in USD for expense replacement cost years
DRUPS & Static UPS
DRUPS $8,374,978 $3,320,600 $70,306,263 $82,001,841
excluding potential
rental gains Static UPS $6,047,104 $2,484,262 $67,578,380 $76,109,746

TCO Savings (Static UPS) excluding potential rental gains over 10 years –
$5,892,095 (approx 7% savings).

Based on this example, the Static UPS system costs less to purchase and
install, maintain, and operate.

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 9

Full Static UPS with Static UPS %

Results
DRUPS Buffer Tanks Savings
Overall UPS system energy losses 341 kW 108 kW 68.3%
2 2
Required plant floor area space 1128 m 1176 m - 4.3%
Table 3
2N architecture overall a) CapEx $8,374,978 $6,047,104 27.8%
comparison summary
b) Maintenance & cyclical
$3,320,600 $2,484,262 25.2%
replacement over 10 years*
c) Electricity cost over 10 years $70,306,263 $67,578,380 3.9%
10 yr TCO (a+b+c) $82,001,841 $76,109,746 7.2%
* It should be noted that for DRUPS, all cyclical replacement/ overhaul and maintenance has to be done by the
authorized dealers only. Thus, it would be more of a challenge to get a competitive quote for replacement,
unlike battery for Static UPS which you can opt for any vendors as long as the technical specs fit.

Table 3 shows that a full 2N Static UPS system’s electrical infrastructure supporting
both the IT and cooling loads of a data centre has a 10 year TCO savings of
$5,892,095 (7.2%) over a full 2N DRUPS electrical infrastructure, but at a small
2
penalty of needing 48 m additional floor space.

So if the data center in our example analysis was a collocation data center, how
might the potential rental gains in the DRUPs scenario affect the overall TCO? We
assumed a rough rental rate estimate of $1285/rack per month. Note that this is
gross revenue and not profit.

Overall 10yr Rental Total TCO including Static UPS

UPS type
Table 4 10yr TCO Gains Rental Gains % Savings
TCO for 10 years DRUPS $82,001,841 ($2,959,400) $79,042,441
including rental gains
Static UPS $76,109,746 $76,109,746 3.7%

For a full 2N Static UPS electrical infrastructure supporting both the IT and cooling
loads of a data centre, the overall 10 year TCO savings using a Static UPS
system vs. a DRUPs system (including money earned through increased rental
space in the case of the DRUPs-based architecture) is $2,932,695 (3.7%
savings).

Consideration of common plants for DRUPS in 2N architecture for

full facility serving both IT and cooling
In designing a common DRUPS plant, i.e. parallel multiple DRUPS, for Data Centre
that requires higher electrical power, consideration has to put into understanding the
limitation of low voltage system. The highest rating of low voltage equipment is
6,300A (which yields approximately 4.36 MVA at 400V) and, therefore, you cannot
have more than two 2 MVA DRUPS connected in parallel on a common electrical
bus.

A way to overcome this power limitation of the low voltage system is to move to a
medium voltage (MV) distribution solution. This topic is the subject of another
research project at Schneider Electric and its analysis will be released in another
white paper in the future.

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 10

Additional In the Appendix, several other analyses are shown in detail to compare DRUPSs
and Static UPSs under different architectures and for specific applications. These
comparisons additional comparisons include:

• 2N architecture supporting IT load only

• Tri-redundant architecture supporting IT load only
• 2N architecture supporting cooling load only (to achieve continuous cooling)
• Tri-redundant architecture supporting continuous cooling load only (to achieve
continuous cooling)

The results can be summarized as follows:

2N architecture supporting IT load only

IT on IT on Static Static UPS %
Results
DRUPS UPS Savings
226 kW
Overall UPS system energy losses 93 kW 58.8%
2 2
Required plant floor area space 680 m 816 m - 20.0%
Table 5
IT load only – 2N a) CapEx $6,004,816 $4,530,012 24.6%
architecture summary
b) Maintenance & cyclical
$2,222,966 $1,877,818 15.5%
replacement over 10 years*
c) Electricity cost over 10 years $50,351,181 $49,748,051 1.2%

10 yr TCO (a+b+c) $58,578,963 $56,155,881 4.1%

NOTE: IF you take the potential additional rental gains into account that the
extra space provides, the DRUPS system 10 year TCO is 10.6% better than the
Static UPS system TCO.

Tri-redundant architecture supporting IT load only

IT on IT on Static Static UPS %
Results
DRUPS UPS Savings
Overall UPS system energy losses 191kW 83 kW 56.5%
Table 6
IT load only – Tri-redundant Required plant floor area space 568 m2 574 m2 - 1.05%
architecture summary
a) CapEx $4,477,926 $3,219,312 28.1%
b) Maintenance & cyclical
$1,667,224 $1,408,363 15.5%
replacement over 10 years*
c) Electricity cost over 10 years $49,869,095 $49,596,000 0.5%
10 yr TCO (a+b+c) $56,014,245 $54,223,675 3.2%

NOTE: With rental gains taken into consideration, the overall TCO savings is
reduced slightly from 3.2% to 2.6%.

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 11

2N architecture supporting cooling load only

Cooling on Cooling on Static UPS %
Results
DRUPS Static UPS Savings
Overall UPS system energy losses 111 kW 15 kW 86.5%
Table 7 Required plant floor area space 448 m2 360 m2 19.6%
Cooling load only –
2N architecture summary a) CapEx $2,369,105 $1,516,330 36.0%
b) Maintenance & cyclical
$1,097,212 $606,131 44.8%
replacement over 10 years*
c) Electricity cost over 10 years $19,230,432 $17,821,814 7.3%
10 yr TCO (a+b+c) $22,696,749 $19,944,275 12.1%

2
NOTE: If rental gains are taken into consideration, the 88 m space savings
will increase overall 10 year TCO savings from 12.1% to 56.3% in favor of
Static UPSs.

Tri-redundant architecture for cooling load only

Cooling on Cooling on Static UPS %
Results
DRUPS Static UPS Savings
Overall UPS system energy losses 169 kW 29 kW 82.8%

Table 8 Required plant floor area space 605 m2 536 m2 11.4%

Cooling load only –
tri-redundant architecture a) CapEx $3,584,160 $2,236,501 37.6%
summary b) Maintenance & cyclical
$1,645,819 $909,197 44.8%
replacement over 10 years*
c) Electricity cost over 10 years $37,633,747 $35,643,627 5.3%

10 yr TCO (a+b+c) $42,863,726 $38,789,325 9.5%

2
NOTE: If rental gains are taken into consideration, the 69 m space savings will
increase overall 10 year TCO savings from 9.5% to 24.7% in favor of Static
UPSs.

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 12

Conclusion Using the common 2N architecture described above as an example, this paper
demonstrated how it is important to consider the entire system and facility when
comparing capital expenses, energy costs, maintenance costs, footprint, and overall
TCO between one data centre backup system type and another. The analysis
described in this paper clearly showed that a data centre physical infrastructure
architecture employing a low voltage Static UPS system outperformed the same
architecture using a low voltage DRUPS on all of these comparison points except
footprint.

About the authors

Han Ming Kuang is the Data Centre Consulting and Solutions Head for DC Centre of Excellence
at Schneider Electric. He has over 16 years of experience in the field of consulting engineering and
project management including extensive experience in the concept and detailed design for various
types of building development in particular mission critical facilities.

Hng Yong Way is the Principal Engineer for DC Centre of Excellence at Schneider Electric. He has
over 13 years of experience in the area of consulting engineering, project and construction
management including extensive experience in the concept and detailed design in particular
mission critical facilities, institutions, residential and commercial developments.

Sanjeet Sandhu is Vice President for the DC Centre of Excellence at Schneider Electric. Sanjeet
heads a team that supports the strategy and delivery of the full lifecycle of Data Centre Services
that include consulting, design, project execution, assessments, software, and managed services
(including critical facility operations). Sanjeet is a registered Professional Engineer in Singapore.

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 13

Comparing UPS System Design Configurations

White Paper 75
Resources
Browse all
white papers
whitepapers.apc.com

Power Sizing Calculator

TradeOff Tool 1

Energy Allocation Calculator

TradeOff Tool 2

Energy Efficiency Calculator

TradeOff Tool 6

Browse all
TradeOff Tools™
tools.apc.com

Contact us
© 2016 Schneider Electric. All rights reserved.

For feedback and comments about the content of this white paper:
Data Center Science Center
[email protected]

If you are a customer and have questions specific to your data center project:
Contact your Schneider Electric representative at
www.apc.com/support/contact/index.cfm

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 14

Appendix

IT load only The IT application comparison looks at the commonly adopted redundancy architec-
tures, i.e. 2N topology. The IT load density is assumed to be 6kW/rack for 500 IT
comparison – racks with a final IT load of 3MW.
2N architecture
Each subsection below details the results of the capital expense, UPS system
energy loss, maintenance costs, footprint, and total cost of ownership (TCO) for the
Design Parameters
2N topology. Schematic diagrams of the two 2N architectures for IT application are
and Assumptions shown above in Figure 1 for DRUPS and Figure 2 for Static UPS in the earlier
• 3MW IT load, 500 racks at section of this paper.
6kW/rack
• Canopy type of generator The ride-through time for DRUPS flywheel is 8 – 12 seconds and the battery runtime
• Canopy type of DRUPS for the Static UPS is a typical 5 minutes, though it can be further reduced depending
• Comparison and TCO on the selected UPS model.
calculation focus only on
the electrical plant equip-
ment supporting the IT See sidebar for all the design parameters and assumptions used in this analysis.
infrastructure only - "varia-
ble" UPS system energy loss analysis
• M&E plant supporting
lighting, general purpose The UPS utilization factor is 47% for both DRUPS and Static UPS under normal
power, cooling load, and operating conditions based on the total IT load of 3000kW with 2N topology. An
other main equipment, e.g. energy loss comparison based on 2N topology shows the difference of energy loss
chiller, CRAC, etc. are between a DRUPS and a Static UPS.
excluded from the case

UPS UPS UPS Part load Total UPS

UPS type
load capacity loading UPS eff. system loss
Table A1
IT application – 2N DRUPS 3000 kW 4 x 2000 kVA 47% 93% 226 kW
architecture UPS system (1600 kW each)
energy loss analysis 4 x 1600 kVA
results Static UPS 3000 kW (1600 kW each) 47% 97% 93 kW
Symmetra MW

DRUPS System – approx. 680m2 (7,319 ft2) Static UPS System – approx. 816m2 (8,783 ft2)

Figure A1
IT application – 2N
architecture overall plant
layout

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 15

UPS System TCO analysis

Notes on TCO For Static UPS, the capital cost includes For DRUPS, the capital cost includes the
Calculations the following equipment: following equipment:
• Maintenance and cyclical
replacement cost consid- • 2250 kVA prime rated canopy • 2000 kVA canopy diesel rotary
ered for main equipment - generator UPS
generator, Static UPS and • Main switchboards /w automatic • Main switchboards
DRUPS transfer switch
• Maintenance cost includes • Main distribution boards
5% CAGR • Main distribution boards • PDU
• For Static UPS, battery • 1600kW Symmetra MW static UPS
replacement every 5 years.
/w 5 min of battery runtime
Total 2 replacements are
included in the 10 years • N+1 CRAC for UPS and battery
TCO room
• For DRUPS, 1 overhaul is
included in the 10 years • PDU with isolation transformer (2%
TCO isolation transformer loss taken in-
to account)

Capital Main. & cyclical Electricity TCO for 10

UPS type
Table A2 expense replacement cost years
IT application – 2N
topology TCO results DRUPS $6,004,816 $2,222,966 $50,351,181 $58,578,963

Static UPS $4,530,012 $1,877,818 $49,748,051 $56,155,881

TCO Savings (Static UPS) over 10 years – $2,423,082 (approximately 4% savings)

From the IT application 2N architecture analysis, it demonstrates that Static UPSs

provide better overall TCO compared to DRUPSs. The negative point in the 2N
architecture for Static UPS is the additional 20% space required.

However, if take rental gains from the additional 20% plant space into consideration,
the 2N DRUPSs have a TCO savings of 10.6% in 10 years over the 2N Static UPSs.

In this case, for a sole IT load application the 2N DRUPSs perform better than the
2N Static UPSs.

IT load only In this section, a tri-redundant design with the same assumptions for the IT load is
analyzed. The same set of design parameters and assumptions were used here as
comparison – in the 2N architecture. Figure A2 and A3 show the basic schematic diagram of the
Tri-redundant architecture
architecture

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 16

Figure A2
DRUPS, tri-redundant
architecture diagram for
IT application

Figure A3
Static UPS, tri-redundant
architecture diagram for
IT application

UPS system energy loss analysis

In reference to Figures A2 and A3, the UPS utilization factor is 63% for both
DRUPSs and Static UPSs under normal operating conditions based on the total IT
load of 3000 kW in a tri-redundant architecture.

An energy loss comparison based on the tri-redundant architecture shows the

difference of energy loss between the DRUPSs and the Static UPSs.

UPS UPS UPS Part load Total UPS

UPS type
Table A3 load capacity loading UPS eff. system loss
IT application – Tri- 3 x 2000 kVA
redundant architecture DRUPS 3000 kW (1600 kW each)
63% 94% 191 kW
energy loss analysis
results 3 x 1600 kVA
Static UPS 3000 kW (1600 kW each) 63% 97.3% 83 kW
Symmetra MW

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 17

Footprint analysis
For the tri-redundant architecture, the footprint difference is marginal. The Static
UPSs take up 1% more space than DRUPSs.

DRUPS System – approx. 568m2 (6,114 ft2) Static UPS System – approx. 574m2 (6,178 ft2)

Figure A4
IT application – Tri-
redundant architecture
overall plant layout

TCO analysis
TCO analysis based on a 10 year life cycle was done using the same assumptions
and parameters.

Capital Maint. & cyclical Electricity TCO for 10

UPS type
Table A4 expense replacement cost years
IT application – Tri-
redundant architecture DRUPS $4,477,926 $1,667,224 $49,869,095 $56,014,245
TCO results Static UPS $3,219,312 $1,408,363 $49,596,000 $54,223,675

TCO Savings (Static UPS) over 10 years – $1,790,570 (approx 3% savings)

From the analysis, it demonstrates that Static UPSs provide better values and better
overall TCO compared to DRUPSs, with a marginally higher space requirement of
an additional 1.05%. The overall TCO savings for 10 years is 3.2%, approximately
$1.79 million.

With rental gains into consideration, the overall TCO savings reduced slightly from
3.2% to 2.6%.

Cooling load only Each subsection below details the results of the capital expense, UPS system
energy loss, maintenance costs, footprint, and total cost of ownership (TCO) for the
comparison – 2N topology. Schematic diagrams of the two 2N topologies for IT application are
2N architecture shown in Figure 1 for DRUPS and Figure 2 for Static UPS in the earlier section of
this paper.

UPS system energy loss analysis

The UPS utilization factor is 42% for the DRUPS and 44% for the Static UPS under
normal operating conditions based on the total cooling load of 945RT. This is the
total cooling load required for a single hall of 3MW IT load and the associated M&E
plant room.

An energy loss comparison based on 2N architecture shows the difference of energy

loss between a DRUPS and a Static UPS.

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 18

UPS UPS UPS Part load Total UPS

UPS type
load capacity loading UPS eff. system loss
Table A5
Cooling load only – 2N DRUPS 1120 kW 2 x 1675 kVA 42% 91% 111 kW
topology UPS system (1340 kW each)
energy loss analysis 2 x 400 kVA
results Static UPS 353 kW (400 kW each) 44% 96% 15 kW
Symmetra PX

Footprint analysis
Based on a 2N architecture, the Static UPSs are more space efficient. It takes
19.6% less space compared with DRUPSs.

DRUPS System – approx. 448m2 (4,822 ft2) Static UPS System – approx. 360m2 (3,875 ft2)

Figure A5
Cooling load only – 2N
architecture overall plant
layout

Design Parameters
and Assumptions
TCO analysis
• Calculated cooling load -
945RT For Static UPS, the capital cost of the For DRUPS, the capital cost of the below
• Water cooled chilled water below main equipment are included main equipment are included
system in N+1,
• 8 - 15°C CHW temperature
• 1600kVA prime rated canopy gen- • 1675kVA canopy diesel rotary UPS
• Canopy type of generator
erator • Main switchboards
• Canopy type of DRUPS
• Comparison and TCO • Main switchboards c/w automatic • Main distribution boards
calculation focus only on transfer switch
the electrical plant equip-
ment supporting the cooling • Main distribution boards
infrastructure only - "varia- • 400kW Symmetra PX static UPS
ble" c/w 6 min battery
• M&E plant supporting IT
load, lighting, general • N+1 FCU for UPS and battery
purpose power and other room
main equipment, e.g. IT
• Buffer tanks 2 x 28.6 m3
UPS, etc. are excluded
from the case study for
simplicity - "constant"

Capital Maint. & cyclical Electricity TCO for 10

UPS type
expense replacement cost years
Table A6
Cooling load only – 2N DRUPS $2,369,105 $1,097,212 $19,230,432 $22,696,749
architecture TCO results
Static UPS $1,516,330 $606,131 $17,821,814 $19,944,275

TCO Savings (Static UPS) over 10 years – $2,752,474 (approx 12% savings)

The cooling application 2N architecture analysis demonstrates that a Static UPS

system provides better value and better overall TCO compared to a DRUPS system.

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 19
Notes on TCO
Calculations The TCO savings for 10 years is very substantial at 12.1% without consideration of
rental gains from the space savings.
• Maintenance and cyclical
replacement cost only 2
considered for main If rental gains are taken in consideration, the 88m space savings will increase
equipment - generator, overall 10 years TCO savings from 12.1% to 56.3% in favor of Static UPSs.
Static UPS and DRUPS
• Maintenance cost includes
5% CAGR
• For Static UPS, battery
replacement every 5 years.
Total 2 replacements are
included in the 10 years
TCO
• For DRUPS, 1 overhaul is
included in the 10 year
TCO

IT load only Same set of design parameters and assumptions were used here in the tri-
redundant architecture where the battery autonomy for the Static UPSs is sized to
comparison – offer 6 minutes to match the short ride-through time of a DRUPSs’ flywheel (8 – 12
Tri-redundant seconds) for a more accurate, “apples to apples” comparison.
architecture The two tri-redundant architectures for the cooling application are shown in the
diagram below; one used a DRUPSs (Figure A6) and the other used a Static UPSs
Design parameters (Figure A7).
and assumptions
Same set of parameters and
assumptions as per the 2N
topology with the exception of
the below:

• 2 halls of 3MW IT load, 500

racks at 6kW/rack
• Calculated cooling load - 2
x 945RT

Figure A6
DRUPS – Tri-redundant
architecture diagram for
cooling application

Figure A7
Static UPS – Tri-
redundant architecture
diagram for cooling
application

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications
 Schneider Electric – Data Center Science Center White Paper 222 Rev 0 20

Cooling application UPS system energy loss analysis

Reference to Figure A6 and Figure A7, the UPS utilization factor is 55% for
DRUPS and 59% for Static UPSs under normal operating conditions based on the
cooling load of 2 x 945RT.

An energy loss comparison based on tri-redundant architecture shows the difference

of energy loss between the DRUPSs and Static UPSs.

UPS UPS UPS Part load Total UPS

UPS type
Table A7 load capacity loading UPS eff. system loss
Cooling application – Tri- 3 x 1675 kVA
redundant architecture DRUPS 2240 kW (1340 kW each)
55% 93% 169 kW
UPS system energy loss
3 x 400 kVA
analysis results Static UPS 706 kW (400 kW each) 59% 96% 29 kW
Symmetra PX

Footprint analysis
In a tri-redundant architecture, the Static UPS system is more space efficient. The
area consumed by the equipment is11.4% less than a comparable DRUPS system.

DRUPS System – approx. 605m2 (6,512 ft2) Static UPS System – approx. 536m2 (5,769 ft2)

Figure A8
Cooling application – tri-
redundant architecture
overall plant layout

TCO analysis
TCO analysis based on a 10 years cycle was done using the same assumption and
parameters.

Capital Maint. & cyclical Electricity TCO for 10

Table A8 UPS type
expense replacement cost years
Cooling application – tri-
redundant architecture DRUPS $3,584,160 $1,645,819 $37,633,747 $42,863,726
capital expense results
Static UPS $2,236,501 $909,197 $35,643,627 $38,789,325

TCO Savings (Static UPS) over 10 years – $4,074,401 (approx 10% savings)

This demonstrates that the Static UPS system provides better value and better
overall TCO compared to a DRUPS system. The TCO savings for 10 years is very
substantial at 9.5%.
2
If rental gains are taken in consideration, the 69m space savings will increase
overall 10 years TCO savings from 9.5% to 24.7% in favor of Static UPSs.

Diesel Rotary UPS (DRUPS) vs. Static UPS:

A Quantitative Comparison for Cooling and IT Applications