APC APPLICATION NOTE #126

Selection Procedure for InRow®

Chilled Water Products
By Greg Uhrhan, P.E.
and Scott Buell

Abstract
Proper selection of InRow chilled water cooling systems requires an engineer or technical salesperson to assess a number of
variables to ensure customer satisfaction. These variables include but are not limited to the room layout, load profile of the
servers and IT equipment, user’s desired cold aisle temperature, and available or desired chilled water temperatures.

Introduction
With new APC InRow products, a number of variables are needed to be calculated. Among those variables are the entering
water temperature, expected water side temperature difference, expected hot aisle temperature, chiller capacity, and Cooling
Distribution Units in use. All of these factors must also be factored in with the overall data center layout as this will also have
significant effect on how the cooling units perform. For information about best practice of data center layout, refer to
Application Note 92, Best Practices for Designing Data Centers with the InRow RC.

Calculation of Hot Aisle Temperature

The biggest and most complex variable is the hot aisle temperature calculation. Generally, servers are rated for typical power
and airflow draw. This makes it simple to calculate a temperature difference for the air through the server.

3162
ΔTair =
(CFM / kW )
In addition to server loads, InRow Power must be considered. Inefficiencies inherent in the transformers, rectifiers, power
distribution units, and UPS units lead to an increased heat load that must be accounted for and neutralized by the cooling
units.

Table 1 – Symmetra PX Thermal Characteristics

Product Heat Output Air Flow
(kW) (SCFM)
[m^3/hr]
Symmetra PX40 3.5 180 [306]

Symmetra PX80 7.0 420 [714]

1
____________________________________________________________________________________________________

©2007 American Power Conversion. All rights reserved. No part of this publication may be used, reproduced, photocopied, transmitted,
or stored in any retrieval system of any nature, without the written permission of the copyright owner. www.apc.com Rev 2006-1
 APC APPLICATION NOTE

Then, for a given cell (probably a set of rows forming a common hot aisle), the hot aisle temperature can be taken as the users

desired cold aisle temp plus the average weighted ΔT . While there is some error associated with using this on a larger scale,

it can be sized down to smaller groups of equipment to provide a good estimate of peak local hot aisle temperatures.

ΔTaverage =
∑ (CFM × ΔT air )
∑ CFM
⎡
⎢ΔTaverage =
∑ (CMH × ΔT air )⎤
⎥
⎢⎣ ∑ CMH ⎥⎦

or its analog

ΔTaverage = 3162 ×
∑ kW
∑ CFM
⎡
⎢ΔTaverage = 1861 ×
∑ kW ⎤⎥
⎢⎣ ∑ CMH ⎥⎦
The Cooling Correction Multiplier (CCM) found in Table 2, accounts for extra room temperature air that is mixed with the hot air
leaving the servers before being conditioned in the air-conditioner.

Table 2 – Cooling Correction Multiplier

Configuration Multiplier,
CCM
Open Aisle .70

HACS .95

RACS .95

The hot aisle temperature can then be calculated as

Thot = ΔTaverage × CCM + Tcold

This Thot is then used as the entering air temp for the air conditioner.

2
___________________________________________________________________________________________________

©2007 American Power Conversion. All rights reserved. No part of this publication may be used, reproduced, photocopied, transmitted,
or stored in any retrieval system of any nature, without the written permission of the copyright owner. www.apc.com Rev 2006-1
 APC APPLICATION NOTE

Water Variables
The water variables are almost entirely dependent on user preferences, or in the case that the user is connecting to a building
chilled water system, building system design and operation. The ACRC and ACRP products can tolerate a wide variety of
entering water temperatures, currently cataloged as 42°F to 60°F [5.6°C to 15.6°C]. If building chilled water is available in
sufficient quantity, it can be used. In that scenario, the entering water temperature would be the chilled water system’s low
temperature main. A facility engineer would need to be consulted to determine what water side temperature rise is acceptable
for smooth and efficient operation of the building system. Also, the facility engineer would need to be consulted to ensure that
the building chilled water system does not go into a “setback” mode during non-occupied hours, as that would adversely effect
the data center operation. Available pump head will also determine the amount of water flow that can be obtained. The
maximum available flow for an ACRC can be back solved by the pressure loss equation:

(.0762 × V& 2 + .1669 × V& )

ΔP =
2.31
[ ΔP = (.9053 × V& 2 + 31.43 × V& ) ]
where ΔP is in psi [Pa] and V& is in Gallons per minute [L/s]. The maximum available flow for an ACRP500 series can be
back solved by the pressure loss equation:

(.0306 × V& 2 + .0537 × V& )

ΔP =
2.31
[ΔP = (.3635 × V& 2 + 10.112 × V& ) ]
where ΔP is in psi [Pa] and V& is in Gallons per minute [L/s].

For dedicated chiller loops that are designed to meet the datacenter’s cooling load only, one should consult the published
Technical Data Manual for performance specifications, water flow rates, and pressure drops. If the user is setting up a
dedicated loop to their datacenter, then nearly any temperature is attainable in the given range. APC recommends use of 45°F
[7.2°C] entering water temperature and a 10°F [5.6°C] temperature rise for optimal performance and minimal latent capacity
waste. For increased electrical efficiency, entering water temperature leaving the chiller can be increased, however the user
must consider the capacity loss of the RC/RP and adjust the datacenter design accordingly if high enough densities are
present. Pumps should be sized to ensure that the pressure drop of not only the air conditioner are included, but also any
building piping, CDU pressure drop (APC Skus ACFD12-T and ACFD12-B), or pex piping pressure loss. CDU pressure loss
is given by :

(.0498 × V& 2 + .0313 × V& )

ΔP =
2.31
[ΔP = (.5916 × V& 2 + 5.8940 × V& ) ]
3
___________________________________________________________________________________________________

©2007 American Power Conversion. All rights reserved. No part of this publication may be used, reproduced, photocopied, transmitted,
or stored in any retrieval system of any nature, without the written permission of the copyright owner. www.apc.com Rev 2006-1
 APC APPLICATION NOTE

And pressure loss through the pex piping is given by:

ΔP (.0542 × V& 2 + .1705 × V& )

=
Leq 2.31

⎡ ΔP ⎤
⎢ = (2.1249 × V& 2 + 105.95 × V& )⎥
⎣⎢ Leq ⎦⎥
There also can be flexible hoses used for vibration isolation and ease of installation and placement. Table 3 shows the
pressure drop induced by the stainless hoses.

Table 3 – Stainless Steel Hose Cv

Length D=1.00” D=1.25”
[DN25] [DN30]
36”[914mm] 35 57

72”[1829mm] 24 38

Glycols
Frequently, in areas where freeze protection is required or desired for safety of system integrity, glycol solutions are used to
depress the freezing point to prevent pipe freezing. Either ethylene or propylene glycol can be used at any proportion,
however, ASHRAE recommends a maximum of 30% ethylene glycol or 35% propylene glycol due to the decrease heat
carrying capacity and higher viscosity of glycol mixtures (ASHRAE Fundamentals Handbook 2001 p. 21.5). Table 4 lists the
derating factors used for glycols.

Table 4 – Glycol Adjustment Multiplier

Percentage Capacity Capacity Pressure Drop Pressure Drop
Glycol (%) Multiplier Multiplier Multiplier Multiplier
Ethylene Propylene Ethylene Propylene
Glycol Glycol Glycol Glycol
0 1.00 1.00 1.00 1.00

10 .97 .96 1.04 1.09

20 .93 .90 1.13 1.20

30 .88 .82 1.21 1.35

40 .81 .77 1.31 1.52

4
___________________________________________________________________________________________________

©2007 American Power Conversion. All rights reserved. No part of this publication may be used, reproduced, photocopied, transmitted,
or stored in any retrieval system of any nature, without the written permission of the copyright owner. www.apc.com Rev 2006-1
 APC APPLICATION NOTE

The above factors assume a constant flow rate as rated for the given conditions in the technical data manual.

Altitude
In properly estimating the capacity of a chilled water unit, many locations may require use of an altitude correction factor. All
values found in APC Technical Data manuals are based on sea level performance. Table 5 gives the Altitude Correction
Factors. Note that while APC can adequately predict cooling equipment performance, we cannot predict how a user’s servers
will perform in high altitude applications. Although, it is generally a fair assumption that as altitude increases, the server outlet
temperature, for a given server, will increase and as a result the hot aisle temperature will slightly increase.

Table 5 –Altitude Correction Factors (ACF)

Altitude (ft [m]) ACF (-)
0[0] 1.00

1000 [305] .98

2000 [610] .96

3000 [914] .93

4000 [1219] .91

5000 [1524] .88

6000 [1829] .86

7000 [2134] .85

8000 [2438] .83

Chilled Water InRow CRACs

ACRC 100 series
The ACRC is a sensible-only half rack wide InRow cooling product. It has a nominal capacity of 17 kW at 85°F [29.4°C]
Entering Air Temperature and 45°F [7.2°C] EWT with a 10° F [5.6°C] temperature rise on the water side and running its full
capacity, 2900 SCFM [4900 CMH], of air. The actual capacity can be modulated from 0 to 30 kW depending on actual
conditions. The unit is fully capable of modulating both airflow and water flow rate to maintain leaving air temperature and cold
aisle temperatures within a user defined range.

5
___________________________________________________________________________________________________

©2007 American Power Conversion. All rights reserved. No part of this publication may be used, reproduced, photocopied, transmitted,
or stored in any retrieval system of any nature, without the written permission of the copyright owner. www.apc.com Rev 2006-1
 APC APPLICATION NOTE

ACRP 500 series

The ACRP is a precision, full rack wide InRow cooling product. It has a nominal capacity of 37 kW at 85°F [29.4°C] EAT and
45°F [7.2°C] EWT with a 10°F [5.6°C] temperature rise on the water side and running its full capacity, 6900 SCFM [11,700
CMH], of air. The actual capacity can be modulated from 0 to 70 kW depending on actual conditions. The unit is fully capable
of modulating both airflow and water flow rate to maintain leaving air temperature and cold aisle temperatures within a user
defined range. Furthermore, the unit is capable of reheat and humidification for precision room control.

ACRC 500 series

The ACRC500 series is alike is capacities to the ACRP, however has no humidity control or reheat ability and therefore is not
precision air conditioning.

Checking a Solution
Three checks must be run to make sure that the system will satisfy all needs and redundancy requirements: a capacity check,
an airflow check, and a supply air temperature check. If any of these fail, application engineering must determine what must be
done to make the system operate as planned or determine whether the other variables are enough to compensate.
The capacity check verifies that the unit performance is adequate to meet the user requirements. If a user requires N
redundancy on a row basis, the engineer must ensure each row has more cooling than power input. If the user requires N+1,
one more unit than is required to meet the cooling needs must be installed. In mixed mode applications, it must be assumed
that the redundant unit is always the highest capacity unit.

The airflow check ensures that the cooling units are drawing at least as much air as required to maintain the calculated
entering air temperature. Therefore, for N redundancy

∑ CRAC
CFM * CCM > ∑Servers CFM
If this is not true, then additional cooling units must be added to ensure all air is captured by the cooling units and hot air is not
recycled into the servers. To define N+1 or greater redundancy, particularly in mixed mode applications, assume a loss of the
highest flow unit and ensure,

∑ CRAC
CFM * CCM − CFM highestflowunit > ∑Servers CFM

Finally, an inlet air temperature check should be run. This ensures that at the maximum air flow rate, the cooling units have
adequate cooling to maintain a sufficient inlet air temperature to the servers. To perform this check,

3415 × ∑ kW
Thot − = Toutlet
1.08 × ∑ CFM

⎡ 3415 × ∑ kW ⎤
⎢Thot − = Toutlet ⎥
⎣⎢ 1.84 × ∑ CMH ⎦⎥

6
___________________________________________________________________________________________________

©2007 American Power Conversion. All rights reserved. No part of this publication may be used, reproduced, photocopied, transmitted,
or stored in any retrieval system of any nature, without the written permission of the copyright owner. www.apc.com Rev 2006-1
 APC APPLICATION NOTE

Verify that Toutlet is less than the cold aisle setpoint.

Sample Procedure
Take Figure 1 as our example data center located in Denver. Each rack is loaded to 5 kW and each have an associated
minimum airflow per kW. User assigns the cold aisle temperature to be 70°F. Four ACRC 100 units are placed in the space
with open aisle configuration, as shown, and supplied with 45°F ethylene glycol at 20% solution and a desired water
temperature rise of 10°F. Determine the cooling system characteristics and redundancy.

Total rack airflow is shown in Table 6.

Table 6 – Rack Power Table

Power Draw Airflow
(kW) CFM/KW (CFM)
Row
A Rack 1 5 160 800
Rack 2 5 160 800
Rack 3 5 105 525
Rack 4 5 160 800
Row
B Rack 5 5 160 800
Rack 6 5 105 525
Rack 7 5 160 800
Rack 8 5 160 800

totals 40 5850

7
___________________________________________________________________________________________________

©2007 American Power Conversion. All rights reserved. No part of this publication may be used, reproduced, photocopied, transmitted,
or stored in any retrieval system of any nature, without the written permission of the copyright owner. www.apc.com Rev 2006-1
 APC APPLICATION NOTE

Figure 1 – Sample Datacenter Cluster

Cold Aisle
FRONT
R C R R C R

100% 100% 100% 100%

160 160 160 160

CFM/kW CFM/kW CFM/kW CFM/kW

BACK
Hot Aisle
BACK

R C R R C R

100% 100% 100% 100%

160 160 160 160

CFM/kW CFM/kW CFM/kW CFM/kW

FRONT
Cold Aisle

Therefore, by calculating, the temperature rise can be determined,

ΔTaverage = 3162 ×
∑ kW = 3162 ×
40
= 21.62 o F
∑ CFM 5850
and

Thot = ΔTaverage × CCM + Tcold = 21.62 × .70 + 70 = 85.1o F

From there, the capacity and performance specifications of the chilled water air-conditioner can be determined. Look up tables
from the technical data manual for pure water at sea level and interpolation indicates a sensible performance of 62,000
BTU/hr. Water flow required is found as 13.2 GPM. Adjusting for the altitude in Denver (5280 ft ASL) yields an altitude
correction factor of .875, and the adjustment for 20% ethylene glycol yields a correction factor of .93. Therefore the actual
capacity is

62,000 BTU / hr × .93 × .875 = 50,400 BTU / hr = 14.8kW

From here a number of checks must be run. First is the airflow check. The racks require 5850 SCFM. Each ACRC unit
produces 2900 SCFM. In order to achieve the required airflow, the use of 3 ACRC units is required, however, the overage will
be 1390 SCFM taking into account the CCM. Therefore, the system, as designed with four RC units, is N+1 from the airflow
vantage. Thermally, the system requires 40 kW of cooling capacity. At 50,400 BTU/hr, the converted capacity is 14.8 kW.

8
___________________________________________________________________________________________________

©2007 American Power Conversion. All rights reserved. No part of this publication may be used, reproduced, photocopied, transmitted,
or stored in any retrieval system of any nature, without the written permission of the copyright owner. www.apc.com Rev 2006-1
 APC APPLICATION NOTE

Therefore, from the thermal standpoint, the system is also N+1. Finally, we must check the leaving air temperature. Maximum
supply temperature will be attained at full airflow rate, therefore, the maximum supply temp is found to be 68.9°F. This value is
below the desired setpoint of 70°F, so the third check is satisfied. The system can be declared to be N+1.

At this point any room or partial room layout needs to be looked at from the perspective of the actual physical configuration.
The proximity of heat loads relative to coolers is critical in an InRow configuration. It is possible that the total cooling capacity is
sufficient to handle the total heat load and yet can not be configured in such a manner that all of the heat will be captured by
the necessary coolers. Please refer to Application Note 92, Best Practices for Designing Data Centers with the InRow RC for
more information.

About the Authors:

Greg Uhrhan is a Mechanical Engineer for APC. He is responsible for cooling system design, technical data, and system
applications. Greg received a Bachelor’s degree in Mechanical Engineering from University of Illinois at Urbana - Champaign in
2001, Masters in Mechanical Engineering from Georgia Tech in 2005, is a member of ASHRAE and ASME, and is a licensed
Professional Engineer in the State of Missouri.
Scott Buell is a Senior Mechanical Engineer for APC. He is responsible for cooling system design, technical data, and system
applications. Scott received a Bachelor’s degree in Mechanical Engineering from the University of Minnesota in 1992.

9
___________________________________________________________________________________________________

©2007 American Power Conversion. All rights reserved. No part of this publication may be used, reproduced, photocopied, transmitted,
or stored in any retrieval system of any nature, without the written permission of the copyright owner. www.apc.com Rev 2006-1