ASHRAE Journal - February 2024 - 10

COLUMN DATA CENTERS
TABLE 1 Properties of all liquids used in the experiment.
CASE
ANNUAL COOLING
PLANT ENERGY, kWh/
RELATIVE PLANT
ENERGY USE
(1) Baseline 60°F CHWS, No
Economizer, No Containment
(2) Baseline w/ Integrated
Economizer, No Containment
(3) X-Factor with No Containment;
Excursions to 85°F
(4) X-Factor with Containment;
Excursions to 83°F
(5) X-Factor with Excursions to
94°F, No Chiller
1,626,025
100%
1,043,390
64%
887,193
55%
761,597
47%
561,303
35%
ANNUAL SERVER
ENERGY, kWh1/
RELATIVE SERVER
ENERGY USE
8,891,027
100.0%
8,891,027
100.0%
8,915,825
100.3%
8,864,217
99.7%
8,983,545
101.0%
ANNUAL FACILITY
ENERGY, kWh/
RELATIVE ENERGY
USE2/PUE
11,850,706
100%
1.33
11,268,072
95%
1.27
11,140,382
94%
1.25
10,955,447
92%
1.24
10,892,379
91%
1.21
MAXIMUM SERVER
AIRFLOW, cfm/
RELATIVE MAX.
AIRFLOW
120,353
100%
120,353
100%
156,381
130%
150,480
125%
185,638
154%
ANNUALIZED
X-FACTOR/EXPECTED
ANNUAL SERVER
FAILURE RATE3
1.14
4.6%
1.14
4.6%
1.08
4.3%
1
4.0%
1.07
4.3%
AVERAGE RACK INLET
TEMP/MAX RACK
INLET TEMP4/WB
THRESHOLD
73.0°F
73.0°F
N/A
73.0°F
73.0°F
66°F-46°F WB
71.2°F
84.6°F
62°F WB
69.3°F
82.7°F
66°F WB
71.6°F
94.0°F
N/A
1 Annual server energy does not include UPS or electrical distribution losses (i.e., " pure " IT energy).
2 Facility energy consumption is defined as the sum of the chiller plant energy + annual server energy with 15% electrical overhead added to account for UPS and distribution losses.
3 Server failure rates assume that an x-factor of 1.0 corresponds to an individual server failure rate of 4.0% per year. Annual server failure rate is proportional to annualized x-factor.
4 Server inlet temperatures assume a 7.0°F (4°C) approach temperature for CRAH units or equivalent cooling coils, and either a 1°F (0.6°C) average temperature rise between the
CRAH discharge and the server inlet for cases with containment, or 6°F (3°C) average temperature rise between the CRAH and the server inlet for cases without containment. The
" max rack inlet temp " as presented here is the maximum average server inlet temp on the " worst case " hour of the year using TMY3 wet-bulb data. Individual server temps may be
higher or lower depending on containment level and position in the rack. The " WB threshold " refers to the ambient wet-bulb temperature at which waterside economizer operation
occurs. In the case of integrated economizer (Case 2), the first value denotes the start of economizer operation and the last value denotes entry into full economizer mode.
Newark, N.J., was used, and an IT facility and cooling
system very similar to that analyzed in the March 2023
ASHRAE Journal column (nominal 1,000 kW IT load,
water-cooled chillers with waterside economizer) was
chosen for simulation. Five different operating scenarios
(Cases 1 to 5) were examined:
1. For the baseline (Case 1), a " legacy chilled water
plant " with 60°F (15.6°C) chilled water and no
economizer, with cooling tower temperature controlled
to 75°F (23.9°C) off the tower whenever possible,
and towers operating at 100% fan speed when wetbulb
temperatures are too high to realize this design
value. The facility has a hot/cold aisle IT equipment
arrangement but does not have significant containment
between the two aisles;
2. Case 2 is identical to Case 1, except that an
integrated waterside economizer has been added. This
could be an expensive retrofit, but is included to allow
for an energy comparison with x-factor Cases 3, 4 and 5;
3. Case 3 is similar to Case 1 (no containment, no
10
ASHRAE JOURNAL ashrae.org FEBRUARY 2 0 2 4
integrated economizer), but a parallel economizer
operation is allowed with the constraint that the
maximum average rack inlet temperature (during
the highest ambient wet-bulb condition) is limited to
85°F (29.4°C).
4. Case 4 is similar to Case 3, except that aisle
containment has been added, and the maximum
average rack inlet temperature (during the highest
ambient wet-bulb condition) is limited to 83°F (28.3°C).
5. Case 5 is similar to Case 4, except that no chiller
operation is allowed.
Highlights from Table 1 include the following:
* For Case 1, note that the baseline x-factor is 1.14.
This is because even though the average discharge
temperature from the CRAHs is 67°F (19.4°C), the lack
of aisle containment results in an average server inlet
temperature of 73°F (22.8°C) (above the 68°F [20°C] that
would align with an x-factor of 1.00).
* For Case 2, note that relative to the baseline, an
integrated economizer drops annual cooling plant

http://www.ashrae.org

ASHRAE Journal - February 2024

Table of Contents for the Digital Edition of ASHRAE Journal - February 2024