ASHRAE Journal - February 2024 - 14

COLUMN DATA CENTERS
TABLE 2 Properties of all liquids used in the experiment.
DETAILS
THRESHOLD
CASE
FOR SWITCH
TO 100%
ECONOMIZER,
WB (°F)
Case 1 (Baseline, No
Containment)
Case 2 (Baseline w/ Integrated
Economizer, No Containment)
Case 4A (w/ Containment)
Case 4B
Case 4C (Case 4 in Table 1)
Case 4D
Case 4E
Case 4F
Case 4G
Case 4H
Case 4I
Case 5 (Chillerless)
N/A
66/46
62
64
66
68
70
72
74
76
78
N/A
CHILLER
PLANT
ENERGY,
% OF
BASELINE
100.0
64.0
53.4
50.8
46.8
43.3
40.6
38.6
36.5
35.1
34.7
34.5
SERVER
ENERGY,
% OF
BASELINE
100.0
100.0
99.2
99.4
99.7
100
100.3
100.5
100.8
101
101
101
FACILITY
ENERGY,
% OF
BASELINE
100.0
95.0
93.0
92.7
92.4
92.2
92.1
91.9
91.9
91.9
91.9
91.9
MAXIMUM
SERVER
AIRFLOW,
% OF
BASELINE
100
100
117
121
125
129
133
137
141
146
150
154
1.140
1.140
0.965
0.979
0.998
1.017
1.031
1.043
1.056
1.065
1.068
1.068
ANNUALIZED
X-FACTOR1
MAX AVG
RACK
INLET
TEMP, °F
73.00
73.00
79.60
81.14
82.70
84.27
85.86
87.46
89.08
90.71
92.36
94.02
AVG RACK
INLET
TEMP,
°F
73.00
73.00
67.05
67.63
68.43
69.23
69.86
70.38
70.96
71.38
71.51
71.56
HIGH UNMET
HOURS AT FULL
LOAD, H/YR AVG
RACK INLET
>80.4°F
388
867
1,296
1,604
1,839
2,081
2,241
2,289
2,304
LOW UNMET
HOURS AT FULL
LOAD
H/YR AVG RACK
INLET <64.4°F
2,995
2,995
2,995
2,995
2,995
2,995
2,995
2,995
2,995
2,995
1 Values assume that Cases 4A - 4I and Case 5 can operate with rack inlet temperatures as low as 59°F (15°C). The x-factor penalty for a minimum inlet temp of 64.4°F (18°C) is
about +0.06.
for all cases in this example analysis, but could also be
further optimized.
* The addition of CRAH units may be possible and
recommended in some cases if the use of redundant
units still results in a shortfall of airflow at high
temperatures, especially if a facility has the desire to
convert to chillerless operation. Upfront measurements
could be made to better determine the increase in
airflow for a specific data center mix of IT equipment.
* As the results of Case 3, 4A - 4I and 5 demonstrate,
a wide range of temperature control strategies can be
considered on a case-by-case basis. One advantage of
chiller operation in Case 3 and 4 relative to integrated
economizer operation is that the chillers do not have to
operate at low condenser water temperatures, which
can be an issue with older chillers.
* Cases 3 and 4 could result in a step function change
in cold aisle temperatures when the chillers are either
turned on or off. Methodology to soften this sudden
rate-of-change is required to avoid IT reliability issues.
Summary
By using x-factor analysis and targeted use of existing
cooling plant equipment, one can likely achieve a
14
ASHRAE JOURNAL ashrae.org FEBRUARY 2 0 2 4
" win-win " situation, saving energy and increasing server
reliability (lowering annualized x-factor) in existing
data centers as well as in new ones.
In an existing facility, incorporating hot/cold aisle
containment is a complementary measure, able to
improve both energy efficiency and server reliability
relative to the " status quo, " or operating x-factor analysis
by itself.
As part of the x-factor optimization process,
improvements in measurement, analysis and control
systems would typically also be recommended.
A change to either the Recommended Range of
ASHRAE's Thermal Guidelines or to Standard 90.4-2022's3
definition of " unmet hours " to allow rack inlet
temperatures in the temperature range of 59°F to
64.4°F (15°C to 18°C) to be counted as acceptable, rather
than as unmet hours, is recommended by the authors.
The cooler winter operation is desirable to counteract
warmer summer conditions and keep annualized server
failure rates low.
In summary, x-factor optimization can be a
central strategy for improving a site power usage
effectiveness/mechanical load component (PUE/
MLC), reducing energy consumption, improving
http://www.ashrae.org
ASHRAE Journal - February 2024

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