ASHRAE Journal - February 2024 - 12

COLUMN DATA CENTERS
energy by 36%, and annual facility energy (which
includes IT energy) by about 5%.
* Case 3 is the first " x-factor optimization " option.
Note that cooling plant energy drops by 45% compared
to the baseline, facility energy use by 6% (better than
Case 2) and annualized x-factor is also reduced even
though there are rack inlet excursions above the
recommended range (to 84.6°F [29.2°C]).
* Case 4 is the second " x-factor optimization "
option. Note that with containment added in addition
to using x-factor excursions, Case 4 provides better
results than Case 3 in all pertinent areas: cooling
plant energy drops by 53% (vs. 45%) compared to the
baseline, facility energy use drops by 8% rather than
6%, annualized x-factor drops to 1.00 rather than 1.08,
and maximum server inlet temperature drops from
84.6°F to 82.7°F (29.2°C to 28.2°C). The complementary
nature of implementing aisle containment at the same
time as x-factor optimization is clearly apparent in
this comparison.
* Case 5 (year-round chillerless operation) has
the lowest energy consumption, but only by about 1%
relative to Case 4. It also has the highest maximum
average server inlet temperature (94.0°F vs. 82.7°F
[34.4°C vs. 28.1°C] for Case 4), and a higher annualized
x-factor (1.07 vs. 1.00 for Case 4). Case 5 would not be
compatible with existing ASHRAE Class 1 IT equipment
(which has a maximum allowable inlet temperature of
89°F [28.3°C]), whereas Case 4 would be.
Parametric Analysis of Cases 4 and 5
Cases 4 and 5 are similar in most respects. The
exception is that Case 4 initiates chiller operation at a
specific ambient wet-bulb temperature (66°F [18.9°C]),
while in Case 5 the chiller is never turned on. Of interest
in this section is the impact of the economizer wet-bulb
temperature threshold on pertinent variables. For this
part of the analysis, the wet-bulb temperature for onset
of chiller operation was varied from 62°F (16.7°C) all the
way up to chillerless operation, in 2°F (1°C) increments.
The results are shown in Table 2 and Figure 2.
Highlights from Table 2 and Figure 2 include:
* Annualized server energy is remarkably constant
over all these solutions, within 1% of the baseline (99.2%
to 101%). It is indicative of that fact that the excursions
in energy associated with limited operation at high
temperatures are reasonably counterbalanced by
12
ASHRAE JOURNAL ashrae.org FEBRUARY 2 0 2 4
extended operation at low temperatures during winter.
* Maximum server airflow is higher than baseline
for all x-factor cases, but there is a wide variation,
dependent on the maximum average server inlet
temperature. The maximum airflow is 117% of baseline
with a maximum server inlet temperature of 79.6°F
(26.4°C) (Case 4A), rising to 154% of baseline with a
maximum server inlet temperature of 94°F (34.4°C)
(Case 5). A facility constrained by CRAH airflow would
thus be better off with a Case 4 option (especially Case
4A) than with chillerless operation (Case 5).
* Since the maximum average rack inlet temperature
of Case 4A falls within the recommended range of
ASHRAE's Thermal Guidelines (64.4°F to 80.6°F [18°C to
27°C]), unmet hours from the standpoint of ASHRAE
90.4-20223 would be zero except for the fact that there
are about 3,000 hours/year of operating hours in the
cooler temperature range of 59°F to 64.4°F (15°C to
18°C). For Case 4A to fully align with Standard 90.4's
unmet hours requirement, the minimum rack inlet
temperature of 59°F (15°C) would have to be raised to
64.4°F (18°C). The energy impact of this change would
be minimum, but the x-factor would increase from
about 0.96 to 1.02 (i.e., about a 6% increase in the server
failure rate would be expected).
Other Considerations
There are several other items consider as part of the
x-factor optimization process with an existing facility:
* In simulating the various options, it is noted that
different cooling tower control strategies were examined
and used for different cases. Strategies examined
included: a) constant cooling tower setpoint (such as
75°F [23.9°C] condenser water supply temperature)
year-round, b) cooling tower approach varies as a
function of wet-bulb temperature, c) cooling tower
approach varies as a function of both wet-bulb
temperature and IT load and d) cooling tower fans
operate at 100% speed except as needed to prevent
overcooling. We found that strategy b) was close to
optimal when not in economizer mode and that
approach d) was close to optimal when in economizer or
chillerless mode. An optimization calculation provides
the best results, as the optimal strategy can also change
with different wet-bulb temperatures and IT loads.
* Both chilled water and condenser water pump flow
rates were assumed to be proportional to cooling load
http://www.ashrae.org
ASHRAE Journal - February 2024

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