ASHRAE Journal - October 2014 - 64

2014 ASHRAE TECHNOLOGY AWARD CASE STUDIES

FIGURE 2 Cooling energy use vs. ambient wet-bulb temperature.

FIGURE 3 Schematic of air-handling unit.

Supply Air

0
20
40
60
80
Daily Average Ambient Wet-Bulb Temperature (°F)

Data Hall Fans Energy

UPS Hall Fans Energy

Cooling Plant Energy

wet/dry cycling of the evaporative media and associated
wide variation in supply air humidity and temperatures.
The simplicity of controls lies in the fact that the return/
exhaust air is mixed with the incoming outdoor cold air
to continually achieve a desired wet-bulb temperature.
The mixed air then passes over a wet media to achieve
stable supply air temperature.
We allowed the mixed air wet-bulb temperature to
float between a minimum and a maximum setpoint.
The maximum setpoint was based on our air-distribution effectiveness. When the outdoor air wet-bulb was
within the range, no mixing was needed. When outdoor
wet-bulb temperature exceeded the maximum setpoint,
the chiller plant was activated to pre-cool incoming air,
which reduced its dry-bulb as well as wet-bulb temperatures. In most scenarios, only small amounts of sensible
cooling of the dry air ambient air was sufficient to bring
its wet-bulb temperature to desired levels.
The amount of cooling required was only that which
was needed to trim outdoor air wet-bulb temperature to
the desired level, which is much smaller than what would
be needed to provide 100% cooling of the recirculation
air. This explains why the chiller plant was not required
for most of the year except during the hot months of July
and August and even then the amount of cooling energy
required was small, as shown in Table 1 and Figures 1 and 2.
Cold air distribution and uncontainment. We turned
a design oversight into an asset. Containment of hot air
or cold air and preventing mixing between the two in the
data center space is now widely accepted. However, we
intentionally chose not to fully isolate cold and hot airstreams, but allowed some directed mixing between the
two in the data center space to overcome design limitation
64

Evaporative Media

0.05

Cooling Coil

Outdoor Air

0.10

Cooling Coil
Air Leaving

0.15

Mixed Air

0.20

0.00

Return Air

Exhaust Air

Energy Use Per Unit of IT Equipment
Energy Use (kWh/kWh)

0.25

ASHRAE JOURNAL

ashrae.org

O CT O B E R 2 0 1 4

TABLE 2 Number of hours in a year when free cooling is available in Salt Lake City.
CONVENTIONAL
AIRSIDE ECONOMIZER

INNOVATIVE
ECONOMIZER

NUMBER OF HOURS OF 'FREE'
COOLING (<67.5°F)

73%

99%

NUMBER OF HOURS BACKUP
COOLING NEEDED

27%

NUMBER OF HOURS
HUMIDIFICATION ENERGY NEEDED

78%

and gain energy efficiency. This may sound counterintuitive to the now widely accepted practice of fully isolating
hot and cold air, but its rationale is explained below and
it is providing excellent energy savings in the field. For
the record, in 2004 the author was likely the first one to
demonstrate and champion the use of physical barriers
between hot and cold airstreams to prevent its mixing at
Oracle's large data center in Austin, Texas.2,3
For proper functioning of any containment, the airflow
across the IT equipment must be equal to the airflow
across the cooling equipment. If not, either the airflow
over the IT equipment will starve or airflow over the
cooling equipment will starve. Therefore, the airflow
of the cooling equipment must be sized to meet the
airflow of the IT equipment. The airflow over IT equipment is derived from the design temperature rise when
air passes through it. The average design temperature
rise in IT equipment is ever-increasing as it becomes
more power consuming. Our data center was designed
to handle the latest IT hardware, and most of these are
designed with lower airflow and higher temperature rise
of ~30°F (17°C) across the IT equipment.
The cooling equipment air-handling units were sized
to match the lower airflow for a 30°F (17°C) temperature
rise. However, we noticed that the average temperature

ASHRAE Journal - October 2014

Table of Contents for the Digital Edition of ASHRAE Journal - October 2014