Cold Aisle Containment – Achieving a Controlled Environment Out of Thin Air By Jeremy Hartley, Managing Director, Dataracks
Aisle containment is an important approach to airflow management, which involves controlling the routes cold air takes from air conditioning units (CRAC) through the data equipment and beyond.
Setup
A free cooling unit was installed inside the test chamber that incorporated a variable, energy efficient plug fan. The unit was connected to a chiller that could simulate a realistic ambient temperature For the main tests, the server load
was simulated at 43kW. This was achieved with 12 simulators, each with 4kW available. Monitoring of the power levels of both the servers and air conditioning unit was achieved by installing power monitoring equipment. Six temperature sensors were placed
in different areas inside the cold aisle. Two sensors were positioned to monitor the return air, and two to monitor the supply air. Load was 43kW. The fan was 1.6kW at 8v Fixed Speed. The water temperature was 15°C with the ambient air temperature at 10°C. Three main tests were performed:
Full containment testing
This series of tests involved achieving a steady cold aisle closure state and then removing various parts of the aisle one by one to measure the effect. The process was then reversed. The cold aisle temperatures had
a difference of 0.1K whilst the open aisle temperatures showed temperature differences of 14K. The return air temperature had also dropped, indicating mixing of air and de-rating of the air conditioning unit. This
also shows that the roof is the most important feature of the cold aisle. The results showed that full aisle
containment is important to achieve the maximum benefits of efficient cooling; the role of the roof system was particularly apparent. With no cold aisle containment,
temperatures at the servers are not as controllable, with temperatures of up to 34°C recorded. 74% of server failures occur in the
top third of racks. Without containment, mixing at the top of the rack will take place and this represents inefficient use of cold air. Cold aisle containment provides
temperature equilibrium across servers, no matter where they are positioned inside the racks.
Failure retention tests
This series of tests involved the air conditioning unit being switched off, simulating a power failure. This gives an indication of the time available to put in place remedial action in the case of a cold aisle closure system and also with an open system. The parameters were the same as Test 1. No Cold Aisle Containment meant
the temperature in the aisle reached 42°C in 4 minutes. With Cold Aisle Containment the
temperature in the aisle reached 39°C in 10 minutes, allowing valuable extra time to react compared with an open system. The test results showed that with an
open setup, temperatures quickly rose and reached the same temperature as the return air. With the cold aisle setup, the aisle temperatures rose more slowly and remained 5K lower than the return air temperature, indicating that servers inside a data centre could survive longer.
Variable load tests
The variable load tests set out to evaluate the benefits of having variable speed plug fans fitted to the air conditioning units in order to react to load changes within a data centre. The fan on the CRAC unit was
set to vary between 8v-9v. With a 43kW load the fan was running at 9 volts, which uses 2.5kW of fan power. The load was then reduced to 33kW and the fan reacted to the change by dropping its speed to 8 volts, using 1.6kW of fan power, meaning 36% less energy was used. The test results showed that with
varying loads in a data centre, energy could be saved by using variable plug fans. As the load drops, the variable fan will slow down to match the new cooling demand. In this case, the fan used 36% less energy to meet the lower cooling demand. Single speed fans will run flat out even
if the load drops. Variable speed fans in a data centre are therefore essential.
Conclusions
These tests have been valuable in demonstrating the benefits of aisle containment, and in quantifying some of the key parameters. The main conclusions were: full
aisle containment with air flow management can result in a much more controlled environment in the data centre. Significant energy savings can be made. Free cooling can be utilised for a greater part of the year. In the event of a power failure within the data centre, the use of aisle containment significantly slows the rate of temperature increase and provides a longer ‘buffer time’ for system repair. Full aisle containment, including a roof system, is imperative in achieving the maximum control and benefits from this approach. Benefits include increasing set points and return temperatures, optimising the use of EC variable speed drive fans, eliminating mixing, short circuiting, hot spots and by pass air, and increasing rack density.
The concepts of enclosing part of an operational data centre as a means of controlling airflow (aisle containment) is not new, but the benefits are not universally appreciated. A trial was set up in order to investigate the parameters which make up effective aisle containment, and to quantify the potential benefits in a realistic setting.
www.netcommseurope.com
NETCOMMS europe Volume I, Issue 6 2011 45
Image:
www.freeimages.co.uk
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52