Data Center Journal

Volume 34 | October 2014

Issue link: https://cp.revolio.com/i/394834

Contents of this Issue

Navigation

Page 12 of 32

10 | THE DATA CENTER JOURNAL www.datacenterjournal.com r esearchers at Villanova University estimate that data centers are responsible for more than two percent of U.S. electrical usage. All of that energy creates massive heat—and servers do not like heat. It can cause them to become slow, or malfunction outright, leading to significant downtime and un- wanted expenses. A properly chosen and maintained cooling tower can help organizations avoid this problem, but becoming familiar with the different types of towers and proper maintenance processes is required. Some towers are more appropriate for data center use than others, and even then, choices must be made on the basis of size, ef- ficiency and other factors. Likewise, main- tenance is not simply about making sure a cooling tower's fan is working properly (al- though that's important). It's about making sure the tower is protected from corrosion, hard water and biological factors. Before data center operators begin to concern themselves with selection and maintenance, they should first develop a baseline understanding of how a cooling tower works—which will help them better grasp how specific towers can work in their operation. reJeCtIng heat anD keePIng thIngs Cool In reality, cooling towers do not keep things cool as much as they reject heat, which they do by transferring heat from the water source to the atmosphere through the process of evaporation (latent) and sensible heat transfer, thereby reducing the water temperature. Pumps circulate the water through the tower and, typically, a chiller that provides chilled water to cool the loads in conventional water-cooled facilities. In this sense, cooling towers are somewhat like traditional HVAC systems. A home air conditioner will use an air- cooled condenser coil and fans located outside to transfer heat to the outside environment. Cooling towers do more or less the same thing, except they use larger fans and different types of heat transfer mediums (especially water) to get the job done. seleCtIng the rIght tower anD CoMPonents Both data centers and cooling towers have unique design and operating condi- tions. Sizes are different, and some designs expend more energy than others. ere- fore, it's important to note the differences to determine the best cooling towers for specific data centers. In terms of sizing, cooling-tower se- lection should be dictated by the following: 1. e heat load that is being gen- erated in the data center 2. Water inlet and outlet tempera- tures (range or delta T) 3. How close the leaving water temperature is to the design wet-bulb (approach) 4. Design wet-bulb temperature, which is read from a wet-bulb thermometer placed in a mov- ing air stream. When any three of these four are held constant, the tower's size will vary 1. Directly with heat load 2. Inversely with range 3. Inversely with approach 4. Inversely with entering wet-bulb Tower efficiency is also important and is affected by fan power, which can be regulated and enhanced through use of a variable-frequency drive (VFD). VFDs au- tomatically adjust fan speeds as conditions change while still maintaining necessary airflow and leaving water temperature. Cooling towers with VFDs should certainly be considered by data center operators looking to maximize efficiencies. Energy consumed by towers, pumps and chillers has gained significant attention over the past decade. Flow rates, approach and range are all variables that can be adjusted to achieve the best efficiency. It's always been an engineering consideration, but new approaches are being tested in an attempt to improve plant efficiency. As such, water flow rates and delta T's—which measure the difference between entering and leaving water temperatures and deter- mine required gallons per minute—should be considered when selecting cooling towers to achieve the best return on invest- ment over the lifespan of the equipment. Cooling-tower selection is more than just basing a decision on size and efficiency, however. Data center operators must also consider additional environmen- tal and mechanical factors that can affect tower performance. One of the biggest risks to plant operation is the loss of makeup water, since water is necessary for heat transfer through evaporation. erefore, it's impor- tant to have more than one water source if possible. For example, recycled water (environmentally friendly) could serve as a primary source, with potable water serving as backup. Alternatively, two potable-water feeds could provide makeup water. If two water sources cannot be achieved, it's prudent to have local water storage onsite to deal with short-term disruptions in water delivery. Data center operators may want to have water storage in underground sumps, or in above ground tanks, and should have arrangements in place for emergency water-delivery services. Remember: no water means no cooling, and critical loads going down. the Most CoMMon Data Center CoolIng towers ere is a handful of cooling-tower models that are most suitable for data cen- ter use. Each has its own benefits, as well as some things to watch out for. Forced-draft towers ese types of towers comprise high-powered centrifugal fans that push air up through the towers' fill. e fill creates a large surface area for the water to run down. Forced-dra towers also feature baffle-like devices called dri eliminators to prevent water from driing outside of the tower. Air travels through the dri eliminators aer leaving the tower's fill and spray zones. Dri eliminators are criti- cal to reducing water loss from the tower. Forced-dra towers are generally noisier than other towers, however. Induced-draft towers Instead of pushing air through the fill, induced-dra towers pull air through. ey generally have a smaller footprint than forced-dra towers and feature large fans at the top. ere's less risk of recircu-

Articles in this issue

Links on this page

Archives of this issue

view archives of Data Center Journal - Volume 34 | October 2014