GLUMAC - Engineers for a sustainable future

by Mike Nichols, P.E., LEED^®AP, San Francisco Associate Principal

For those who own and operate mission critical facilities, complex and dynamic cooling requirements continue to present a challenge, and the need for a more sophisticated cooling solution. This resolution must deal with the diverse power densities and uneven loading that occurs in a majority of existing data center facilities. Conventional data centers operate at a power density range of 30 to 150 Watts per square foot (W/sf). Power densities of the €next generation’ of data centers will be double or even triple current densities, due to more concentrated heat output of blade servers and other new data processing equipment being installed.

Increased power densities create an increase in thermal mixing and reduce operating efficiencies. Thermal mixing is common in data centers because in most cases the cool conditioned space is not specifically isolated from the hot return air, with the exception of some thermal stratification that occurs due to differences in air density. This allows hot air discharged from the racks to temper air serving the cold aisle. To increase cooling performance where it is needed, minimize construction costs, and reduce thermal mixing, mechanical design engineers are turning to CFD modeling (computer simulation) as a design tool.

Samuel Graves, a Senior Mechanical Engineer at HP says, "Three years ago, HP started an initiative to bring energy efficiencies into our data centers. At the core of this work is the managing of thermal mixing. In some cases, if thermal mixing can be eliminated, we are doing so and in some instances where thermal mixing can not be eliminated, we are attempting to manage it in a way that still increases our overall energy efficiency. The management of thermal mixing and energy efficiency is at the core of the design of the 6 mega Data Centers HP is currently building."

Computational Fluid Dynamics (CFD) uses numerical methods and algorithms to solve and analyze problems that involve fluid flows. The problem is maintaining proper server rack temperatures to achieve required reliability. A CFD model allows us to analyze the pressure, temperature, and flow characteristics of air above and below the raised floor. With this information in hand, we can take the necessary steps to prevent hot spots and reduce thermal mixing. Computers are used to perform the millions of calculations required to simulate the heat transfer that takes place between the cooling fluids, and the complex surfaces of the equipment that they cool.

The process of HVAC design has historically been accomplished by an application of empirical guidelines and prior experiences that are later refined and confirmed by engineering calculations. With respect to cooling high-density data centers, there are a great number of parameters that are counter-intuitive and prevent the use of a standard "rule of thumb" approach to the layout of server racks and cooling equipment.

To ensure that these computer systems operate properly and don’t fail prematurely, they must be adequately cooled. Each server rack has air flow requirements that must be met in order to maintain the desired rack temperature. A majority of data centers are designed and built with an under floor air distribution system, which allows the end user the flexibility to relocate equipment and increase power densities, without modifying expensive ductwork and relocating major cooling equipment.

Data centers are frequently in a state of flux, and the way cabinets and racks are populated rapidly changes to accommodate advancing technology and client needs. The change in cooling load requirements is dealt with by way of adding, eliminating, or relocating floor registers to deliver cooling to the load. The difficulty lies in getting the correct quantity of registers and supply air to the load. With CFD modeling, this task becomes much easier.

Glumac utilizes CFD modeling in the design of new data centers as well as re-arranging existing data centers in order to optimize performance. In addition, this tool can be used for studying failure scenarios, identifying cooling problems and evaluating possible solutions. The ability to have millions of engineering calculations represented graphically and dynamically with computer simulation, provides a valuable tool to the engineer to rapidly evaluate the results of a given engineering decision. This allows the engineer the flexibility to perform "what if" scenarios that would otherwise be impossible due to time constraints.