Application of Computational Fluid Dynamics Analysis for Data Center Cooling: Strategy Evaluation and Efficient Optimization

Alexy Kolesnikov, CPP, Inc.

Over the years, computational fluid dynamics (CFD) have been used to simulate increasingly more complex fluid/gas flows and heat exchange processes for a variety of applications encountered in, most notably but not limited to, the aerospace, automotive and nuclear industries. Aided by rapid advances in the two areas most pertinent to HVAC applications, namely turbulence modeling and parallel computing technology, CFD finds itself uniquely positioned to provide high fidelity airflow distribution data. Within the data center application framework CFD can be used to examine the air movement and heat stratification within the floor data storage space during the preliminary design phase or post design with the aim of ventilation and space cooling performance characterization and optimization. With the output data resolution provided by a CFD study being unavailable otherwise, one of the key benefits of a numerical simulation is its inherent ability to directly correlate airflow and heat load distributions integral for cooling performance characterization.

Presented is a CFD study performed to assess heating/cooling operation parameters of the data center floor space by facilitating visualization and characterization of hot/cold air convection within the domain, thus addressing cooling efficiency of the proposed design configuration. A detailed three-dimensional computer model of the data center floor space, including equipment racks with corresponding flow rates and heat loads, relevant furniture, additional heating/cooling units and air supply and exhaust diffusers, was built to facilitate the simulation. The final CFD model contains approximately 12 million fluid cells to simulate airflow and temperature distribution within the facility covering approximately 90,000 square feet of floor space. Evaluation of the symmetry of the geometric arrangement within the facility resulting in airflow and temperature distribution similarity among the repeatable sections of the space is addressed, aimed at restricting the CFD study to a section of the actual floor space. Data available from the study includes detailed velocity, temperature, and pressure distributions within the space allowing for identification of high/low velocity and high heat load regions with potentially poor cooling performance. Recommendations aimed at improving system's heat load distribution and enhancing energy efficiency are conceptualized in the presentation based on the obtained simulation results.

Biography:

Alexy Kolesnikov was a research and development manager at Flow Sciences, Inc. (2000-2004). He lead the company's CFD-driven research and development program and was responsible for providing technical direction in new product design, including identification, assessment, and implementation of new design concepts aimed at maximizing product performance, performing computer modeling, and optimizing airflows in ventilated enclosures, interfacing with manufacturing, and delivering client presentations. He holds a patent covering the CFD-optimized laminar low airflow hood design.

He was also a supervisor within the Worldwide CFD Consulting Services group at CD-adapco (2004-2007), where he managed multiple projects performed for established clients focusing on a variety of automotive applications (e.g. underhood modeling, engine cooling, interior airflow, and acoustics).

Kolesnikov joined CPP, Inc. in 2007 as a senior associate with responsibility for CFD business segment development, project management, and execution. He has authored peer reviewed papers in the Journal of Computational Physics and the American Biological Safety Association Journal among others, and presented at numerous conferences including Labs21, Annual ABSA, AIHce and ASME/JSME conferences. He is also an ASHRAE member.

Back to Agenda