Compact Models for Rapid Thermal Modeling of Data Centers: Phase 2


Thermal design and operation of air cooled data center facilities incorporating new features such as aisle containment and improved air distribution using new tile designs can benefit significantly from rapid thermal modeling. Proper orthogonal decomposition (POD) methodology has been shown to be highly effective in reducing model size, and hence, computational time required to adequately capture the flow and temperature distributions within data center facilities.

Sample Project

Tile/Rack Level Air Flow Model Development

Team:?

Goals:

  • Explore the extension of POD-based modeling of contained aisle facilities.
  • Consider models of new passive and active tiles for improved air delivery.

Outcome:

  • Modified body force model suggests better prediction than generally used porous jump model.
  • Much lower mesh size for modified body force model as compared to geometrical resolution. (4.2M vs. 45K per rack)
Tile Rack

Air flow / rack flow (%)

Geometrical resolution

Modified body force

Porous jump

Aisle top

-46.18

-43.48

-0.77

Aisle sides

35.77

35.31

12.26

Aisle Center

7.02

4.78

-11.41

Gap

3.39

3.32

0.15

Center for Energy-Smart Electronic Systems

Villanova ES2 Team Areas of Expertise
Thermal management
Electronics cooling
Fluid mechanics
Convective heat transfer
Sustainable energy technology
Multiscale systems
Computational fluid dynamics
Modeling

Contact

Director: Dr. Amy Fleischer, Professor and Chair, Department of Mechanical Engineering, Director of the NovaTherm Laboratory

Dr. Amy Fleisher, ES2 Director
Email: amy.fleischer@villanova.edu 
Phone: (610) 519-4996

Associate Director
Dr. Aaron Wemhoff, ES2 Associate Director
Email: aaron.wemhoff@villanova.edu 
Phone: (610) 519-8045