Real time experimentation without affecting the actual system performance is a big challenge and expensive for huge systems. The data centers are one of such systems as well as the highest power consuming entities in the world. In such cases the scaled down testing facility provides the platform to test the actual system with its all possible emergencies. There is no attempt of building the prototype testing facility of the data center in research community which will be useful for operator to manage the system efficiently. We have design and developed a unique prototype testing facility to achieve the efficient thermal management of data center. The dimensional analysis has been done by keeping Richardson's number constant and accordingly cold-air flow-rates range were defined for 20 times scaled down prototype. The CFD modeling of the actual system and scaled down system has shown the similar thermal behavior. Furthermore, the prototype model was developed for conventional raised floor plenum (2 rows each with 5 racks) and S-Pod (3 pods with 4 racks) rack structure with 4 CRAC units. To maintain the similar environment in prototype as the actual data center the structural modification includes the four server compartments each with porous sponge (around 60 % porosity), open tiles in front of racks, the hot air supply through pipes with pin holed at each server compartment. In the scaled down structure, the heating through servers has been emulated by supplying heated air (at 100°C) in each server rack compartments with 1.67×10-4 m3/s (10 SLPM). Similarly, filtered atmospheric air supplied with 10 SLPM through the CRAC in the plenum chamber as the cold air. The exhaust has been provided through a vent at the top of the system. Twelve thermocouples have been deployed to track the temperatures at server inlet-outlets, plenum, CRAC & overhead chamber. The whole system was insulated to avoid the leakages. It was observed that the S-Pod structure has superior performance than the conventional raised floor. The major parameters affecting the data center performance such as workload on servers (hot air passed through servers) has been considered in prototype by changing the inlet flow rate. The other complexities such as server fan, perforation of tiles will be introduced in near future. This study will be a breakthrough for actual testing the critical factors of any data center.
[1]
Suhas V. Patankar,et al.
Airflow distribution through perforated tiles in raised-floor data centers
,
2006
.
[2]
Atul Bhargav,et al.
Effect of plenum chamber obstructions on data center performance
,
2015
.
[3]
B. Launder,et al.
The numerical computation of turbulent flows
,
1990
.
[4]
Hazim B. Awbi,et al.
Scale effect in room airflow studies
,
1990
.
[5]
S. Halgamuge,et al.
Can a data center heat-flow model be scaled down?
,
2012,
2012 IEEE 6th International Conference on Information and Automation for Sustainability.
[6]
Mahmoud Ibrahim,et al.
Numerical modeling of data center with transient boundary conditions
,
2010,
2010 12th IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems.
[7]
U. Ghia,et al.
Procedure for Estimation and Reporting of Uncertainty Due to Discretization in CFD Applications
,
2007
.
[8]
Ali Mubarak Almoli,et al.
Air flow management inside data centres
,
2013
.
[9]
Yogendra Joshi,et al.
Emerging Data Center Thermal Management and Energy Efficiency Technologies
,
2012
.
[10]
Atul Bhargav,et al.
Advances in data center thermal management
,
2015
.
[11]
S. Nada,et al.
Experimental investigations of air conditioning solutions in high power density data centers using a scaled physical model
,
2016
.
[12]
Marta Chinnici,et al.
Thermal Metrics for Data Centers: A Critical Review☆
,
2014
.