N. Srinarayana

Low-Reynolds-number fountain behaviour

Experimental evidence for previously unreported fountain behaviour is presented. It has been found that the first unstable mode of a three-dimensional round fountain is a laminar flapping motion that can grow to a circling or multimodal flapping motion. With increasing Froude and Reynolds numbers, fountain behaviour becomes more disorderly, exhibiting a laminar bobbing motion. The transition between steady behaviour, the initial flapping modes and the laminar bobbing flow can be approximately described by a function FrRe2/3 =C. The transition to turbulence occurs at Re > 120, independent of Froude number, and the flow appears to be fully turbulent at Re ≈2000. For Fr > 10 and Re 120, sinuous shear-driven instabilities have been observed in the rising fluid column. For Re 120 these instabilities cause the fountain to intermittently breakdown into turbulent jet-like flow. For Fr 10 buoyancy forces begin to dominate the flow and pulsing behaviour is observed. A regime map of the fountain behaviour for 0.7Fr 100 and 15Re 1900 is presented and the underlying mechanisms for the observed behaviour are proposed. Movies are available with the online version of the paper.

Thermal Performance of an Air-Cooled Data Center With Raised-Floor and Non-Raised-Floor Configurations

In this paper, the thermal performances of an air-cooled data center with raised-floor and non-raised-floor configurations are compared with respect to the room and ceiling return strategies. The thermal performance of the data center is evaluated in terms of supply heat index, rack cooling index, total irreversible loss, and the number of racks with at least one server exceeding the maximum recommended and allowable inlet air temperature according to American Society of Heating, Refrigeration, and Air-Conditioning Engineers (ASHRAE) thermal guidelines. The numerical simulations are conducted providing an insight into the flow and temperature distributions, and thus giving a better understanding of the cooling issues. It is found that using a ceiling return strategy for the return of hot exhaust air to the computer room air conditioning units gives a better thermal performance of the data center, for both raised- and non-raised-floor strategy, as compared to the room return. The findings are then extended to a geometrically complex operational data center to improve its cooling effectiveness. The paper also highlights the drawback of using supply heat index alone as a performance metric.

A comparative study of raised-floor and hard-floor configurations in an air-cooled data centre

The thermal performance of an air-cooled data centre with raised-floor and hard-floor configurations are compared with respect to the room and ceiling return strategies. The supply heat index, rack cooling index and number of racks that exceed the maximum recommended and allowable rack inlet air temperature according to ASHRAEs thermal guidelines, are used to evaluate the thermal performance. Numerical simulations are conducted to obtain thermal and velocity fields. Results show that using a ceiling return strategy for the return of hot exhaust air to the CRACs gives a better thermal performance of the data centre, for both raised and hard-floor strategy as against the room return. The paper also addresses an important limitation of the supply heat index performance metric.

Analysis of exergy destruction in data centres

IT equipment and systems, housed in data centres, consume a considerable amount of electricity. Most of the electrical energy consumed by the data centre IT equipment is released in the form of heat. From a second law of thermodynamics analysis point of view, the mixing of hot and cold air streams in the room caused by hot air recirculation is an irreversible process, leading to wasted work potential in data centres. In the work presented here, a numerical analysis of flow and temperature distribution of a raised-floor data centre is conducted in order to evaluate the thermal performance of the data centre. Subsequently, from flow patterns and temperature profiles, a detailed exergy analysis of the data centre is performed to get a better understanding of the room airspace irreversibilities. The amount of the exergy as well as the exergy destroyed in the airspace is evaluated in order to cast a light on the nature of the irreversibilities. Finally, new performance metrics based on the second law of thermodynamics are proposed and used to assess the performance of a data centre. A comparison between the performance metrics based on the first and second law of thermodynamics as well as the limitations of the newly derived performance metrics are discussed.

Thermal Performance of Data Centers-Rack Level Analysis

This paper analyzes the thermal performance of a data center on a rack level, by utilizing racks stocked with 1U servers. Eleven different rack models covering a wide range of server arrangements with void spaces are presented and modeled in a raised-floor configuration. The objective is to investigate the thermal performance of semipopulated racks with different server arrangements. The inlet and outlet temperature profiles of the models are presented and the overall effect of each model on the thermal performance of a specific rack is discussed. Findings point toward the detailed modeling of semipopulated racks to investigate the main locations of the temperature spikes adjacent to the rack inlet and outlets as well as the recirculation zones.