'Data Center Cooling Strategies' simulation project by gschiaffini


#1

I created a new simulation project called 'Data Center Cooling Strategies':

Data Center Cooling Strategies


More of my public projects can be found here.


#2

Project overview

The purpose of this project is to investigate two different data center cooling designs and check for reduction of energy consumptions. Two design scenarios are as shown in the figure below:

1

  • Data center Design 1: Raised floor configuration

  • Data Center Design 2: Hot aisle containments, raised floor and lowered ceiling configuration

The cooling system in a data center is often based on a raised floor configuration. When this technique is used, the cold air enters the room through the perforated tiles on the floor, that in turn cools the server racks. The hot air is then removed from the surrounding through the CRAC’s.

With the help of hot aisle containment, the hot aisle does not spread within the room and is directed directly to the CRAC’s. This increases the efficiency of cooling by preventing the mix of hot and cold air together.

CFD Simulation Results

The results presented show the temperature and velocity contours for the initial and modified design configurations.

Solution Description

The numerical study investigated the temperature distribution and the velocity field inside of the server room for both the design configurations. The figure below shows the velocity and temperature fields for a mid section of the baseline design. It is observed that the hot air is not only found to the rear of the racks but also present in the region of cold aisle. This is due to the mixing of both the cold and hot aisles within the data center surrounding. The maximum velocity for this baseline design is found to be 0.44 m/s with a temperature range of 28.6 to 49.7 degree Celsius.

It can also be seen, considering a different section plane, which is the direction of the flow due to convection phenomena, using a vector plot. The temperature contour shows that the zones between two server racks are much cooled with respect to the other ones. The reasons for this can be understood by looking at the flow patterns.

3

It can be seen in the above image that in the server rows where inlets are present, the top of the racks see descending flow direction, instead of the desirable ascending flow from the inlets themselves. This is due to the strong recirculation currents driven by thermal buoyancy forces. This effect is very undesirable, as it reduces cooling effectiveness specifically for the top shelves of the server racks. This effect could be minimized by allowing for proper air flow above the racks, either by increasing ceiling height, placing more distributed outlets on ceiling, or using some kind of active flow control system (fans) to direct flow above the server racks. Or more simply, by preventing the hot air coming from the racks from freely circulating.

4

The temperature plot shows the significant temperature stratification that is to be expected given the large recirculation currents. It can be seen that only the lower most servers are receiving appropriate cooling.
The next two pictures show the velocity and temperature distribution for the improved design scenario, for a middle section pane.

The velocity field shows that the flow now is well driven to the outlets, due to the presence of containment on top of the racks. This fact results also in a better temperature distribution. In particular, the cold zones, between the server racks, are extended.

6

The above figure shows how the hot containment prevents the ascending flow from recirculating back to the inlet rows. This results in a cleaner overall flow pattern than was seen in the previous design. In the figure below where the temperature field can be seen, it is also evident how the new design reduces temperature stratification, particularly in the contained regions between the servers.

7

Final Result comparison

The next two plots explain how the average temperature calculated for each rack is in general lower for the improved design that is about 23%.

8

This previous result is reflected in the increase of power that has to be supplied to the server because of servers overheating. In the average, considering the improved design, it can be achieved a saving of 63% of power supplied.

9