Compiled by Eamonn Ryan based on the SAIRAC Johannesburg Centre Tech Talk on September 19 by Michael Young, mechanical engineer
As we explore the design of data centres, it’s crucial to understand the role of various components and how they interconnect. This is Part 9 of a ten-part series.
Michael Young, mechanical engineer. © RACA Journal
One key element in this design is the heat exchanger, which dictates operating temperatures on both the facility and technology sides.
Defining system parameters
Let’s consider a scenario where we need to achieve 650kW of cooling. This is the heat that must be transferred from the equipment to the water flow returning to the central system. The starting point is identifying the appropriate cooling class; for instance, we might begin with a W2 class, which has a supply water temperature of 27°C.
In designing the cooling system, we need to establish the approach temperature—the temperature difference between the water leaving the chiller and the water exiting the cooling unit. Typically, we design this approach temperature to be between 4 to 5°C. For our scenario, if we set our approach temperature to 5°C, the outlet water temperature would be 32°C.
Next, we must consider flow rates. The relationship between temperature difference and flow rate determines the required cooling capacity. If we set a flow rate of 600 liters per second, we need to calculate the associated pressure drop across the heat exchanger. Excessive pressure loss can lead to increased energy consumption, as the pump will need to work harder to maintain flow.
When selecting a flow rate, it’s essential to ensure that the pressure loss remains within acceptable limits. If the pressure drop becomes too high, it indicates that the selected flow rate exceeds the pump’s capabilities. This iterative process involves adjusting flow rates and temperatures to find a balance that meets cooling requirements while respecting system limitations.
ASHRAE provides the foundational guidelines for determining ambient temperatures and operational standards. However, it’s important to understand that each component in the cooling system interacts with the others. Thus, while ASHRAE gives us a starting point, the specifics of the design come from thorough calculations and simulations tailored to the facility’s unique needs.
The data centre of the future
As we look ahead, data centres will evolve to accommodate various types of technology. Future facilities are likely to feature hybrid systems that incorporate both traditional CPUs and specialised AI chips. This necessitates separate cooling systems optimised for each type of technology.
For instance, a data centre could house multiple cooling units designed to handle different temperature requirements. Standard CPU systems might operate at a water inlet temperature of 20°C, while AI chips could function effectively at higher temperatures, such as 30°C. This flexibility allows for greater energy efficiency, especially when integrating free cooling modes.
The design of future data centres hinges on understanding the intricate relationships between cooling systems, technology types, and environmental conditions. As we move toward more sophisticated cooling solutions, the focus will be on optimizing energy efficiency while accommodating a diverse range of computing needs.
