A data center consists of a set of computers and electrical devices that generate heat while operating, making it necessary to use a cooling system to maintain optimal working conditions. This article discusses the types of cooling systems and their characteristics, outlines the appropriate cooling system selection process, and highlights the advantages of simulating these systems with ANSYS Fluent.
What is a Data Center?
A data center is a facility housing a large group of networked computer servers usually used by organizations for remote storage, processing, and distribution of vast amounts of data.
A typical data center
What is Data Center Cooling?
Devices in a data center generate a lot of heat during operation. This heat can reduce the lifespan of and damage hardware, sometimes even leading to fires. On the other hand, optimal data center performance occurs within a certain temperature range. To maintain optimal performance, a cooling system is used to manage temperature and humidity.
Click to access the HVAC CFD
Why is Efficient Data Center Cooling Needed?
Efficient data center cooling is needed for several critical reasons:
- Safety: There is a possibility of fire when the temperature of the equipment rises excessively.
- Lifespan: The lifespan of the equipment is maximized within a certain temperature range.
- Balanced erosion: If the temperature difference inside the data center is large, different devices will have different lifespans. As a result, some devices may get damaged while others remain healthy, leading to increased repair and replacement costs.
- Optimum performance: The data center operates at optimal performance within a specific temperature range.
- Reliability: In the event of a disruption in the performance of the data center, many problems can arise for users. For example, if a bank’s data center is interrupted even momentarily, it can cause significant issues. To ensure the stable operation of equipment, it is necessary to use a cooling system.
Types of Data Center Cooling Systems
Key methods of data center cooling include:
- Free Cooling: Free cooling utilizes ambient air for data center thermal management. This method is viable in regions with consistently low temperatures throughout the year. It can either directly introduce external air or cool internal air via a heat exchanger. The primary advantage of this cooling method is low energy consumption.
- Computer Room Air Conditioner (CRAC): these units are similar to traditional air conditioners, employing a refrigeration cycle to lower air temperature. Its components typically include a compressor, condenser, expansion valve, and evaporator. The refrigerant absorbs heat from the data center air, then releases it outside. CRAC units also manage humidity levels and are known for reliability but can be energy-intensive.
- Computer Room Air Handler (CRAH): CRAH systems utilize chilled water from an external plant to cool air through a cooling coil. Fans propel air over the coil, entering the data center at a lower temperature. This method is cost-effective in regions with cooler climates. The term “chilled water systems” often refers to the water-cooled segment of CRAH units.
- Direct-to-Chip Liquid Cooling (Data Center Liquid Cooling Systems): In this method, coolant directly contacts heat-absorbing plates attached to equipment, efficiently managing high heat generation. Fluid temperature is regulated via a refrigeration cycle, making it ideal for high-performance data centers.
- Immersion Cooling: Here, equipment is submerged in dielectric fluid coolant, ensuring direct contact with chips to absorb heat. Similar to direct-to-chip liquid cooling, fluid temperature control is achieved through a refrigeration cycle. The coolant’s dielectric properties prevent chemical reactions with the equipment.
These methods encompass common data center cooling approaches. Other variants may involve different airflow patterns, equipment arrangements, or configurations, often categorized under these primary methods.
Click to access the Heat Transfer CFD
Best Data Center Cooling Systems
The decision to choose the type of cooling system varies depending on different factors such as budget constraints, cooling system power consumption, ambient temperature, data center size, and the amount of heat produced within the data center.
It can be argued that immersion cooling is the most efficient method for systems that generate a lot of heat. On the other hand, direct air cooling has much lower installation, design, and usage costs compared to other methods. However, its reliability is affected by ambient air temperature fluctuations, which can lead to system malfunctions.Data Center Cooling Design
First, the amount of heat generated in the data center should be estimated. The cooling system should be selected based on the maximum heat produced, ensuring that it can handle even temporary spikes in heat generation. The cooling system must be prepared for the worst-case scenario to maintain optimal performance and prevent overheating.
Next, it is necessary to determine the optimal temperature range and humidity level in the data center. This information is typically available in most standards.
Next, based on the obtained information, the budget, and the ambient air temperature, the cooling system is selected. In the next step, the data center is designed to maximize cooling performance. This part involves arranging devices and directing fluid flow, whether air or liquid, to achieve maximum heat absorption.
Finally, the performance of the cooling system needs to be checked in real-time so that any issues can be promptly addressed. Additionally, regular maintenance and servicing of the system are essential. The system should be designed for easy servicing and maintenance.
Click to access the CFD Fluid Mechanics
Data Center Cooling System Diagram
The diagram for a data center cooling system can be depicted in various formats. Typically, these diagrams illustrate the path of fluid movement inside the data center in two or three dimensions. In some cases, the diagram also includes a section depicting the refrigeration cycle.
A cooling loop for a data center. From “Data center cooling using model-predictive control” by Nevena Lazic et al.
Data Center Cooling Fluids
Air is the most important and versatile fluid in data center cooling systems. Water is used, but not directly in liquid cooling methods or refrigeration cycles; it is mostly used in CRAH units. Common refrigerants are used in all refrigeration cycles. Dielectric fluids are used in immersion cooling due to the direct contact of the fluid with the equipment. Advanced cooling fluids, such as liquid metals or phase-change fluids, are used in specialized cooling applications.
Examples of different cooling architectures of data centers. From “Effect of data center servers’ power density on the decision of using in-row cooling or perimeter cooling” by Abdullah Mohamed et al.
Data Center Cooling Systems Simulations by ANSYS Fluent
All parts of the cooling system, whether inside the data center or in the refrigeration cycle outside the data center, can be simulated with ANSYS Fluent. With the ability to simulate materials with complex properties, this software can analyze all types of cooling systems and predict their performance. Most of the research in this field is related to equipment arrangement and fluid path determination to maximize cooling system efficiency.
Our specialists at CFDLAND can perform simulation projects for all types of data center cooling systems using ANSYS Fluent software. Your desired simulations will be completed with high accuracy and speed. Just order your project at ORDER PROJECT. Delve into our CFDSHOP, where you will find many completed ANSYS Fluent projects. You will surely find something similar to your desired work.
Conclusion
In conclusion, efficient data center cooling is critical to ensuring the safety, reliability, and optimal performance of data centers, which house and manage vast amounts of data. The selection of an appropriate cooling system, whether it be free cooling, CRAC, CRAH, direct-to-chip liquid cooling, or immersion cooling, depends on factors such as budget, energy consumption, and ambient conditions. Simulation of these systems using ANSYS Fluent provides a precise and effective means to design and optimize cooling solutions, ensuring they meet the specific needs of the data center while maintaining high performance and cost-effectiveness.