Computational Fluid Dynamics (CFD) for Data Center Thermal Management

As power and heat loads rise, efficient cooling is becoming more challenging. Developing an airflow improvement plan is difficult as it is not perceptible by the eyes. One way to address this challenge is the use of Computational Fluid Dynamics. Together with a temperature sensor, managers can have more comprehensive information valuable for thermal management. This empowers them to proactively manage the data center cooling system. 

As a data center is critical in business operation, IT equipment functions together 24/7 generating heat simultaneously. Using a laptop computer for a prolonged time can heat a person’s lap. What more for rows of servers? The heat generated is enough to pose serious threats to data centers. Higher electricity expense due to increasing energy consumption is not only the issue. If cooling is inadequate, the equipment can overheat. Hotspots can also develop somewhere inside the facility. These problems are the main culprit for data center downtime.

What is CFD?

CFD modeling is a tool that uses an algorithm to simulate thermo-fluid in a system. It includes chemical reactions, evaporation, melting, and freezing. CFD is capable of analyzing complex data on fluid to fluid, fluid to solid, or fluid to gas interaction. The speed and higher capacity in computing data contribute to CFD accuracy as a mathematical model. 

The CFD simulation process consists of three primary steps. These are:

• Pre-Processing – the geometry is described and will be divided into elements (mesh)
• Solver – the fluid material properties, flow, and boundary conditions are set to solve through a computer
• Post-Processing – after solving, data will be analyzed with different methods for a comprehensive representation

CFD in Data Center

CFD is used in a variety of fields including the data center specifically in thermal management. Computer-Aided Design (CAD) focuses on the solid objects found inside the computational domain. On the other hand, CFD works on the spaces between or inside the objects. This makes it useful in maintaining proper airflow in the data center. CFD software has 3D objects that are usually found inside the data center. These include cooling units, racks, and perforated floor tiles. However, it should be noted that in CFD, the smaller the computational domain, the more accurate the result. For bigger data centers, dividing the space into smaller parcels is recommended. This aims to achieve an accurate result. 

While DCIM software can keep track of temperature and power, it cannot represent airflow and predict the impact of future installations. For mission-critical facilities like the data center, the in-depth data of CFD plays an important role in maintaining a continuous business operation. 

Benefits of using CFD in Data Center Thermal Management

1. Identifying Potential Hot spots

Temperature is a critical part of data center management. It is the major cause of equipment failure. The study of Uptime Institute found that almost 10% of the racks run hotter than what is stated on the reliability guidelines. A hotspot is defined as “a location at the intake of IT equipment where the measured temperature is greater than the expected value as recommended by ASHRAE TC 9.9”. The ASHRAE thermal guidelines recommend an allowable temperature between 5 to 45°C (41 to 113°F). If prolonged, hotspots will affect the equipment’s performance and warranty agreement between the manufacturer. 

The CFD can help managers in handling hotspots problems. The 3D simulation of the airflow movement provided by the CFD will show the areas receiving inadequate airflow. CFD allows virtual re-arrangement of the physical layout. Managers can virtually relocate the problem loads in the server room. This is done by moving the equipment from rack to rack to see how to get rid of the hotspot. Then, the high-power servers can be placed in the colder are with higher cooling capacity. Another solution adding open tiles on the hotter rack. The CFD will predict the airflow and temperature reduction through an interactive representation when these actions are taken. AKCP sensorCFD™ uses live sensor data to constrain the CFD model for a realistic view of the data centers performance.

2. Pre-construction Design Planning

Poor design planning can lead to post-construction complications. Knowing the impact of airflow distribution is an important part of design planning. Engineers can identify the place of the cooling equipment near the IT loads. However, computing the actual distribution of air can be a challenge for them.

The CFD’s ability to predict it can help engineers in designing the right layout. The proposed design can be reviewed by modeling the data hall. As a result, engineers can see thermal management problems that may arise in the future. Some of these are air recirculation, mixing of hot air and cold air, and bypass. With an advanced visualization, engineers can have insight into the worst-case failure scenarios. They can redesign the layout if necessary and study different solutions to avoid them. This will save time and money for possible repair if problems are discovered after construction. Another is the huge downtime cost ($5,600/minute) for issues during the operation. CFD is also used to understand the effectiveness of cooling varying IT layouts and determine the feasibility of supporting proposed cabinet deployments. 

3. Installing Additional Equipment

Deciding where to place new equipment is another critical aspect of data center management. Additional equipment increases the hot airflow. There are a lot of considerations including the floor space, size of the equipment, power requirement, and sufficient cooling. 

The CFD determines if additional cooling is necessary to suffice the heat. More importantly, it guides managers in identifying the optimal location for the new equipment. The Rack Cooling Index (RCI) can be computed when CFD predicts the equipment inlet temperature. For instance, if two CRACs are positioned on opposite walls in one section of a data center, new equipment with high density is placed near the ends of the racks. This is because air recirculation is more likely to happen in the center. Using CFD, managers can attest to the impact of this method and assure the long-term survival of the new equipment. 

4. Optimizing Airflow

Data centers experience hotspots 10% of the time even after providing 2.6 times more than the cooling enough for a heat load. This calls for an effective method of optimizing airflow. In some cases, hotspots happen not because of inadequate airflow. A recent study shows that this problem happens because air can’t flow to the necessary areas. Generally, managers either organize the data center into hot and cold aisles or improve the floor space. However, these modifications require a high operational cost. One mistake can make huge money go to waste. 

Therefore, it’s a practical action to test these modifications before implementation. Modeling the airflow distribution first and understanding it is an important pace. Through a CFD simulation, data is analyzed without actually performing them. It is possible to see the amount of server temperature is reduced. This benefits the managers by pre-determining the most effective method in addressing disrupted airflow. 

5. Reduce Excess Cooling

A data center is an energy-intensive facility considering all the equipment housed inside. A typical data center can consume as much energy as 25,000 households. The managers’ priority is to provide sufficient cooling to protect the equipment from overheating. But since both airflow and temperature are not visible to the eyes, the amount of cooling required remains guess-estimated. It has been a practice for them to set their cooling units at maximum. Due to the over-estimated cooling requirement, the energy consumption will be higher. This results in to increase in operating expenses. 

Since CFD can provide simulation on what is not visible to the eyes, managers can know if the cooling is already sufficient. This prevents excess cooling, saving up the money that is supposed to be paid for the high electric bills. 

Temperature Monitoring 

Real-time temperature monitoring and CFD are often used interchangeably. These two are technically different but are supplemental to each other. While CFD predicts the temperature and airflow distribution, temperature monitoring provides real-time data from environmental sensors. Managers maintain a low ambient temperature in compliance with the ASHRAE. Over the years, it has been one of the most important aspects of data center management. Nonetheless, both CFD and temperature monitoring are valuable in creating a thermal management strategy, protecting the data centers from cost and threats. 

Right Sensor for an Accurate CFD Simulation

Temperature sensors measure the temperature in the environment and convert the data into a record. They can be placed on walls or racks in the data center. Sensors are connected to an alarm and notify the managers if the temperature reached its threshold. 

The role of a temperature sensor is critical in CFD simulation. When the temperature sensor gathered data, it will be then directed to the CFD modeling system to predict scenarios. Environmental parameters are used in calibrating the model. Therefore, the accuracy of the CFD simulation can be based on the accuracy of data provided by the temperature sensor. 

 

AKCP offers wireless temperature sensors that can be installed in your data center facility to measure the in-rack and ambient temperature. Its wireless feature allows managers to install it without cable works and construction costs. It is manufactured using highly integrated, low-power surface mount technology to ensure long-term reliability.  Using AKCP wireless temperature sensor, managers can be more confident in achieving superior accuracy on the CFD simulation. Thus alleviate all the threats of temperature and airflow on the data center. 

Protect your data center through an effective thermal management plan. Message AKCP now!

Reference Links:

https://tc0909.ashraetcs.org/documents/ASHRAE_TC0909_Power_White_Paper_22_June_2016_REVISED.pdf

https://www.esmagazine.com/articles/101279-the-benefits-of-indoor-cfd-modelling-in-data-center-applications

https://www.downloads.siemens.com/download-center/d/111_White-Paper–Using-CFD-for-Optimal-Thermal-Management-and-Cooling-Design-in-Data-Centers_A6V10910773_us-en.pdf?mandator=ic_bt&segment=HQ&fct=downloadasset&pos=download&id1=A6V10910773

https://coolsimsoftware.com/Portals/0/PDF/WP107_CFD_for_Data_Centers.pdf

https://www.datacenterdynamics.com/en/opinions/how-useful-is-cfd-for-data-centers/

https://www.hitechcfd.com/cfd-knowledgebase/seven-stages-of-a-typical-cfd-simulation.html