A bird’s view of the data center industry

“Who would not want to be a part of a rising economy, be in any capacity, everyone finds a way to enter an explosive industry like data center”

An industry that consumes 1 % of the global energy and generates 360 million tonnes of CO2, expends around 50 % of expenditure and energy on cooling, just think about how much scope for improvement there is. The major pain points of the data center industry are 1) achieving higher efficiency and performance, 2) lowering energy consumption, and 3) lowering the carbon footprint. Why take it lightly when the industry spends 50 % of its energy on cooling? Keeping the data centers cool while recycling the evolved heat should be the prime motive. On top of it using renewable energy sources to drive your data centers is the need of the hour for the environment and for humankind.

Paanduv's capabilities for Datacenter modeling

“Computational fluid dynamics in the integration of AI can change the way data center industry is currently operating”

The pandemic forced companies to function in a different manner than they used to. This has posed new challenges for the companies to function in a way without hampering overall productivity. Most practices are done on a remote basis and thus DATA has become more useful than it ever was. Data is all that we all are relying on. Data centers are meant to save data and operate on them. Data is all that we generate from computational modeling (CFD). Data is all that is required for Artificial Intelligence (AI) to train the agent.

Globally growing data center economy

CFD and AI-assisted expert system (C-AES) developed by Paanduv Applications is capable of predicting the arrangement of dry coolers for heat recovery from data farms. It elaborates on the thermal data from the new dry cooler's configuration and moreover tells you about the heat island effect of the dry coolers. It also demonstrates the effect of wind direction, wind speed, and local temperature on the heat islands and hot air from the dry cooler's outlet.

C-AES interface developed by Paanduv Applications (CFD-AI-based expert system)

The data center services market valuation is $48.9 billion in 2020. It is projected that by 2026 this will increase to $105.6 billion. Considering the high Capex in this industry computational modeling is the most effective way to run this industry efficiently. No other industry relies on computational modeling as much as this one does. Computational modeling has answers to all your questions.

Which cooling unit should be used?
How should the servers be arranged in a room
What should be the location of the cooling units
How many cooling units will be required to lower the temperature
Which cooling mechanism would best suit and is most efficient for cooling
Is there a need for perforated tiles in the architecture
How efficient my fan coil assembly would be
Will there be a backflow of air from the cooling units' inlet
How would the airflow in the room
Where are the dead zones
How does the thermal; profile looks like
What if one of the units fails
What will be the thermal effects when the units are running at the highest peak load?
What if units are running at 50 % and any cooling capacity
How is the HVAC of the electrical room
Will immersion cooling be more efficient
Can I see the bubble evolving from tow phase immersion cooling
What will be the average temperature of the room, the maximum temperature of the room, and the temperature at any time and space?
What should be the dry cooler arrangement in the data farm
How well the heat can be dissipated from the system
What will be the effect of wind speed and direction on the heat evolved from dry coolers
Which heat exchanger will be more suited for the architecture
How can I minimize the heat island effect
What can be the best way to move towards a greener approach
If I have data for multiple scenarios what will be the results for the next scenario

And many more…….

Most of the questions are already answered by Paanduv to their clients. Our solutions to these problems are actually deployed by the data center industry folks and based on that they are making their upcoming data center and data farm strategies.

Broadly compartmentalizing data centers into three areas 1) Power 2) Information technology, and 3) Cooling equipment. All three need computational modeling to function smoothly and efficiently. Datacenter modeling covers physics such as conjugate heat transfer including all modes of heat transfer, concept and analogy of porous media, turbulence, bubble dynamics, and multiphase modeling.

Data center architecture and computational modeling strategy

“Power Usage Efficiency (PUE) is the quantitative factor that defines the energy efficiency of the data center. The lower the PUE greener the data center.”

“Water Usage Effectiveness (WUE) is the ratio between the use of water in data center systems (water loops, adiabatic towers, humidification, etc.). WUE is used as a metric to measure data center sustainability in terms of water usage and its relation to energy consumption.”

Sustainability in data centers

Industries are now more aware of climate change and their responsibility. That is the reason why energy consumption is not increasing proportionally to the increase in workload on the data centers. Companies like Microsoft, Meta, and Google are building data centers in isolated places such as Finland and Sweden using renewable energy such as ocean wave energy, and hydroelectric power to run data centers and utilizing the heat coming out of the centers to cater to the city's electricity requirements. More and more such efforts are made and data centers are deployed with smaller carbon footprints. Lafdal data center in Norway is the largest data center in Europe with low PUE, zero CO2 footprint, and zero WUE.

Sustainable solutions to data centers (Quanchao Technology mountain data center, China(left), Lefdal mine data center, Norway (right))

The upcoming blog will include some technical case studies that provide the solution to reduce the temperature of an electrical room and the effect of ambient conditions on the heat island effect in a data farm.

“Data centers currently account for 4% of the total greenhouse emissions worldwide. The CFD-AI-based expert system can help you achieve the full potential of the data center industry.”

“Paanduv Applications is helping industries reach their full potential and upgrade their cutting edge technologies by their computational modeling consulting and R&D services”

Written by

Dr. Rimzhim Gupta

Research Scientist,

Paanduv Applications

Page updated

Report abuse