Digital Twins To Improve Data Center Operations
![Digital Twins To Improve Data Center Operations](jpg/digital-twin.jpg)
Digital twin technology offers exciting possibilities to many industries to know what is happening in real-time, plan strategies and predict problems. Among them are data centers looking for new solutions to improve their operations and reduce their energy consumption and carbon footprint but are not ready to use other next-generation keys, such as artificial intelligence.
In recent years, the data center industry has been improving its facilities’ monitoring systems and implementing new technologies to optimize operations and reduce consumption. Many operators are considering using artificial intelligence to manage energy consumption and carbon emissions, but there are numerous challenges related to data quality and availability. In many cases, the information that can be obtained is limited to homogeneous environments and does not represent the current complexity and diversity in the facilities.
Therefore, the most important challenge for most data centers is improving their operations’ monitoring at multiple levels. This problem could be solved with new technologies such as digital twins. These are fed with all the data related to the operations and allow the creation of a virtual representation of the physical environments with great detail and different uses. On the one hand, there is real-time or near real-time monitoring of what is happening in the data center. On the other hand, they allow simulations to be carried out to discover what would happen if problems of a different nature occur or new systems are implemented.
This proposal is advancing thanks to the work of certain specialized companies, such as the Singapore firm Red Dot Analytics (RDA), which has created an artificial intelligence-driven digital twin platform specifically focused on data centers. Its creators claim that it allows operators to simulate their operations in great detail, helping to manage their carbon footprint and reduce energy consumption.
This company was born at the Nanyang Technological University (NTU). Since its creation, it has focused on the simulation of actions aimed at improving operational and energy efficiency and helping operators better understand the costs and risks of each step in this path. Additionally, this solution can be used to implement machine learning models that would help improve data center operations at different levels.
Regarding Computer Weekly, RDA chief scientist and NTU professor Wen Yonggang explains that “different people have different interpretations of what a digital twin should be, and that has been a big challenge when we work with partners from the industry, customers and stakeholders. He explains that most digital twins are “just” virtual representations of the physical infrastructure, but he sees this as just the first layer of this platform.
It highlights that many other capabilities can be integrated into it, such as the overlay of operational data to carry out statistical analysis and diagnostics and achieve predictive and prescriptive capabilities that drive decision-making for data center operators. And he claims his digital twin can help operators reduce facility power consumption by 40% without having to modify the hardware.
Another approach they have taken at this company is to help operators better understand their carbon emissions and identify ways to reduce them. And they’re also working with other customers on asset management to minimize downtime in data centers. In any case, the objective of this digital twin platform is to be used as a support for decision-making that allows converting the best practices of the industry into a more scientific way of carrying out operations.
The challenge that data centers face is that, in many cases, they do not have enough sources of information since they have obsolete data collection systems and multiple sources that are unable to unify. But When explains that digital twins offer the ability to start with a small amount of data to build the foundation of the virtual representation and then feed in more data to scale and refine the model.