In this highlight clip, Professor Yonggang Wen of Nanyang Technological University (Singapore) and the Chairman of Red Dot Analytics (RDA) explained how Cognitive Digital Twin technology can be used to make data centers more sustainable.
Data centers accounted for about 7% of Singapore’s total electricity consumption in 2021. These data centers consume huge amounts of electricity because a lot of electricity is required to cool the servers. To cool these servers, an enormous amount of cold energy must be generated to extract the heat from them. Professor Wen also mentioned that there is a lot of wastage of energy in the process of cooling because servers are often overcooled in order to keep them at a constant temperature. This is regardless of the load of the server, hence even when the server’s load is low, it would still be overcooled.
Professor Wen then emphasized the inefficiency of this – not just for the environment but also for businesses because it would cost a lot of money to continuously keep the servers cooled.
As such, Red Dot Analytics has used AI and its cognitive twin technology to enable data centers to be more energy-saving. Their cognitive engine is a machine learning-based engine that utilizes a deep reinforcement learning approach and neural network to train the learning algorithms continuously. This is complemented by their digital twin – a visualization platform that is calibrated with a physical sensor data to provide powerful real-time insights, which enables data center operators to save electricity and reduce the chance of outages.
Professor Wen also highlighted the importance of the process of the Dual-Cycle Loop with Human-in-the-Loop Architecture, which refers to the process of working in the physical world and the cyber world in conjunction with human experts to provide the best and most trustworthy actions for infrastructure management.
“Digital model first, before we get into the physical reality of the world”Professor Yonggang Wen
In closing, Professor Wen emphasized the importance of the metaverse approach to address these industrial issues and highlighted that testing models in the digital realm first before implementing them could save companies a lot of money while also ensuring that data centers are more sustainable.