The new algorithm is able to analyze energy consumption by tapping on sensors in computer chips already found in equipment such as computers, servers, air conditioning systems and industrial machinery.
Such computer chips are needed for a host of functions such as to measure temperature, log data traffic and monitor the workload of computer processors.
By combining it with externally-placed sensors, such as those that monitor ambient temperature, the new technology can integrate and analyze all the operational data and recommend energy-saving solutions with almost no upfront cost.
The algorithm which extracts all readily available data and turns it into a treasure trove of information that can be studied and analyzed has been developed by Asst Prof Wen Yonggang from NTU’s School of Computer Engineering.
The algorithm has been licensed by an NTU-incubated company, Evercomm Singapore.
In a typical semiconductor factory which produces computer chips and components for computers and mobile devices, the annual electricity bill could easily reach S$50 million and more.
Ted Chen, co-founder and product architect of Evercomm Singapore, who worked with Asst Prof Wen to commercialize the technology, said: “With NTU’s new analytic engine, such large semiconductor factories and campuses could save up to S$1 million a year without a need to change much of their hardware, and instead, tune their operation and time their energy usage.”
“The new algorithm allows us to use the most cost-effective way to find out where we can save energy, and our performance can be guaranteed by using real-time data.”
Evercomm, a two-year-old company, already has a few semi-onductor manufacturers as its clients, of which one is a heavy electricity user in Singapore, GlobalFoundries, the second largest foundry in the world.
GlobalFoundries’ management team is committed to sustainable energy consumption by providing ample opportunities to adopt local innovations, as proven by being the first in the industry to engage Evercomm’s energy analytic services.
In Taiwan, Evercomm is engaged by National Dong Hwa University and Chunghwa Telecom to deploy their energy analytic engine across the entire university campus.
“By combining the software algorithm with hardware sensors, we can find out exactly how much cooling a room needs, whether there is an oversupply of cooling and so adjust the air flow and temperature to achieve the best balance,” said Chen, an alumnus of NTU’s School of Electrical and Electronic Engineering.
Even without deploying external sensors, Evercomm can achieve up to five per cent energy savings for companies, which are facing increased regulations worldwide on their energy usage and resulting carbon footprint.
Evercomm is looking to expand its expertise into data centre industry and has deployed a pilot test at the NTU Green Datacentre, saving five per cent of its monthly electricity bill.
In a datacentre consisting of hundreds of computer servers, more than half of the energy costs are spent on cooling these servers through air-conditioning.
“Servers which are performing intensive computing will generate a lot of heat,” added Prof Wen, an expert in cloud computing and green data centres. “If we know which of these servers are, we can spread out the computing load and so reduce the heat emitted by the servers, in turn reducing the energy needed to cool them.”
Related articles and links: