Researchers from SIT, NUS and Power Automation plan to put their findings on smart energy management systems into practice.

(From left) The research team: Dr Yu Ming, Principal Investigator from Power Automation, Associate Professor Xiang Cheng, Co-Principal Investigator from the National University of Singapore, Associate Professor Tan Kuan Tak, Co-Principal Investigator from SIT and Zhao Yunyi, research engineer from SIT, at the building-level islanded microgrid at SIT@NYP Building. (Photo: Tan Kuan Tak)

Over a decade ago, microgrids were a novel concept in Singapore. But now, these self-sufficient energy systems, capable of supplying solar electricity to small communities, could become an important part of Singapore’s efforts to go green – with testbeds on Pulau Ubin and at the Singapore Institute of Technology’s (SIT) upcoming Punggol Campus.

Microgrids, which generate electricity from renewables such as solar and wind power, are capable of operating independently. They can also be connected to a central grid, either drawing power from or supplying surplus energy to it.

“There are many control algorithms that optimise the operations of a single microgrid,” noted SIT’s Associate Professor Tan Kuan Tak, an electrical engineer by training. “But how about when you have many microgrids coming into the network? How do we rely less on the main grid and let microgrids help one another meet the load demand?”

Relying less on the main grid and decentralising energy systems with more localised microgrids could reduce transmission losses when electricity travels long distances from centralised power plants to end-users, defer infrastructure upgrade costs, improve management of localised energy demand and reduce electricity charges.

Traditional control strategies are typically centralised and designed to optimise the operation of individual microgrids. Therefore, they face limitations in adapting to microgrids' dynamic and changing nature when dealing with multiple, independent operating microgrids. For three years, A/Prof Tan and a team of researchers from SIT, Power Automation ( a joint venture between SP Group and Siemens), and the National University of Singapore have been working to create a system that optimises energy efficiency, adapts to varying conditions, and remains robust against environmental uncertainties.

The research team at the Electrical Power Engineering Lab at SIT@NYP Building. (Photo: Tan Kuan Tak)

Their solution: a smart energy management system (EMS) that can control several microgrids at once. Developed with the help of artificial intelligence (AI) and predictive control technology, the cloud-based system anticipates supply and demand, responds to real-time data such as solar power and electricity prices, and directs electricity to consumers cost-effectively – with potential energy cost savings of 10 per cent.

Unlike traditional fixed control methods designed for single microgrids, the new EMS is smart, integrated, and can optimise energy efficiency in real-time. It is suitable for managing different microgrids with varying characteristics, which are influenced by varying weather conditions, fluctuating energy demands, energy sources, etc.

“We wanted to develop a controller that could consider the operating scenarios of different microgrids, with different energy generation mix and demands,” added A/Prof Tan, Co-Principal Investigator for the project titled “Optimisation of Energy Management in Multiple Microgrids System Based on Predictive Control and Artificial Intelligence”, which was funded by the Exploiting Distributed Generation (EDGE) programme from 2020 to 2023.

Tech Talent

Technical challenges are par for the course for research projects. But the biggest obstacle faced was finding the right tech talent – engineers.

The COVID-19 manpower crunch in 2020, the year the project began, made things worse. “When we managed to get someone suitable, usually they were overseas,” lamented A/Prof Tan, whose first research engineer only came on board six months into the project. “It was quite stressful. There were deliverables and milestones we had to achieve, but we did not have engineers to support us.”

In the interim, he was assisted by SIT students who helped with the research as part of their final-year projects and master’s dissertations.

Navigating Complexity

The team of about 10 researchers, spearheaded by Principal Investigator Dr Yu Ming from Power Automation, a joint venture between SP Group and Siemens in the energy sector, grappled with complications. Having enough data to train the software through a process known as reinforcement learning was crucial. The team gathered data on solar irradiance, daily ambient temperature, market electricity pricing, and load demand from industry partners, colleagues, and sensors across SIT.

An overview of the Energy Management System for optimising and monitoring microgrid operations.

For instance, the intermittency of solar energy made it hard for them to predict how much power could be generated from solar panels. Another was how far ahead the system should forecast load demands.

“If the prediction horizon is very long, it could result in heavy computation burdens for our system. Whereas if we decrease the prediction horizon, it might not be so accurate,” A/Prof Tan explained. “The quality of available data may influence the effectiveness of the proposed solution.”

Mr Benjamin Tay, CEO of Power Automation, said “It has been exciting for my team to contribute to this endeavour as they enthusiastically embrace AI to advance our sustainable technology practices and explore innovative approaches to unleash the potential of microgrids. With each complication we overcome, we gain valuable insights to craft even better solutions in the future.”

SIT Punggol Campus as Testbed

Solar Photovoltaic panels on the building rooftops at SIT Punggol Campus.

The team is looking for partners to let them test their solutions on actual microgrids.

“We want to work with different partners to have different microgrid systems to continue testing our algorithm and refining it,” said A/Prof Tan.

SIT’s Punggol Campus, which will be powered by the largest private microgrid in Singapore when it is ready in the second half of 2024, is an ideal place for such an experiment. The upcoming grid will be able to integrate electricity, thermal and renewable energy sources into its network.

“The special feature of this Punggol microgrid is that it can be reconfigured into multiple interconnected nano-grids, pico-grids, and DC grids. It will be a good testbed for our EMS to demonstrate its strength in controlling multiple microgrids,” said A/Prof Tan.

Meanwhile, the team is also helping Jurong Port calculate the expected energy savings by introducing more solar panels and energy storage batteries. In exchange, the researchers receive data which they can use to refine their algorithms.

A/Prof Tan is heartened by the strides made in AI over the past decade. “When I started my PhD, microgrids were new in Singapore; we were just talking and researching. Now, bit by bit, all these things are falling into place.,” he shared.

I am very happy to see that AI technology can make a real difference in the green energy drive of Singapore,” said Associate Professor Xiang Cheng, Co-Principal Investigator from the National University of Singapore.